Topic: VST plugins & mixing/mastering workflow in your DAWs

[if_then_else]

FWIW: I recently discovered a few new (to me) VST plugins during the Black Friday sales (e.g. TBProAudio DSEQ3 and ABML2, the Sonible smart bundle, OEKSound Spiff and Soundtheory Gullfoss), which have found their permanent place in my DAW for the music I usually record (mostly SBD+AUD matrices and multi-track SBD recordings).

Even though many tasks could already be done manually in the DAW in the past, these tools make many things easier and, above all, lead to a simplified, fixed workflow and a huge time saving.

However, I have found that the sequence in which you load the plug-ins in the mix busses and especially on the master bus makes a huge difference.

I.e.: Filters and (Dynamic Spectral) EQs first. Then transient shaping and saturation/distortion plug-ins. Then reverb and stereo imaging and finally the limiter. For loudness matching, you need two instances of ABML2 on the master bus (one as the first and one as the last plug-in in the chain). Otherwise you are simply under the illusion that “louder is better”.

I have collected the plugins in the list over several years and some of the VSTs are certainly redundant in function. But sometimes a more user-friendly interface or more precise manipulation options were worth the switch for me.

Question for the others here: What do you use for what purposes and are there any plugins that are game-changers for you in terms of your workflows (either in terms of precision or time savings)?

## Alignment and Phase Correction
- Sound Radix Auto-Align2: Automatically time-aligns and phase-corrects multi-microphone recordings.

## Audio Restoration and Editing
- Acon Digital Acoustica Premium: Comprehensive audio restoration suite for removing noise, clicks, and other artifacts from recordings. Feature-wise, it's comparable to RX Standard but all of the yearly upgrades to version 7 were free over the past 5+ years.
- iZotope RX Standard: Industry-standard audio repair toolkit offering modules for noise reduction, de-clicking, spectral repair, and more.

## Channel Strip Emulation / Expanders
- SSL Sourcerer: Advanced channel expander specifically designed to reduce mic bleed in multi-track recordings. Provides precise control over unwanted audio leakage without harsh gating artifacts.

## Compressors
- Fabfilter Pro C2: Versatile compressor with multiple compression styles and extensive visual feedback.
- Fabfilter Pro MB: Multiband compressor allowing frequency-specific dynamic control.
- Klanghelm MJUC: Vintage-style variable-mu compressor emulation with a warm, musical character.
- Sonible smartComp 2: AI-driven compressor that analyzes the input and suggests appropriate compression settings.

## Deverb
- Acon Digital deVerberate: Specialized plugin for reducing or removing unwanted reverb from audio recordings.

## Dynamics Processors
- Fabfilter Pro DS: Precision de-esser for taming harsh sibilance in vocal recordings.
- Fabfilter Pro G: Versatile gate and expander for noise reduction and dynamic shaping.
- Wavesfactor Trackspacer 2.5: Carves out space for vocals in a mix by dynamically reducing conflicting frequencies in other tracks.

## Enhancers
- AudioThing Type A: Type A uses the Dolby A encoding method without subsequent decoding, applying its compression and equalization techniques to enhance audio signals. This results in added warmth and character by boosting specific frequencies while compressing others.

## Equalizers
- Fabfilter Pro Q4: High-precision equalizer with dynamic EQ capabilities and extensive visual feedback.
- Sonible smartEQ 4: AI-powered equalizer that analyzes and suggests EQ curves based on the input signal.
- TBProAudio DSEQ 3: Dynamic spectral equalizer that applies EQ adjustments based on the input signal's dynamics.
- Voxengo TEOTE: Another dynamic spectral EQ that automatically balances the frequency spectrum of audio material.

## Filters
- TDR Special Filters Bundle (Arbiter, Elliptical, Infrasonic, Ultrasonic): Collection of high-quality filters for precise frequency shaping and clean-up tasks.

## Intelligent Processing
- Soundtheory Gullfoss: AI-powered equalizer that dynamically enhances clarity and balance in real-time.

## Limiters
- Fabfilter Pro L2: Transparent limiter with extensive metering and oversampling options.
- Sonible smartLimit: Intelligent limiter that automatically adapts to the input signal for optimal loudness.

## Loudness Matching
- TBProAudio ABLM2: ABLM2 is a gain control plugin that enables level-matched A/B comparisons of audio tracks. It measures loudness before and after processing, automatically adjusting levels to eliminate loudness bias for more objective evaluations of effects. With multiple measurement modes and auto-linking features, it's an essential tool for mixing and mastering.

## Metering and Analysis
- Sonible trueBalance: Analyzes and visualizes the spectral balance of audio signals across the frequency spectrum.
- Sonible trueLevel: Provides loudness metering and analysis tools for achieving consistent levels across tracks.

## Reverb
- LiquidSonics Reverberate 2: Convolution reverb with modulation capabilities for realistic space emulation.
- LiquidSonics Seventh Heaven: High-end reverb plugin emulating the Bricasti M7 hardware reverb unit.
- Valhalla Room: Algorithmic reverb known for its natural sound and extensive customization options.
- Valhalla Vintage Verb: Emulates the character of classic digital reverbs from the 1970s and 1980s.

## Saturation and Distortion
- Fabfilter Saturn 2: Multiband distortion and saturation processor with modulation capabilities.
- Kazrog True Iron: Emulates the subtle saturation and harmonics of audio transformers.
- Klanghelm SDRR2: Versatile saturation plugin with multiple modes for various types of analog-style distortion.
- Kush Omega bundle (458A, A, N, TWK): Collection of analog-modeled processors for adding warmth and character to audio. I prefer 458A and TWK for the type of music I record. Your mileage may vary.
- Soundtoys Decapitator: Popular analog saturation emulator known for its aggressive, punchy character.
- Soundtoys Radiator + Little Radiator: Emulations of vintage Altec console preamps for adding tube-like warmth.

## Stereo Processing
- Nugen Monofilter 4: Controls low-frequency content to ensure mono compatibility in stereo mixes.
- Nugen Stereoizer 3: Expands the stereo image of audio signals while maintaining mono compatibility.
- Nugen Stereoplacer 3: Allows precise positioning of frequency bands within the stereo field.
- Ozone Imager 2: Stereo widening tool for enhancing the perceived width of audio signals.
- Stereo Savage 2: Enhances stereo width and depth through mid-side processing techniques.

## Tape Emulation
- Softube Tape: Adds analog tape warmth and compression to digital recordings.
- u-he Satin: Versatile tape emulation with extensive control over tape machine characteristics.
- UADx Ampex ATR-102 Master Tape: Detailed emulation of the Ampex ATR-102 mastering tape machine.
- UADx Oxide Tape Recorder: Simulates the warmth and cohesion of recording to tape.

## Transient Shaping
- OEKSound Spiff: Transient shaper that allows precise control over the attack and sustain of audio signals.

[admkrk]

Ozone for dither and light limiting, mostly because I cannot figure out how to get the dither to work otherwise.

Rx to remove the occasional clink or clank.

[morst]

## Alignment and Phase Correction
- Sound Radix Auto-Align2: Automatically time-aligns and phase-corrects multi-microphone recordings.

https://www.soundradix.com/products/auto-align/


Wow I need to check this out.
I wanna compare this to using something like SMAART versus my manual method. I bet this and SMAART beat me but I'd like to hear by how much... $199 huh? The pros think it's a great deal for their 48 tracks... that would be a hassle to do manually.

[goodcooker]

On almost every recording I use -

*deClick -
*deNoise - this also includes a dePlop that decreases the woof from low bass that seems to escape my EQ so I put it in pre EQ
*EQ - I use several different ones but unless I need a multiband parametric EQ I use API 550 or 560
*SSL Channel Strip - this is a simple but flexible EQ and dynamics processor

My magic bullet is Waves L3 Ultramaximizer - it's a dynamics processor that uses simple presets and has only two controls - an activation threshold and an output limiter setting. It's great glue for the master.

[if_then_else]

## Alignment and Phase Correction
- Sound Radix Auto-Align2: Automatically time-aligns and phase-corrects multi-microphone recordings.

https://www.soundradix.com/products/auto-align/


Wow I need to check this out.
I wanna compare this to using something like SMAART versus my manual method. I bet this and SMAART beat me but I'd like to hear by how much... $199 huh? The pros think it's a great deal for their 48 tracks... that would be a hassle to do manually.

I actually only discovered Auto-Align 2 thanks to another thread here - about 2 or 2.5 years ago, and I've used it on all my SBD+AUD matrix recordings since then. I also tried it on some multi-track matrix recordings, but with more than 32 channels the system load eventually got very high. It might be smarter to use it on groups of channels rather than individual channels.

The thing is: Previously, I always aligned my SBD+AUD matrix recordings manually with élastique Pro in the DAW. In Samplitude, this is accurate to 5 or 6 decimal places. This means that even if the result is “almost” exact, there is still a small delta after 30 minutes or 60 minutes and the phases no longer match exactly.

I still do the initial, rough alignment manually but then let Auto-Align 2 do the surgical work. The effect is clearly noticeable: i.e. any comb filter effects are eliminated and the phases match exactly. The program constantly matches samples from the individual channels and corrects them. The result is much clearer for me.  And since then I've spared myself the discussions with some band members about the fact that something sounds “off”.

Whether this is worth $200 for everyone, I don't know. My time was well worth it.

Sound Radix always have a plugin sale around Black Friday (I think reduced by about one third of the asking price). I don't know if there are any other sales throughout the year. If you need it (and can wait) you should go for it then - as with most other plugins.

There is also a specialised, even more expensive “Auto-Align Post”, but this is more intended for film sets where talents move around the room. I think even the manufacturer recommends Auto-Align 2 for music, because it is more suitable for this kind of content.

[if_then_else]

Surprisingly, perhaps the most useful plugin for me was TBProAudio ABLM2. (The price is quite fair for such a plugin and the manufacturer never has any sales promotions).

I had always wondered why the resulting mix was so much quieter before the final limiting than when I was monitoring it. This is because many of the other plugins in the chain add gain and when you lower the level to avoid clipping, it suddenly doesn't sound so great. The perception is that louder is always "better".

ABLM2 equalizes the gain of the individual plug-ins and even adds a safety margin if desired.

This means that what I hear when monitoring in the DAW is what I get when mastering. And I can correct it in real time without having to change and re-render anything later.

[morst]

There is also a specialised, even more expensive “Auto-Align Post”, but this is more intended for film sets where talents move around the room. I think even the manufacturer recommends Auto-Align 2 for music, because it is more suitable for this kind of content.

This might be particularly useful to folks attempting to synchronize direct board recordings with those made using body worn mics.

[phil_er_up]

 if_then_else - thanks for posting these. It is appreciated.

DL'ed the  Sound Radix Auto-Align2 plugin and could not get wavelab to see it as a plugin.

Tried the TBProAudio ABLM2 as a plugin in and managed to get it to work in wavelab and it does look like a good tool.

Thanks again.

[if_then_else]

if_then_else - thanks for posting these. It is appreciated.

DL'ed the  Sound Radix Auto-Align2 plugin and could not get wavelab to see it as a plugin.

Tried the TBProAudio ABLM2 as a plugin in and managed to get it to work in wavelab and it does look like a good tool.

Thanks again.

Not sure about the issue with Wavelab and Auto-Align 2... It could be because it might not have been installed to the default Windows folder. (I think the default folder for VST plugins would be C:\Program Files\Steinberg\VstPlugins. In my specific case Auto-Align 2 is installed in C:\Common Files\VST3\AutoAlign2.vst3\Contents\x86_64-win\AutoAlign2.vst3). Maybe just see if that's the same issue with your installation and, if so, add the path to your VST search path in Wavelab and then re-scan the plugins.

Re. TBProAudio ABLM2: I noticed today that there's a plugin called "Letimix GainMatch" that supposedly serves a similar purpose. At a first glance, it looks less advanced/customisable than ABLM2, but it's also just $19 compared to the €49 for ABLM2.

[boomfizzle]

## Alignment and Phase Correction
- Sound Radix Auto-Align2: Automatically time-aligns and phase-corrects multi-microphone recordings.

Whether this is worth $200 for everyone, I don't know. My time was well worth it.

Sound Radix always have a plugin sale around Black Friday (I think reduced by about one third of the asking price). I don't know if there are any other sales throughout the year. If you need it (and can wait) you should go for it then - as with most other plugins.

There is also a specialised, even more expensive “Auto-Align Post”, but this is more intended for film sets where talents move around the room. I think even the manufacturer recommends Auto-Align 2 for music, because it is more suitable for this kind of content.

I've posted about Auto Align Post 2 here in the past.  I actually just recently purchased Auto Align 2 as well, for tracking live drums.   (They had a bonus Black Friday discount for existing customers)
It's worth calling out that Auto Align 2 calculates the sample shift required to align phase, and that's it.  If you're trying to align two sources from decks with different clocks, the ability to have the "Post" version do any time stretching (difference between Static and Variable) necessary to align transients is worth it in time saved to me, as well.
It's expensive software, but try the demo and if it's not for you, save the money.

I mainly use FabFilter Pro L-2 to grab some stray peaks and dither.   I much preferred the results I got this way than by relying on normalization. 

I also want to mention the two attached pics...I used these Fairchild settings (taken from Chase's picture this past fall on Phish Tour) in my UA Fairchild plugin, along with these UA Pultec settings (stolen from a Spector murder doc), and loved what they each added to the last show I recorded.  YMMV, but it feels like I'm standing on the shoulders of giants on playback.

[phil_er_up]

if_then_else - thanks for posting these. It is appreciated.

DL'ed the  Sound Radix Auto-Align2 plugin and could not get wavelab to see it as a plugin.

Tried the TBProAudio ABLM2 as a plugin in and managed to get it to work in wavelab and it does look like a good tool.

Thanks again.

Not sure about the issue with Wavelab and Auto-Align 2... It could be because it might not have been installed to the default Windows folder. (I think the default folder for VST plugins would be C:\Program Files\Steinberg\VstPlugins. In my specific case Auto-Align 2 is installed in C:\Common Files\VST3\AutoAlign2.vst3\Contents\x86_64-win\AutoAlign2.vst3). Maybe just see if that's the same issue with your installation and, if so, add the path to your VST search path in Wavelab and then re-scan the plugins.

Re. TBProAudio ABLM2: I noticed today that there's a plugin called "Letimix GainMatch" that supposedly serves a similar purpose. At a first glance, it looks less advanced/customisable than ABLM2, but it's also just $19 compared to the €49 for ABLM2.

Thanks for the reply If_then_else.

Did add "C:\Common Files\VST3\' to the path in wavelab and rescanned plugins and it still did not see it. Tried this and some other methods for about an hour and could not get it to work. Have an older version of wavelab so that might be the issue.

Would the auto-align2 fix the combing effect with 2 sets of cardioid  mics from the same mic stand...like one set of cards with Supercardioid  or wide card?
Also, like the idea of using it with SBD and aud recordings.

[if_then_else]

Would the auto-align2 fix the combing effect with 2 sets of cardioid  mics from the same mic stand...like one set of cards with Supercardioid  or wide card?
Also, like the idea of using it with SBD and aud recordings.

Yes. That's actually my use case. I mainly record 4 or 5 audience channels + a stereo soundboard feed and matrix them in post.
Cf. the screenshot below. The darker areas in the Correlation diagram show the corrections.

[Gutbucket]

Following..

Will be curious to try Auto-Align2 at some point. In particular to see what it does with the multi-mic stereo configs I run

[Ronmac]

Melda Production make some very fine, intelligent plug-ins at reasonable prices. I have been using them for over a decade and highly recommend picking up their free bundle.

https://www.meldaproduction.com/MAutoAlign?gad_source=1&gbraid=0AAAAAD3G1YeVoaZ69kRNLND2uu8PuHPjP&gclid=CjwKCAiApsm7BhBZEiwAvIu2XwjTwJpJ_30PvFNRN5fASvb515UblA8k1aMe0JlNXbbQwqzWvdjsqBoCHhkQAvD_BwE

[phil_er_up]

Went online and looked for other Alignment and Phase Correction tools and found some.

Does anyone recommend another tool besides  auto-align2 that does Alignment and Phase Correction?

Really want to try this. Ran wide cards and cards with omni mics for many years on one mic stand. 

[if_then_else]

Personally, I haven't used MAutoAlign. The reviews on Gearspace and other sites appear to be "mixed" (no pun intended). In other words: it appears to have worked fine for some people on some sources and not for others.

https://gearspace.com/board/music-computers/1279132-sr-auto-align-vs-mautoalign.html

I think there's also a similar plugin from Nugen Audio called "Aligner". I haven't used that one, either.

Give them a try (if there are trial versions available) and see what works best for you...
The SoundRadix version appears to be the industry standard but, admittedly, it doesn't come cheap.

[phil_er_up]

Thanks If_then_else and ronmac.

Finally got something to install. Tried again to get auto-align to work in wavelab and could not.

Went a different route and installed and used MAutoAlign. Though could not get wavelab or spectralayers to see the MAutoAlign plugin.

Went to Davinci resolve in fairlight and MAutoAlign is there as a VST plugin.
Figured out how to use MAutoAlign  in  Davinci resolve fairlight.

Figured out how to export the wav file in DR though it will only output 24/48 unless you buy the full version for $300. The Full version will output 24/96.

So at a crossroads as how to move forward. Will buy DR if I have too. Though would rather get it to work in wavelab so I don't have to jump to multiple apps.

Anyone have any suggestion on how to proceed besides buying DR full version?

[if_then_else]

Try Reaper first. It's unrestricted "freemium" / shareware.
A non-commercial license will set you back $60.

Reaper loads Auto-Align 2 well on my PC.

[Ronmac]

MAutoalign works well in Reaper.


I encourage anyone who uses it to pony up the $60 for it, after giving it a good trial. That very reasonable license fee will allow you to legally use the product for two full versions. Currently at V7, you will not have to renew until V9 comes out, likely 3-4 years from now.

[phil_er_up]

Thanks again If_then_else and ronmac for reaper recommendation.

Installed reaper and guess what it sees all my VST plugins including MAutoAlign and TBProAudio ABLM2 immediately without me setting path for the vst plugins. Arg. Been using wavelab for over 20 years so know it really well. Though reaper looks pretty good. Just will have to get use to it.

Ran a test took card and supercard recordings from the same mic stand and ran the  MAutoAlign and TBProAudio ABLM2 plugins. Yikes. Would A/B it with the card and supercards recordings and the output from Mautoalign. There was quite a difference in the A/B comparison. The  MAutoAlign seemed to fix the Alignment and Phase Correction between the 2 mics. Processed file sounded cleaner. The whole file seemed to come into focus. Seemed to fix some or all of the combing effect from 2 different types of cardiod mics.  Then ran another test with front of lip mics on stage with a SBD feed. Same thing. Processed file sounds cleaner and whole file again came into focus.

Will have to purchase reaper, MAutoAlign and TBProAudio ABLM2 plugins.

Thanks again If_then_else for sharing your plugins recommendations and ronmac for his help too.

==========================================================================================
Did not mean to hijack the thread with my problems with wavelab seeing the plugins.

Lets get back to what  If_then_else posted.

Any other plugins from that list you would recommend  If_then_else ?

[if_then_else]

Any other plugins from that list you would recommend  If_then_else ?

Depends on the musical genres you're recording, how often you're going to use the plugins and on your budget.
Many plug-in manufacturers charge exorbitant prices for their flagship products throughout the year and then discount them heavily at Easter, Black Friday or Christmas. If you don't need the plugins immediately, buy them at a time when they are on sale.

Apart from the (really decent) stock plugins of Reaper, there are a few good, free VST plugins. In many cases these are some stripped-down but fully working versions of the individual company's flagship products, e.g:

https://klanghelm.com/contents/freeware.html (MJUC jr., IVGI, DC1A)
https://www.tokyodawn.net/tokyo-dawn-labs/ (TDR Nova, TDR Kotelnikov)
https://techivation.com/t-de-esser/ (Techivation T-De-Esser 2)
https://www.izotope.com/en/products/free-audio-plug-ins.html (Edit: I forgot about iZotope Imager 2...)

Re. the "special" plugins category which are almost always part of my own workflow: I'd recommend the Kush Omega amp simulators 458a and TWK (priced at $29 each). I don't really use the other two. Again, this depends on the type of music you're recording.

https://thehouseofkush.com/products/omega-458a
https://thehouseofkush.com/products/omega-twk

The other plugin that I used a lot lately is TBProAudio DSQE3, the main alternative to OEKSound Soothe v2. This is a surgical, dynamic spectral EQ. Basically, its purpose is to remove digital hashness/muddiness/boominess. Be careful as it's easy to overdo things with this kind of plugins. TBProAudio never run sales but the plugin is fairly priced at €79. There is a newer, similar one from Three-Body-Technology (named "SpecCraft") that supposedly does the same thing and for which they're running an intro offer at $69. I haven't tried it as I'm more than happy with DSEQ3 - but it might be a valid alternative. When in doubt, just try it it. I wouldn't spend $200 on Soothe 2.

https://www.tbproaudio.de/products/dseq

In terms of limiters: Sonible Smart:Limit is really good, it also includes their metering technology and AI assistants. It's super easy to get to the loudness levels of professionally mastered albums. I think the regular price is €69. They're running occasional promos on their plugin bundles. So possibly worth waiting, if you're interested in their EQ and Compressor as well.

https://www.sonible.com/smartlimit/

[phil_er_up]

Thanks again for your thoughts on these plugins. Looked at all the ones you posted.

DL'ed https://klanghelm.com/contents/freeware.html (MJUC jr., IVGI, DC1A).
Not sure if I would use those or not.

Also got auto-align 2 to work in reaper too. Will try and compare output from MAutoAlign and auto-align 2.

Mostly record jam, jazz, reggae and blues. Though mostly the first 2.

[if_then_else]

These some screenshots of the effects / plugins I was talking about.

With DSEQ3, I tend to use the preset "Master - balance soft" that's basically transparent. (As previously mentioned, it's quite easy to overdo it with these dynamic spectral EQ plugins.) Select 4x/4x for oversampling (OS) and Ultra/Ultra for Quality. Depending on the source you'd either want to use "linear phase" (which is the default) or "natural phase" (NP). You can customise the frequency range(s) for which the equalisation will be applied.

Re. Kush Omega: These amp simulation plugins are great to make guitars, bass etc shine. Always make sure to enable "phase" and "-20dB" and, if necessary, reduce the input and/or output volume.

For Auto-Align 2: Just add the plugin to each of the mix busses, play back the audio in your DAW (it must be running for Auto-Align to analyse the tracks) and when done click on "Align".

[if_then_else]

Update: I also tested the demos of Soothe 2 and SpecCraft yesterday and DSEQ3 is clearly the better alternative for me, both in terms of sound and the usefulness of the presets that are included.

[phil_er_up]

Ran auto-align 2 and Mauto-align. Thought Mauto-align sounded good till I heard the auto-align 2. Auto-align 2 file sounded better IMO.
The low end of the file from the auto-align 2 processed file was there and sounded more natural the Mauto-align file sounded more mid-range and high end not very much low end for the file I was running. YMMV.

Also used TBProAudio DSEQ3 and it is a very good plugin and I like it and would use this one. Is easy to use and sound output was musical.

Tried Fabfilter Pro C2 and Fabfilter Pro L2 and both are really good though both are expensive.
Really liked the Pro C2 compressor and would see if I use the Pro L2 limiter that much though both seemed to do a really good job and sound output was musical.

Have anyone tried the fabfilter plugins and is it worth the $179 for each?

Saw online people raved about the Fabfilter Pro Q4 though have not tried that one yet.

=================================================================

Could not get Kush Omega and Sonible smartLimit to work. So can not comment on those.

[if_then_else]

Ran auto-align 2 and Mauto-align. Thought Mauto-align sounded good till I heard the auto-align 2. Auto-align 2 file sounded better IMO.
The low end of the file from the auto-align 2 processed file was there and sounded more natural the Mauto-align file sounded more mid-range and high end not very much low end for the file I was running. YMMV.

Also used TBProAudio DSEQ3 and it is a very good plugin and I like it and would use this one. Is easy to use and sound output was musical.

Tried Fabfilter Pro C2 and Fabfilter Pro L2 and both are really good though both are expensive.
Really liked the Pro C2 compressor and would see if I use the Pro L2 limiter that much though both seemed to do a really good job and sound output was musical.

Have anyone tried the fabfilter plugins and is it worth the $179 for each?

Saw online people raved about the Fabfilter Pro Q4 though have not tried that one yet.

=================================================================

Could not get Kush Omega and Sonible smartLimit to work. So can not comment on those.

The Fabfilter plugins are the de-facto industry standard. I've got them but the equivalent plugins by Sonible are pretty good as well and possibly better value. I think the current sale is still ongoing.

The reason why you might not have gotten the Kush and Sonible plugins to work might be iLok. You don't have to have an iLok USB key but you'd need to install iLok license manager and set up a free account to demo the plugins.

Fabfilter often run discounts and you will ge substantial loyalty discounts as well, if you own their other plugins. But yes, they're expensive. On the positive side, they don't use iLok.

[phil_er_up]

Was looking through your list again and saw this plugin:

Soundtheory Gullfoss: I-powered equalizer that dynamically enhances clarity and balance in real-time.

Have you used this plugin if_then_else?

Not sure if it would be worth installing the "try now" version. Have DL'ed so many other plugins that I need to run to see if I would purchase them before the time limit runs out on the free version. Found have to use a plugin multiple times to see what it really does and sometimes use it with different genres to see if it works the same on each of them.

[if_then_else]

It's one of those plugins that definitely have an effect - but in most cases, the effect is "homeopathic".

I.e. you can hear it in an A/B test but it's nothing that would make or break your mix. Given a fixed budget, I'd rather spend my money on the plugins that make the biggest difference.

[morst]

Anyone tried one of these automatic alignment tools on the two channels of a near-coincident or wider stereo pair?
I wonder if it makes a difference when the recording is played back in summed mono?

Edit: reading the manual for AA2 I see this:

"Keep in mind that AA2 won’t stretch or “warp” the audio in order to align it.
The Time Alignment algorithm simply applies Time Offsets to the playback of each track, allowing them to play tightly in sync."

However, it will do phase rotation...
"Even in Groups with large numbers of tracks, AA2 will continue to automatically determine the best combination of Time Offsets, polarity reversals and phase rotations needed to lock each track into the tightest possible phase relationship with the whole of the Group."

[Gutbucket]

Looking forward to trying those alignment tools myself. 

TLDR- short discussion of alignment esoterics. Feel free to ignore if you prefer to pretend it's simple.

This gets me wondering what alignment might be best for some stereo combinations.  Most of the time it's likely to be best straight-aligned (in reference to a planar wavefront arriving from 0-degrees directly in front)*, but I'm curious to hear from fellow tapers using these tools or doing it manually, of experiences of where a slightly different alignment works out best.  The three+ mic position arrangements suggested by the extended Stereo Zoom (see the Michael Williams papers, the Schoeps image assistant, 3-mic position improved PAS) push the center forward a bit to get the image linking across the playback stage to work best. To put it a different way, is it best to align for a planar wavefront arriving from directly ahead or align some of the elements for a wavefront arriving from the angle of the PA off to one side, and vice versa on the other side.. or the angle of the playback speakers.. or?  Impossible to align for all directions unless the mics are all coincident.


*But what is "straight-aligned"? Any non-coincident two microphone array with spacing between the microphone pair can only be fully aligned along one axis of arrival (lets assume directly forward/backward).  Yet sound arrives from all angles.  Even if we don't care about or perhaps even desire a useful misalignment of the far off-axis stuff, a simple 2-mic microphone config using spacing between the mics will not be fully aligned for sources positioned somewhat over to either side of the alignment axis.  Introduce additional channels or pairs and it can get very complex, particularly if the microphones are not all arranged along a single line as many stereo multi-microphone configurations suggest.  We might arrange and align for arrival from only one way along the axis - say directly forward but not backward.  And the non-alignment relationship between channels for differnt off-axis angles will shift around in various ways.

Aligning mono channels that are positioned non-coincidently (rather than spaced stereo pairs) is a bit more conceptually straightforward and probably the original intent of manu of these tools, yet still when the sound sources are distributed in space one can only align for a single source among them.  If a aligning a bunch of close mics on a drum kit, one presumably needs to choose which drum (snare? kick?) on which to base the alignment and let the alignment of all others fall where they may.

I don't mean to cast shade on these amazing tools, and enjoy thinking through and discussing the implications.  As always, what sounds right is right.  You needn't think too deeply about all this.  Just listen carefully when making the decisions to be sure of doing more good than harm.  Please let me know if/when you come across outlier situations when using these tools. Thanks.

[phil_er_up]

Would the auto-align2 fix the combing effect with 2 sets of cardioid  mics from the same mic stand...like one set of cards with Supercardioid  or wide card?
Also, like the idea of using it with SBD and aud recordings.

Yes. That's actually my use case. I mainly record 4 or 5 audience channels + a stereo soundboard feed and matrix them in post.
Cf. the screenshot below. The darker areas in the Correlation diagram show the corrections.

In your screenshot of your DAW you have a "tape" plugin at the end of bus one and bus two plugin stack.
What is that "tape" plugin?

[morst]

I don't mean to cast shade on these amazing tools, and enjoy thinking through and discussing the implications.  As always, what sounds right is right.  You needn't think too deeply about all this.  Just listen carefully when making the decisions to be sure of doing more good than harm.  Please let me know if/when you come across outlier situations when using these tools. Thanks.

No matter what they say, no matter what they do,
Gonna end up in the middle of that rock & roll stew

~Rock & Roll Stew [Traffic]

[Gutbucket]

 

[if_then_else]

Would the auto-align2 fix the combing effect with 2 sets of cardioid  mics from the same mic stand...like one set of cards with Supercardioid  or wide card?
Also, like the idea of using it with SBD and aud recordings.

Yes. That's actually my use case. I mainly record 4 or 5 audience channels + a stereo soundboard feed and matrix them in post.
Cf. the screenshot below. The darker areas in the Correlation diagram show the corrections.

In your screenshot of your DAW you have a "tape" plugin at the end of bus one and bus two plugin stack.
What is that "tape" plugin?

Those "tape plugins" can help "glue" audio tracks together and add a polished, cohesive sound to the mix.
Think of these like one type of (gentle) saturation plugins...

Tape plugins emulate the (subtle) harmonic distortion of specific analog tape machines:

• They add warmth and depth to digital recordings
• They smooth out harsh transients, creating a more cohesive sound
• They can be used on individual tracks, busses, or the master channel

I think there was a thread somewhere on here where people used real tape machines to add that kind of "warmth". High-end tape machines have become really expensive these days. These are the software equivalents. I don't know if they sound exactly like the specific types of hardware tape machines but at least they lend the same type of "warmth" to a mix/master.

There are even a few decent free tape plugins. Proper reply after work.

[if_then_else]

Re. tape plugins: There are some decent free ones. So, try them out first:

• Chow Tape Model: https://chowdsp.com/products.html
• FerricTDS mkII (version 2): https://varietyofsound.wordpress.com/2009/10/29/ferrictds-released-today-withi-the-kvr-dc09/ (This is was originally created for 32-bit DAWs. In the meantime Variety Of Sound also created a 64-bit version which is available from the website's Downloads section.)

Among the commercial plug-ins, which are often designed to emulate specific hardware, there are some that are more suitable for the mix bus and others that are more suitable for the master bus. Some sound “dirtier” (suitable for heavy metal or noise rock) and others sound cleaner.

An example of tape plug-ins for the mix bus would be the Studer A800 tape recorder emulation from UAD. I personally don't like it at all in the master bus, but it sounds good for electric guitars in the mix bus. For the master, for example, UAD's Ampex ATR-102 would be a good choice. The tape plug-ins from u-he (“Satin”) or Softube (“Tape”) would be suitable for both the mix bus and the master bus. In general: If you decide on one of these commercial plug-ins, only buy them on sale. Some of them are reduced from over a hundred dollars to 30 or 40 dollars. I wouldn't spend more than that under any circumstances.

As for Gullfoss: As already mentioned, this is more of a plugin that makes small cosmetic improvements. Don't push it to its limits. Gullfoss comes in three versions: one for individual instruments, one with an optimized footprint for real-time applications (e.g. for live mixes) and the master version. I have only ever used the master version. The parameters are explained quite well here: https://www.soundonsound.com/reviews/soundtheory-gullfoss.

The two main parameters are Recover and Tame. Recover accentuates those elements that are in danger of getting masked by other sounds, while Tame suppresses those over-dominant sounds responsible for the masking. Both have their own amount percentage controls, and a third control called Bias determines how Gullfoss reacts in situations where it is less clear whether Recover or Tame would be most appropriate. Positive Bias values favour the Recovery mode at lower signal levels, while negative values lean towards the Tame mode. While Gullfoss matches the subjective level of the output to the input, it is designed not to adversely impact the dynamics of the treated signal — indeed, in the case of an over-compressed or over-limited signal, a useful degree of dynamic improvement can be made to transients. This was certainly borne out in my tests, where some delicate percussion that was in danger of getting buried in the mix came across much more cleanly.

The next control, Brighten, adjusts the overall tonality of the processed signal; as its name suggests, it can be used to lift the top end, but it can also be set to a negative value to reduce the highs if the source is already too bright. Finally, there’s a Boost setting, but this isn’t just a level control; it also affects the overall tonal balance in a way that reflects the way human hearing perception changes with level. Its action is to some extent based on the familiar Fletcher-Munson equal loudness curves, but there’s more to it than that, as it also has a dynamic impact on the EQ rather than applying a static curve — increasing Boost emphasises bass frequencies while cutting back the mids. Actual output level can be adjusted using the vertical slider at the right of the plug-in window.

[phil_er_up]

Thanks If_then_else for the info, pics and links.

Do have 2 tape plugins currently - Ozone vintage tape and PSP vintage warmer 2.

Early on in taping career used the PSP vintage warmer 2 though not sure if I was improving the sound or not. Never used the Ozone vintage tape until yesterday after reading this thread. Never thought about using it at the end of the bus stack. So this thread to me is about the VST plugin though also where and what ones to use in the bus 1, bus 2, master bus etc. This info could help newbies think about which plugins to use and where in your bus stacks. Good stuff.

You are very helpful and really thank you for that.

[if_then_else]

Thanks. I'm not going to pollute this thread further with my specific preferences in terms of plugins or workflows. This was just intended as an overview of some plugins I found useful. I understand that people's priorities and preferences in terms of post-processing are highly subjective and the kind of music I'm recording as well as matrices/multi-track recordings might only appeal to a certain subgroup of people here. That said, I've got two shows coming up next month where I've got open taping (+ filming) permission. I'll try to document my post-processing workflow in a separate thread while I'm on it. Just in case that it might be helpful to others.

[Gutbucket]

Thanks for the detailed info if_then_else. It's great to get this kind of insight into what fellow TS'ers are currently using, how they are using it and for what. Particularly so due to this area constantly evolving with the introduction of new tools.

[opsopcopolis]

Ran auto-align 2 and Mauto-align. Thought Mauto-align sounded good till I heard the auto-align 2. Auto-align 2 file sounded better IMO.
The low end of the file from the auto-align 2 processed file was there and sounded more natural the Mauto-align file sounded more mid-range and high end not very much low end for the file I was running. YMMV.

Also used TBProAudio DSEQ3 and it is a very good plugin and I like it and would use this one. Is easy to use and sound output was musical.

Tried Fabfilter Pro C2 and Fabfilter Pro L2 and both are really good though both are expensive.
Really liked the Pro C2 compressor and would see if I use the Pro L2 limiter that much though both seemed to do a really good job and sound output was musical.

Have anyone tried the fabfilter plugins and is it worth the $179 for each?

Saw online people raved about the Fabfilter Pro Q4 though have not tried that one yet.

=================================================================

Could not get Kush Omega and Sonible smartLimit to work. So can not comment on those.

The Fabfilter stuff is absolutely worth the money IMO. Especially the EQ, Comp, and Limiter. Pro Q4 is a game changer for audience tapes imo. Allows you to perform wide q boosts, while also dynamically reducing resonances

[Billy Mumphrey]

Thanks. I'm not going to pollute this thread further with my specific preferences in terms of plugins or workflows. This was just intended as an overview of some plugins I found useful. I understand that people's priorities and preferences in terms of post-processing are highly subjective and the kind of music I'm recording as well as matrices/multi-track recordings might only appeal to a certain subgroup of people here. That said, I've got two shows coming up next month where I've got open taping (+ filming) permission. I'll try to document my post-processing workflow in a separate thread while I'm on it. Just in case that it might be helpful to others.

Feel free to share any and all preferences, in this thread or another as I've been (slowly) following along and learning A LOT. The pace and progress of these digital tools is scary, in (mostly) good and bad ways.

The alignment tools are of particular interest, as they do seem to contradict what we try to do with stereo imaging: introduce time arrival differences for a fuller, more complete sound. I do believe in going with what sounds best, and look forward to hopefully trying these out myself.

[if_then_else]

Speaking of (multi-)track alignment (not phase-alignment): There seem to be some freeware/open source programs that are supposed to handle this. Among others, two stand-alone projects on github and a Nyquist plugin for Audacity. I haven't tried any of them yet, but I'll definitely have a look at the software.

https://audionyq.com/aligning-multiple-tracks/
https://github.com/protyposis/AudioAlign
https://github.com/benfmiller/audalign/

[morst]

Speaking of (multi-)track alignment (not phase-alignment): There seem to be some freeware/open source programs that are supposed to handle this. Among others, two stand-alone projects on github and a Nyquist plugin for Audacity. I haven't tried any of them yet, but I'll definitely have a look at the software.

https://audionyq.com/aligning-multiple-tracks/
https://github.com/protyposis/AudioAlign
https://github.com/benfmiller/audalign/

Looks like a good start.




Potential workflow:

• [size=78%]Align near start of show using plugin[/size]
• [size=78%]Mark start or end points of every track[/size]
• [size=78%]Align near end of show using plugin[/size]
• [size=78%]Feed offsets into my spreadsheet[/size]

https://docs.google.com/spreadsheets/d/1pQGfYwPgBFFzcY5m6aRj-Zbu9HsRumLy-tJB1d8Eufg/edit?gid=583050244#gid=583050244

• [size=78%]Sync[/size]
• [size=78%]Save work[/size]
• [size=78%]Listen and see if everything is correct...[/size]

I really hate the way this website changed at some point...

[Gutbucket]

More on channel alignment (please ignore this post if not your bag)-

A very interesting way to gain additional insight into some of the peculiarities of channel alignment is to mess around with applying a very slight delay to the Side (difference) channel verses the Mid (Sum) channel and listen for what that does.  The recording need not have been be made with a Mid/Side or X/Y microphone setup.  You can use any stereo recording by simply converting it to Mid/Side, applying the delay and converting back to L/R again.  That's most easily done in a single time delay plugin which includes Mid/Side routing options as well as Left/Right.  Here's a free one that can do that, Voxengo Sound Delay- https://www.voxengo.com/product/sounddelay/

Tiny delays of a fraction of a millisecond do interesting things in changing the perception of depth in the recording, particularly over headphones. Too much screws up the left right imaging while creating an overemphasised sense of font-back depth.  Way too much will create an echo effect. But a just touch can help bring the imaging of headphone listening "out of head" and increase the sense of 3d dimension without altering the Left/Right imaging significantly.  Whether one wants to do this to their recordings or not isn't really the point, it's more about hearing what it does to the perceived stereo imaging to gain insight into the nature of time alignment and how the best sounding alignment may not be one that is perfectly in-phase (which for a spaced pair of microphones can only be achieved along a single axis of arrival, as discussed in previous posts above), and how the choice alignment can legitimately vary over some range.

This all came to mind again today upon coming across this Youtube video- Headphones Are Not Stereo (mid side phase trickery) https://youtu.be/uZ9WQDojQt8?si=2UuKHi4Ecbq_z4CZ
The most relevant part in that video about the effect of delayed time/phase alignment starts here- https://youtu.be/uZ9WQDojQt8?si=jYi74B0U_i3VlMmi&t=442
If you need a primer on how sum/difference(Mid/Side) stereo works, just back up a bit along the timeline.

[Gutbucket]

^
This is also related to how alignment is intentionally varied in some microphone arrays by positioning the center microphone or center stereo pair slightly forward of a flanking pair, as we've discussed in other threads.  A wave-front arriving from directly ahead will reach the center microphone or pair a fraction of a millisecond before the flanking mics. When mixed down to stereo, that center mic or pair provides the majority of the Mid (sum) stereo content while the flanking mics provide most of the Side (difference) stereo content.  Positioning the center mic position slightly forward of the flanking pair effectively delays arrival at the flanking pair that is providing the majority of the Side (difference) content by some fraction of a millisecond.

I don't think I made this particular association clear in my posts when we were discussing center mic position in those other treads, although we did certainly talk around it. It similarly came up when discussing the extension of Improved PAS technique from 2 channels to additionally include PAS solutions for 3-microphone positions for the use of 3 or 4 microphones.  In that case the microphone angle/spacing/position solutions from the Michael Williams multi-mic Stereo Zoom data, the Sengpielaudio Visualizer and the Schoeps Image Assistant all call for pushing the center microphone position forward of the flanking mics by something like ~8-20 cm or so, as do most surround recording microphone configurations.  That effectively ends up sort of creating two axes of in-phase alignment angled somewhat right and left of center, rather than a single axis for sound arriving from directly ahead. (More technically correct - one axis angled right of center where the right spaced mic and center mic (or pair) is in full alignment, and another angled left of center where the left mic and center mic (or pair) are in alignment.)

Here are a few of the threads in which we've discussed center microphone (or pair) positioning:
> Relationships between spaced omnis and center mic- https://taperssection.com/index.php?topic=182579.0
> 3 mic (LCR) distance- https://taperssection.com/index.php?topic=200878.0
> Re: Improved PAS technique- now as setup diagrams & extended to 3-4 microphones- https://taperssection.com/index.php?topic=167549.msg2411846#msg2411846
(fair warning, some of the above threads get wordy!)

[if_then_else]

I've just tried my luck with Audalign (https://github.com/benfmiller/audalign/). There's good news and bad news. The good news is that it works perfectly. The bad news: it is not trivial to install.

The installation instructions on the project page should be taken with a grain of salt. If you install the whole thing with a current Python version (e.g. under a current Anaconda on Windows or Python 3.12 or 3.13 on Linux), you will encounter errors because some versions of the libraries used are no longer compatible with each other.

The last update of Audalign released a year ago is specific to Python 3.11.

The solution is to create a virtual environment with Python 3.11 and install the modules. I did this on Linux, it should work similarly on Windows:

Code: [Select]

pyenv install 3.11
/home/markus/.pyenv/versions/3.11.10/bin/python3.11 -m pip install --upgrade pip
/home/markus/.pyenv/versions/3.11.10/bin/pip3 install audalign
/home/markus/.pyenv/versions/3.11.10/bin/pip3 install audalign[visrecognize]

Do not install audalign[noisereduce]. It will pull some *big* Nvidia specific modules that would probably require a Nvidia graphics card and drivers.

This is a small sample Python script to align two files that are located in the folder "bak" and places the aligned files in a directory called "destination".

Code: [Select]

import audalign as ad

# Create a recognizer (in this case, using the FingerprintRecognizer)
fingerprint_rec = ad.FingerprintRecognizer()

# Optionally, configure the recognizer
fingerprint_rec.config.set_accuracy(3)

# Align files in a folder
results = ad.align("target/folder/", recognizer=fingerprint_rec)

# Or align specific files
results = ad.align_files(
    "bak/20160117_CHVE_44.1Khz-16bit.wav",
    "bak/AUD.wav",
    destination_path="destination/",
    recognizer=fingerprint_rec
)

# For fine-tuning the alignment
fine_results = ad.fine_align(
    results,
    recognizer=ad.CorrelationSpectrogramRecognizer()
)

Call the script with Python 3.11 from your virtualenv environment

Code: [Select]

/home/markus/.pyenv/versions/3.11.10/bin/python ./align_files.py

The sample output would look similar to the following.

Code: [Select]

Directory contains 0 files or could not be found
No matches detected
0 out of 0 found and aligned
Fingerprinting 20160117_CHVE_44.1Khz-16bit.wav
Fingerprinting AUD.wav
Finished fingerprinting 20160117_CHVE_44.1Khz-16bit.wav
Finished fingerprinting AUD.wav
20160117_CHVE_44.1Khz-16bit.wav: Finding Matches...  Aligning matches
AUD.wav: Finding Matches...  Aligning matches
Writing destination/20160117_CHVE_44.1Khz-16bit.wav
Writing destination/AUD.wav
Writing destination/total.wav
2 out of 2 found and aligned

Total fingerprints: 10395127
Fine Aligning...
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 3480.75it/s]
Comparing AUD.wav against 20160117_CHVE_44.1Khz-16bit.wav...
Comparing 20160117_CHVE_44.1Khz-16bit.wav against AUD.wav...
Calculating correlation... Finding Local Maximums... done
Calculating correlation... Finding Local Maximums... done
2 out of 2 found and aligned

I've loaded the original files in Reaper. You can see that the original files (track 1 and 2) are unaligned.
The two tracks taken from the "destination" directory are aligned though. One of the files (track 2) has been padded with silence at the beginning.

Again, this is for "track align" purposes. For phase alignment, you might want to use one of the phase alignment plugins that have been discussed earlier in this thread.

Oh, and unless I've done anything wrong, the output in destination appears to be in mono. (Which would be backed by the animated gif / demo on the project website. I'm going to contact the author about it anyways.) You might want to use mono stems as input. This could also be scripted:

Code: [Select]

ffmpeg -i input.wav -filter_complex "[0:a]channelsplit=channel_layout=stereo[left][right]" -map "[left]" left.wav -map "[right]" right.wav
To join two mono channels back to stereo:

Code: [Select]

ffmpeg -i left.wav -i right.wav -filter_complex "[0:a][1:a]amerge=inputs=2[a]" -map "[a]" output_stereo.wav
Alternatively, there is an internal workaround with Audalign:
https://github.com/benfmiller/audalign/issues/67

I'll try to come up with a Python script tailored to "our" workflow when I have time.

[if_then_else]

So... I've tested the revised script on two soundboard/audience matrix recordings and I'm now pretty confident that it works. But it was a pretty frustrating experience overall.

I tried the two fixes/workarounds suggested in the github issue and neither of them worked. After that I tried it with the help of ChatGPT 4o and Claude and it was maddening, because the AI was so stupid. Some of the workarounds suggested by the AI drove my laptop to a memory usage of 100% (32 GiB).

Since neither the library nor the LLMs could reliably generate stereo files, I had the input files split into mono channels, which I then reassembled at the end. This works fine as long as one of the channels of the reference file does not contain only silence. I think most of the relevant cases are now covered by the Python script below.

Code: [Select]

import os
import sys
import argparse
import logging
from pydub import AudioSegment
from audalign import FingerprintRecognizer, align_files

class LoggerWriter:
    """Redirects stdout/stderr to logging system with timestamps"""
    def __init__(self, level):
        self.level = level
   
    def write(self, message):
        if message.strip():
            logging.log(self.level, message.strip())
   
    def flush(self):
        pass

def configure_logging(destination_dir):
    """Set up logging to both console and file with timestamps"""
    logger = logging.getLogger()
    logger.setLevel(logging.INFO)

    # Clear existing handlers
    for handler in logger.handlers[:]:
        logger.removeHandler(handler)

    # Create formatter with timestamps
    formatter = logging.Formatter('%(asctime)s - %(levelname)s - %(message)s')

    # File handler (log to destination directory)
    log_file = os.path.join(destination_dir, 'alignment.log')
    file_handler = logging.FileHandler(log_file)
    file_handler.setFormatter(formatter)
    logger.addHandler(file_handler)

    # Console handler
    console_handler = logging.StreamHandler()
    console_handler.setFormatter(formatter)
    logger.addHandler(console_handler)

    # Redirect standard streams
    sys.stdout = LoggerWriter(logging.INFO)
    sys.stderr = LoggerWriter(logging.ERROR)

    logging.info(f"Logging system initialized. Full log at {log_file}")

def list_root_wav_files(directory):
    """List all .wav files in directory root"""
    try:
        return [
            os.path.join(directory, f)
            for f in os.listdir(directory)
            if os.path.isfile(os.path.join(directory, f)) and f.lower().endswith(".wav")
        ]
    except Exception as e:
        logging.error(f"File listing error: {e}")
        return []

def split_stereo_to_mono(input_file, temp_dir):
    """Split stereo file to mono channels with existing file check"""
    try:
        base_name = os.path.splitext(os.path.basename(input_file))[0]
        left_output = os.path.join(temp_dir, f"{base_name}_L.wav")
        right_output = os.path.join(temp_dir, f"{base_name}_R.wav")
       
        # Check if files already exist
        if os.path.exists(left_output) and os.path.exists(right_output):
            logging.info(f"Using existing split files for '{input_file}':")
            logging.info(f"  - Left channel: '{left_output}'")
            logging.info(f"  - Right channel: '{right_output}'")
            return left_output, right_output
           
        audio = AudioSegment.from_file(input_file)
        channels = audio.channels

        if channels == 1:
            # If mono, check if temp file exists
            mono_output = os.path.join(temp_dir, os.path.basename(input_file))
            if os.path.exists(mono_output):
                logging.info(f"Using existing mono file: {mono_output}")
                return mono_output, None
               
            # If not exists, create it
            audio.export(mono_output, format="wav")
            logging.info(f"Copied mono file '{input_file}' to '{mono_output}'.")
            return mono_output, None

        elif channels == 2:
            # Split stereo into left and right channels
            left_channel = audio.split_to_mono()[0]
            right_channel = audio.split_to_mono()[1]

            left_channel.export(left_output, format="wav")
            right_channel.export(right_output, format="wav")

            logging.info(f"Split stereo file '{input_file}' into:")
            logging.info(f"  - Left channel: '{left_output}'")
            logging.info(f"  - Right channel: '{right_output}'")

            return left_output, right_output

    except Exception as e:
        logging.error(f"Splitting error: {e}")
        return None, None

def align_channels(ref_path, input_left, input_right, dest_dir, channel_mode, temp_dir):
    """Align channels with reference using specified mode"""
    try:
        recognizer = FingerprintRecognizer()
        ref_audio = AudioSegment.from_file(ref_path)
       
        # Handle reference channels
        ref_base = os.path.splitext(os.path.basename(ref_path))[0]
        ref_left = os.path.join(temp_dir, f"{ref_base}_L.wav")
        ref_right = os.path.join(temp_dir, f"{ref_base}_R.wav")
       
        if ref_audio.channels == 1:
            ref_left = ref_path
            ref_right = ref_path
            logging.info("Using mono reference for both channels")
        else:
            if channel_mode == 'L':
                ref_audio.split_to_mono()[0].export(ref_left, format="wav")
                ref_right = ref_left
                logging.info("Using left reference channel for all alignments")
            elif channel_mode == 'R':
                ref_audio.split_to_mono()[1].export(ref_right, format="wav")
                ref_left = ref_right
                logging.info("Using right reference channel for all alignments")
            else:
                ref_audio.split_to_mono()[0].export(ref_left, format="wav")
                ref_audio.split_to_mono()[1].export(ref_right, format="wav")
                logging.info("Using stereo reference channels independently")

        # Perform alignments
        if input_left:
            align_files(
                filename_a=str(ref_left),
                filename_b=str(input_left),
                destination_path=dest_dir,
                write_extension="_L_aligned.wav",
                write_multi_channel=False,
                recognizer=recognizer
            )
        if input_right:
            align_files(
                filename_a=str(ref_right),
                filename_b=str(input_right),
                destination_path=dest_dir,
                write_extension="_R_aligned.wav",
                write_multi_channel=False,
                recognizer=recognizer
            )

    except Exception as e:
        logging.error(f"Alignment failed: {e}")
        raise

def rename_aligned_files(destination_dir):
    """Renames aligned mono files to follow the desired naming convention."""
    try:
        for filename in os.listdir(destination_dir):
            if "_L._L_aligned.wav" in filename:
                new_name = filename.replace("_L._L_aligned", "_L_aligned")
                os.rename(
                    os.path.join(destination_dir, filename),
                    os.path.join(destination_dir, new_name),
                )
                logging.info(f"Renamed '{filename}' to '{new_name}'.")
            elif "_R._R_aligned.wav" in filename:
                new_name = filename.replace("_R._R_aligned", "_R_aligned")
                os.rename(
                    os.path.join(destination_dir, filename),
                    os.path.join(destination_dir, new_name),
                )
                logging.info(f"Renamed '{filename}' to '{new_name}'.")
    except Exception as e:
        logging.error(f"An error occurred during renaming: {e}")
        raise

def merge_aligned_channels(destination_dir):
    """Merges aligned left and right mono channels into stereo files."""
    try:
        aligned_files = [f for f in os.listdir(destination_dir) if f.endswith("_aligned.wav")]
       
        grouped_files = {}
        for f in aligned_files:
            base_name = f.split("_")[0]
            if base_name not in grouped_files:
                grouped_files[base_name] = {}
           
            if "_L_aligned" in f:
                grouped_files[base_name]["left"] = os.path.join(destination_dir, f)
            if "_R_aligned" in f:
                grouped_files[base_name]["right"] = os.path.join(destination_dir, f)

        for base_name, channels in grouped_files.items():
            if "left" in channels and "right" in channels:
                output_stereo = os.path.join(destination_dir, f"{base_name}_aligned_stereo.wav")
               
                left_audio = AudioSegment.from_file(channels["left"])
                right_audio = AudioSegment.from_file(channels["right"])
                stereo_audio = AudioSegment.from_mono_audiosegments(left_audio, right_audio)
                stereo_audio.export(output_stereo, format="wav")
               
                logging.info(f"Merged '{channels['left']}' and '{channels['right']}' into stereo file '{output_stereo}'.")
               
                os.remove(channels["left"])
                os.remove(channels["right"])
                logging.info(f"Deleted intermediate mono files")

    except Exception as e:
        logging.error(f"An error occurred during merging: {e}")
        raise

def process_workflow(input_dir, reference_basename, temp_dir, destination_dir, channel_mode='auto'):
    """Full workflow: Split -> Align -> Rename -> Merge"""
    os.makedirs(temp_dir, exist_ok=True)
    os.makedirs(destination_dir, exist_ok=True)
   
    ref_path = os.path.join(input_dir, reference_basename)
    wav_files = list_root_wav_files(input_dir)
   
    had_errors = False
   
    try:
        for wav_file in wav_files:
            if os.path.basename(wav_file) == reference_basename:
                continue
            try:
                input_left, input_right = split_stereo_to_mono(wav_file, temp_dir)
                if input_left is None and input_right is None:
                    had_errors = True
                    continue
                   
                align_channels(ref_path, input_left, input_right, destination_dir, channel_mode, temp_dir)
            except Exception as e:
                had_errors = True
                logging.error(f"Failed to process file {wav_file}: {e}")
                continue
           
        try:
            rename_aligned_files(destination_dir)
        except Exception as e:
            had_errors = True
            logging.error(f"Failed during renaming: {e}")
           
        try:
            merge_aligned_channels(destination_dir)
        except Exception as e:
            had_errors = True
            logging.error(f"Failed during merging: {e}")
           
        if had_errors:
            logging.warning("Processing completed with errors")
        else:
            logging.info("Processing completed successfully")
       
    except Exception as e:
        logging.critical(f"Workflow failed: {e}")
        raise

def main():
    parser = argparse.ArgumentParser(description="Audio alignment workflow")
    parser.add_argument("-i", "--input", required=True, help="Input directory with WAV files")
    parser.add_argument("-r", "--reference", required=True, help="Reference filename")
    parser.add_argument("-t", "--temp", required=True, help="Temporary directory path")
    parser.add_argument("-o", "--destination", required=True, help="Output directory")
    parser.add_argument("-c", "--channel", choices=['L','R','auto'], default='auto',
                      help="Reference channel selection")

    args = parser.parse_args()
   
    # Create directories and configure logging
    os.makedirs(args.destination, exist_ok=True)
    configure_logging(args.destination)

    try:
        process_workflow(args.input, args.reference, args.temp, args.destination, args.channel)
    except Exception as e:
        logging.critical(f"Fatal error in main workflow: {e}")
        sys.exit(1)

if __name__ == "__main__":
    main()

Parameters:

The script accepts four required command-line parameters:

Input Directory (-i, --input)

• Path to the directory containing WAV files to be processed
• Contains both the reference file and files to be aligned
• Only processes WAV files in the root directory (subdirectories are ignored)

Reference File (-r, --reference)

• Basename (filename) of the reference WAV file
• This file serves as the timing reference for aligning other files
• Must be located in the input directory
• Can be either mono or stereo

Temporary Directory (-t, --temp)

• Path where intermediate mono files are stored
• Used for storing split channels from stereo files
• Stores both reference and input file channels during processing
• Files in this directory are reused if they already exist

Destination Directory (-o, --destination)

• Path where the final aligned files are saved
• Contains the alignment log file
• Stores the final merged stereo files after alignment
• Intermediate aligned mono files are created here before merging

Optional Channel Mode (-c, --channel)

• Specifies which channel to use for alignment: 'L' (left), 'R' (right), or 'auto'
• In 'auto' mode, left channels are aligned with left, right with right
• When using 'L' or 'R', all channels are aligned against the specified reference channel
• Particularly useful when one channel of the reference might be silent or corrupted

Example: I think 3 minutes for the automated Alignment of a SBD/AUD matrix aren't too bad. I think memory usage didn't exceed 7-8 GiB with this algorithm. Of course, all of this wouldn't have been necessary if the Audalign library supported stereo output out of the box.

Code: [Select]

/home/markus/.pyenv/versions/3.11.10/bin/python ./trackalign.py  -i bak/  -t temp/ -o destination -r

20160117_CHVE_44.1Khz-16bit.wav -c auto
2025-02-01 14:17:24,761 - INFO - Logging system initialized. Full log at destination/alignment.log
2025-02-01 14:17:52,828 - INFO - Split stereo file 'bak/AUD.wav' into:
2025-02-01 14:17:52,828 - INFO -   - Left channel: 'temp/AUD_L.wav'
2025-02-01 14:17:52,828 - INFO -   - Right channel: 'temp/AUD_R.wav'
2025-02-01 14:17:54,119 - INFO - Using stereo reference channels independently
2025-02-01 14:17:54,484 - INFO - Fingerprinting 20160117_CHVE_44.1Khz-16bit_L.wav
2025-02-01 14:17:55,574 - INFO - Fingerprinting AUD_L.wav
2025-02-01 14:18:56,511 - INFO - Finished fingerprinting AUD_L.wav
2025-02-01 14:19:01,108 - INFO - Finished fingerprinting 20160117_CHVE_44.1Khz-16bit_L.wav
2025-02-01 14:19:01,635 - INFO - 20160117_CHVE_44.1Khz-16bit_L.wav: Finding Matches...
2025-02-01 14:19:01,844 - INFO - Aligning matches
2025-02-01 14:19:01,929 - INFO - AUD_L.wav: Finding Matches...
2025-02-01 14:19:02,041 - INFO - Aligning matches
2025-02-01 14:19:03,873 - INFO - Writing destination/20160117_CHVE_44.1Khz-16bit_L._L_aligned.wav
2025-02-01 14:19:03,954 - INFO - Writing destination/AUD_L._L_aligned.wav
2025-02-01 14:19:04,802 - INFO - Writing destination/total._L_aligned.wav
2025-02-01 14:19:04,891 - INFO - 2 out of 2 found and aligned
2025-02-01 14:19:04,891 - INFO - Total fingerprints: 949603
2025-02-01 14:19:05,282 - INFO - Fingerprinting 20160117_CHVE_44.1Khz-16bit_R.wav
2025-02-01 14:19:06,359 - INFO - Fingerprinting AUD_R.wav
2025-02-01 14:20:07,225 - INFO - Finished fingerprinting AUD_R.wav
2025-02-01 14:20:12,142 - INFO - Finished fingerprinting 20160117_CHVE_44.1Khz-16bit_R.wav
2025-02-01 14:20:12,699 - INFO - 20160117_CHVE_44.1Khz-16bit_R.wav: Finding Matches...
2025-02-01 14:20:12,820 - INFO - Aligning matches
2025-02-01 14:20:12,896 - INFO - AUD_R.wav: Finding Matches...
2025-02-01 14:20:13,145 - INFO - Aligning matches
2025-02-01 14:20:14,859 - INFO - Writing destination/20160117_CHVE_44.1Khz-16bit_R._R_aligned.wav
2025-02-01 14:20:14,943 - INFO - Writing destination/AUD_R._R_aligned.wav
2025-02-01 14:20:15,797 - INFO - Writing destination/total._R_aligned.wav
2025-02-01 14:20:15,881 - INFO - 2 out of 2 found and aligned
2025-02-01 14:20:15,881 - INFO - Total fingerprints: 1898676
2025-02-01 14:20:15,983 - INFO - Renamed 'AUD_R._R_aligned.wav' to 'AUD_R_aligned.wav'.
2025-02-01 14:20:15,983 - INFO - Renamed '20160117_CHVE_44.1Khz-16bit_L._L_aligned.wav' to '20160117_CHVE_44.1Khz-16bit_L_aligned.wav'.
2025-02-01 14:20:15,983 - INFO - Renamed '20160117_CHVE_44.1Khz-16bit_R._R_aligned.wav' to '20160117_CHVE_44.1Khz-16bit_R_aligned.wav'.
2025-02-01 14:20:15,983 - INFO - Renamed 'AUD_L._L_aligned.wav' to 'AUD_L_aligned.wav'.
2025-02-01 14:20:16,859 - INFO - Merged 'destination/20160117_CHVE_44.1Khz-16bit_L_aligned.wav' and 'destination/20160117_CHVE_44.1Khz-16bit_R_aligned.wav' into stereo file 'destination/20160117_aligned_stereo.wav'.
2025-02-01 14:20:16,906 - INFO - Deleted intermediate mono files
2025-02-01 14:20:17,889 - INFO - Merged 'destination/AUD_L_aligned.wav' and 'destination/AUD_R_aligned.wav' into stereo file 'destination/AUD_aligned_stereo.wav'.
2025-02-01 14:20:17,932 - INFO - Deleted intermediate mono files
2025-02-01 14:20:18,804 - INFO - Merged 'destination/total._L_aligned.wav' and 'destination/total._R_aligned.wav' into stereo file 'destination/total._aligned_stereo.wav'.
2025-02-01 14:20:18,853 - INFO - Deleted intermediate mono files
2025-02-01 14:20:18,855 - INFO - Processing completed successfully


[if_then_else]

I've created a stand-alone Windows version of my command-line tool for the track alignment. (I had to compile it from Python to C++ and into an *.exe file. So there is some performance overhead compared to the original Python script and the file size is also quite big. About 50 MB. Sorry, there was no other way. I had to download GBs worth of dependencies amd C++ compilers to compile it on Windows in the first place.)

EDIT: I uploaded the binary stand-alone version to GitHub releases:

https://github.com/mmathis76/TrackAlign/releases/tag/v1.0.0

The docs and the Python code are also available on Github:

https://github.com/mmathis76/TrackAlign

[phil_er_up]

I've created a stand-alone Windows version of my command-line tool for the track alignment. (I had to compile it from Python to C++ and into an *.exe file. So there is some performance overhead compared to the original Python script and the file size is also quite big. About 50 MB. Sorry, there was no other way. I had to download GBs worth of dependencies amd C++ compilers to compile it on Windows in the first place.)

EDIT: I uploaded the binary stand-alone version to GitHub releases:

https://github.com/mmathis76/TrackAlign/releases/tag/v1.0.0

The docs and the Python code are also available on Github:

https://github.com/mmathis76/TrackAlign

Thank you for this. It is appreciated. Will take a look at it.

Tried to get one of the other "align" tools on github and could not get it to work. See what you mean by tricky to install.

[if_then_else]

Running the exe file I had created is straightforward. That was the very reason I created it.

If you wanted to run the original Python script (e.g. because it's more memory efficient or because you prefer one of the other supported alignment algorithms) things are a bit different.

I can post the specific installation steps to get it working here, but it was even more complicated on Windows than on Linux: To install one of the optional components of Audalign you would need a C++ compiler to compile the code. Linux comes with gcc. On Windows you would have to Install the free C++ compilers from Visual studio and Clang. In other words, several GB worth of dependencies.

The exe file I uploaded to github is quite big but it comes with all of the dependencies out of the box. Because all of the dependencies are included there's also an overhead in terms of memory consumption. 16GB of RAM recommended. 8 GB would be suffient for the original Python version.

I figured out that the Audalign library on which it is based could produce multichannel output. But it is a bit useless because it would be a multichannel file composed of mono tracks of all input files. My wrapper script preserves proper stereo output by splitting the input files to mono and merging them again after the alignment has taken place.

[if_then_else]

Added the full installation procedure (on Windows 11) for the original Python script here:

https://github.com/mmathis76/TrackAlign/blob/main/README.md

If you don't want/need Python installed, feel free to use the compiled *.exe file from https://github.com/mmathis76/TrackAlign/releases/.

Again, the Windows *.exe will probably need twice the RAM compared to the Python script at run-time.

Recommendations: 8 GB of RAM for the Python script. 16 GB of RAM for the exe version.

(This is largely due to the fingerprinting and cross-correlation algorithms used by the Audalign package.
They appear to be precise but at the same time very demanding in terms of memory. 

[phil_er_up]

Added the full installation procedure (on Windows 11) for the original Python script here:

https://github.com/mmathis76/TrackAlign/blob/main/README.md

If you don't want/need Python installed, feel free to use the compiled *.exe file from https://github.com/mmathis76/TrackAlign/releases/.

Again, the Windows *.exe will probably need twice the RAM compared to the Python script at run-time.

Recommendations: 8 GB of RAM for the Python script. 16 GB of RAM for the exe version.

(This is largely due to the fingerprinting and cross-correlation algorithms used by the Audalign package.
They appear to be precise but at the same time very demanding in terms of memory.

Tried to run the .exe and received the following message:

==============================================

From webroot: Threat: trackalign.exe - infection - win.Trojan.Gen

Webroot deleted the file.
==============================================
 DL'ed it again and webroot says this time:

Malware Group:
Win32.LocalInfect.2

Location:
To my local drive c:\....
===============================================

Not sure how to proceed now...

[if_then_else]

I've scanned my original file with ClamTK on Linux and MS Defender on Windows and no Trojan was found.

The MD5 sum is

Algorithm       Hash                                                                   Path
---------       ----                                                                   ----
MD5             E9FE34B6E5EA791655E1D203D33073D7

Check if your's is identical or if the file has been tempered with.

Otherwise, the heuristics of your anti virus tool might have jumped on the exe version because of the way Nuitka had compiled the dependencies.

If that's an issue, you can still install the Python script and its dependencies based on the how-to.

I'm definitely not distributing any Trojan code.

[phil_er_up]

Was not accusing you of building  a .exe with a trojan virus just wanted to let you know what happened when I tried to run it.

Tried again to DL the file and webroot will not let me now. Can not run a md5 to see if they are the same as what you posted.

============================================================================================

Must be a hassle to do work, document it, put it out there and first person who runs it get a potential virus. Really sorry about this.

Want to thank you for all you have done with this thread and informing us about the ever changing world of audio VST plugins.

[if_then_else]

No worries. A friend of mine wanted to try it on some of his own matrix recordings. I'll have him beta-test it.

I can only tell you it took me hours to figure out the installation steps on Windows and to script it in the first place. If it helps others: the better.

I just don't understand why the Audalign package doesn't support proper stereo output in the first place. Otherwise, it looks like a great and useful tool.

[phil_er_up]

I can only tell you it took me hours to figure out the installation steps on Windows and to script it in the first place. If it helps others: the better.

Use to program myself and know how much work goes into creating software. Did not really want to tell you do to I imagine your heart just sank when you saw my post about potential virus.

I know you put a lot of work into this and want to thank you for it.

[TheJez]

Hi if_then_else,
Thanks for your work on this tool! It is much appreciated...
I tried the *.exe over the weekend on a gig from 2017 of which I had the mixing desk recording and an audience recording. Unfortunately the program reported an error at the end, when merging the mono tracks to a stereo wav file for the mixing desk recording. However, the mono-tracks were still there in the output folder, so I still had high hopes. After loading the stereo output audience recording and the two mono output mixing desk tracks into Audition, I noticed they were not very well in sync. The original input files were ~20s out of sync, the mono 'alligned files' were still ~120ms and ~50ms out of sync. Of course it was very nice that it more or less alligned them, but the remaining mismatch is too much to be usable. Any idea what the issue here is? Are the recordings too different for the algorithm to do a more accurate allignment?

I'll add a screenshot of the result and the logging... Thanks again!

.\trackalign.exe -i in -r aud.wav -t temp -o out

2025-02-08 16:10:41,809 - INFO - Logging system initialized. Full log at out\alignment.log
2025-02-08 16:13:26,608 - INFO - Split stereo file 'in\aud.WAV' into:
2025-02-08 16:13:26,608 - INFO -   - Left channel: 'temp\aud_L.wav'
2025-02-08 16:13:26,608 - INFO -   - Right channel: 'temp\aud_R.wav'
2025-02-08 16:15:51,392 - INFO - Using stereo reference channels independently
2025-02-08 16:21:42,813 - INFO - aud_L.wav: Finding Matches...
2025-02-08 16:21:42,813 - INFO - Aligning matches
2025-02-08 16:21:42,813 - INFO - aud_L.wav: Finding Matches...
2025-02-08 16:21:42,813 - INFO - Aligning matches
2025-02-08 16:21:42,814 - INFO - No matches detected
2025-02-08 16:21:42,814 - INFO - 0 out of 1 found and aligned
2025-02-08 16:26:28,959 - INFO - aud_R.wav: Finding Matches...
2025-02-08 16:26:28,968 - INFO - Aligning matches
2025-02-08 16:26:28,968 - INFO - aud_R.wav: Finding Matches...
2025-02-08 16:26:28,968 - INFO - Aligning matches
2025-02-08 16:26:28,969 - INFO - No matches detected
2025-02-08 16:26:28,969 - INFO - 0 out of 1 found and aligned
2025-02-08 16:29:00,523 - INFO - Split stereo file 'in\mx.WAV' into:
2025-02-08 16:29:00,523 - INFO -   - Left channel: 'temp\mx_L.wav'
2025-02-08 16:29:00,523 - INFO -   - Right channel: 'temp\mx_R.wav'
2025-02-08 16:31:30,948 - INFO - Using stereo reference channels independently
2025-02-08 16:37:49,428 - INFO - aud_L.wav: Finding Matches...
2025-02-08 16:37:50,860 - INFO - Aligning matches
2025-02-08 16:37:51,429 - INFO - mx_L.wav: Finding Matches...
2025-02-08 16:37:52,353 - INFO - Aligning matches
2025-02-08 16:38:05,107 - INFO - Writing out\aud_L._L_aligned.wav
2025-02-08 16:38:05,527 - INFO - Writing out\mx_L._L_aligned.wav
2025-02-08 16:38:10,992 - INFO - Writing out\total._L_aligned.wav
2025-02-08 16:38:11,459 - INFO - 2 out of 2 found and aligned
2025-02-08 16:38:11,459 - INFO - Total fingerprints: 4171050
2025-02-08 16:43:15,120 - INFO - aud_R.wav: Finding Matches...
2025-02-08 16:43:16,153 - INFO - Aligning matches
2025-02-08 16:43:16,684 - INFO - mx_R.wav: Finding Matches...
2025-02-08 16:43:18,160 - INFO - Aligning matches
2025-02-08 16:43:30,448 - INFO - Writing out\aud_R._R_aligned.wav
2025-02-08 16:43:30,873 - INFO - Writing out\mx_R._R_aligned.wav
2025-02-08 16:43:36,293 - INFO - Writing out\total._R_aligned.wav
2025-02-08 16:43:36,761 - INFO - 2 out of 2 found and aligned
2025-02-08 16:43:36,762 - INFO - Total fingerprints: 8284363
2025-02-08 16:43:37,565 - INFO - Renamed 'aud_L._L_aligned.wav' to 'aud_L_aligned.wav'.
2025-02-08 16:43:37,565 - INFO - Renamed 'aud_R._R_aligned.wav' to 'aud_R_aligned.wav'.
2025-02-08 16:43:37,566 - INFO - Renamed 'mx_L._L_aligned.wav' to 'mx_L_aligned.wav'.
2025-02-08 16:43:37,566 - INFO - Renamed 'mx_R._R_aligned.wav' to 'mx_R_aligned.wav'.
2025-02-08 16:43:41,200 - INFO - Merged 'out\aud_L_aligned.wav' and 'out\aud_R_aligned.wav' into stereo file 'out\aud_aligned_stereo.wav'.
2025-02-08 16:43:41,257 - INFO - Deleted intermediate mono files
2025-02-08 16:43:42,250 - ERROR - An error occurred during merging: attempt to assign array of size 259154417 to extended slice of size 259158513
2025-02-08 16:43:42,250 - ERROR - Failed during merging: attempt to assign array of size 259154417 to extended slice of size 259158513
2025-02-08 16:43:42,340 - WARNING - Processing completed with errors

[if_then_else]

Hi if_then_else,
Thanks for your work on this tool! It is much appreciated...
I tried the *.exe over the weekend on a gig from 2017 of which I had the mixing desk recording and an audience recording. Unfortunately the program reported an error at the end, when merging the mono tracks to a stereo wav file for the mixing desk recording. However, the mono-tracks were still there in the output folder, so I still had high hopes. After loading the stereo output audience recording and the two mono output mixing desk tracks into Audition, I noticed they were not very well in sync. The original input files were ~20s out of sync, the mono 'alligned files' were still ~120ms and ~50ms out of sync. Of course it was very nice that it more or less alligned them, but the remaining mismatch is too much to be usable. Any idea what the issue here is? Are the recordings too different for the algorithm to do a more accurate allignment?

I'll add a screenshot of the result and the logging... Thanks again!

.\trackalign.exe -i in -r aud.wav -t temp -o out

2025-02-08 16:10:41,809 - INFO - Logging system initialized. Full log at out\alignment.log
2025-02-08 16:13:26,608 - INFO - Split stereo file 'in\aud.WAV' into:
2025-02-08 16:13:26,608 - INFO -   - Left channel: 'temp\aud_L.wav'
2025-02-08 16:13:26,608 - INFO -   - Right channel: 'temp\aud_R.wav'
2025-02-08 16:15:51,392 - INFO - Using stereo reference channels independently
2025-02-08 16:21:42,813 - INFO - aud_L.wav: Finding Matches...
2025-02-08 16:21:42,813 - INFO - Aligning matches
2025-02-08 16:21:42,813 - INFO - aud_L.wav: Finding Matches...
2025-02-08 16:21:42,813 - INFO - Aligning matches
2025-02-08 16:21:42,814 - INFO - No matches detected
2025-02-08 16:21:42,814 - INFO - 0 out of 1 found and aligned
2025-02-08 16:26:28,959 - INFO - aud_R.wav: Finding Matches...
2025-02-08 16:26:28,968 - INFO - Aligning matches
2025-02-08 16:26:28,968 - INFO - aud_R.wav: Finding Matches...
2025-02-08 16:26:28,968 - INFO - Aligning matches
2025-02-08 16:26:28,969 - INFO - No matches detected
2025-02-08 16:26:28,969 - INFO - 0 out of 1 found and aligned
2025-02-08 16:29:00,523 - INFO - Split stereo file 'in\mx.WAV' into:
2025-02-08 16:29:00,523 - INFO -   - Left channel: 'temp\mx_L.wav'
2025-02-08 16:29:00,523 - INFO -   - Right channel: 'temp\mx_R.wav'
2025-02-08 16:31:30,948 - INFO - Using stereo reference channels independently
2025-02-08 16:37:49,428 - INFO - aud_L.wav: Finding Matches...
2025-02-08 16:37:50,860 - INFO - Aligning matches
2025-02-08 16:37:51,429 - INFO - mx_L.wav: Finding Matches...
2025-02-08 16:37:52,353 - INFO - Aligning matches
2025-02-08 16:38:05,107 - INFO - Writing out\aud_L._L_aligned.wav
2025-02-08 16:38:05,527 - INFO - Writing out\mx_L._L_aligned.wav
2025-02-08 16:38:10,992 - INFO - Writing out\total._L_aligned.wav
2025-02-08 16:38:11,459 - INFO - 2 out of 2 found and aligned
2025-02-08 16:38:11,459 - INFO - Total fingerprints: 4171050
2025-02-08 16:43:15,120 - INFO - aud_R.wav: Finding Matches...
2025-02-08 16:43:16,153 - INFO - Aligning matches
2025-02-08 16:43:16,684 - INFO - mx_R.wav: Finding Matches...
2025-02-08 16:43:18,160 - INFO - Aligning matches
2025-02-08 16:43:30,448 - INFO - Writing out\aud_R._R_aligned.wav
2025-02-08 16:43:30,873 - INFO - Writing out\mx_R._R_aligned.wav
2025-02-08 16:43:36,293 - INFO - Writing out\total._R_aligned.wav
2025-02-08 16:43:36,761 - INFO - 2 out of 2 found and aligned
2025-02-08 16:43:36,762 - INFO - Total fingerprints: 8284363
2025-02-08 16:43:37,565 - INFO - Renamed 'aud_L._L_aligned.wav' to 'aud_L_aligned.wav'.
2025-02-08 16:43:37,565 - INFO - Renamed 'aud_R._R_aligned.wav' to 'aud_R_aligned.wav'.
2025-02-08 16:43:37,566 - INFO - Renamed 'mx_L._L_aligned.wav' to 'mx_L_aligned.wav'.
2025-02-08 16:43:37,566 - INFO - Renamed 'mx_R._R_aligned.wav' to 'mx_R_aligned.wav'.
2025-02-08 16:43:41,200 - INFO - Merged 'out\aud_L_aligned.wav' and 'out\aud_R_aligned.wav' into stereo file 'out\aud_aligned_stereo.wav'.
2025-02-08 16:43:41,257 - INFO - Deleted intermediate mono files
2025-02-08 16:43:42,250 - ERROR - An error occurred during merging: attempt to assign array of size 259154417 to extended slice of size 259158513
2025-02-08 16:43:42,250 - ERROR - Failed during merging: attempt to assign array of size 259154417 to extended slice of size 259158513
2025-02-08 16:43:42,340 - WARNING - Processing completed with errors

It's difficult to tell without access to the original source files. I know that Audalign supports multiple algorithms. I used "fingerprinting" because it was working fine with the sources I had tried it on. I think "cross-correlation" is more accurate.

I might make the algorithms configurable in a forthcoming version, so that you could chose the one that would work best with your sources.

Would it be possible to share e.g. 5-10 minutes long clips of the original source files to see what might have gone wrong here? (Obviously, the split and aligned mono files were of different lengths, so they couldn't be joint anymore.)

If it's OK for you, let's take it to PM.

[TheJez]

If it's OK for you, let's take it to PM.

Will do, thanks...

[if_then_else]

I fixed the issue with the missing samples. Will double-check and try out several of the other supported algorithms before posting an updated version after work.

The pydub library appears to have an implicit dependency on ffmpeg or avconv installed locally. I hadn't noticed that as ffmpeg was installed by default on Linux.

[TheJez]

I fixed the issue with the missing samples. Will double-check and try out several of the other supported algorithms before posting an updated version after work.

The pydub library appears to have an implicit dependency on ffmpeg or avconv installed locally. I hadn't noticed that as ffmpeg was installed by default on Linux.

Thanks... Indeed I had to install an ffmpeg executable along with the tool to make it work. I used the most recent version of ffmpeg for that. Would that be ok? Or will the next executable contain an ffmpeg (or avconv) executable in it?
Maybe the align tool deserves a thread of its own, doesn't it?

[Gutbucket]

A question for if_then_else and TheJez specifically because they are actually building some of these tools, but open to replies from all-

Any insight into how I might create or have made a VSTplugin (ideally) or some other software tool that does something I've not found implemented elsewhere?

What I am looking to create is a four channel input > four channel output "circular panning" control or more accurately a "4 channel circular balance control" routine that uses a constant-power panning law so that perceived level does not change as the angular position of the balance pot is moved around the circle. 

[Gutbucket]

[Edit- There may already be something that does this that I haven't come across.  I'll search ambisonic tools, and maybe ask on the sursound mailing list.]

More detail (too much?) -
The circular balance control should achieve the following (I'm using numbered inputs and lettered outputs here, both of which "wrap fully around the circle" from input ch-4 back to ch-1 and from output ch-D to A.

The following cardinal orientations correspond to "hard-panned" balance positions:

1) With a panpot orientation of 0 degrees the signals are routed directly from input to output:
ch-1 > A 
ch-2 > B   
ch-3 > C 
ch-4 > D 

2) With a panpot orientation of 90 degrees, the signals are routed directly to the next "adjacent" output clockwise around the circle. ch-4 "wraps back around" to ch-A:
ch-1 > B 
ch-2 > C 
ch-3 > D 
ch-4 > A   

1) With a panpot orientation of 180 degrees, the signals are routed directly to the output on the opposite side of the circle:
ch-1 > C
ch-2 > D
ch-3 > A
ch-4 > B

4) With a panpot orientation of 270 degrees, the signals are routed directly to the next "adjacent" output counter-clockwise around the circle. ch1 "wraps" to ch4:
ch-1 > D
ch-2 > A
ch-3 > B
ch-4 > C

For panpot orientations exactly halfway between the positions above, the two immediately adjacent inputs are summed with an equal reduction in gain to both as required to achieve a constant power output, like panning a mono signal to the center position halfway between left and right by using a constant-power pan law:

5) With a panpot orientation of 45 degrees, the signals are routed with the following sums and gains:
ch-1 (-XdB) + ch-2 (-XdB) > A
ch-2 (-XdB) + ch-3 (-XdB) > B
ch-3 (-XdB) + ch-4 (-XdB) > C
ch-4 (-XdB) + ch-1 (-XdB) > D
where X = the gain reduction needed for equal power output of the sum.

etc..

For panpot orientations somewhere between those above, all three immediately adjacent inputs get summed, with unequal reductions in gain to each as required to achieve a constant power output:

6) With a panpot orientation somewhere between 0 and 45 degrees, the signals are routed with the following sums and gains:
ch-1 (smallest reduction in level) + ch-2 (intermediate reduction in level) + ch-3 (greatest reduction in level) > A
ch-2 (smallest reduction in level) + ch-3 (intermediate reduction in level) + ch-4 (greatest reduction in level) > B
ch-3 (smallest reduction in level) + ch-4 (intermediate reduction in level) + ch-1 (greatest reduction in level) > C
ch-4 (smallest reduction in level) + ch-1 (intermediate reduction in level) + ch-2 (greatest reduction in level) > D
where the reduction in level produces an equal power output in the sum.

etc..

I need to brush up on pan/balance laws to determine the appropriate constant power function and what reduction in gain that applies to each of the summed inputs.  In other words, the value of X and the value of smallest, intermediate, and greatest in the descriptions above.  With that I could draw a graph of balance position and the corresponding gain values applied to or each channel, looking sort of like a set of overlapping sine curves.  Such a graph would better describe the operation visually.

[TheJez]

Hi Gutbucket,
Unfortunately I don't have any practical experience with VST development. When googlin' a bit I see many options, e,g, downloading a SDK (software development kit) that helps to create a VST plugin from the bottom up. Something that seems very appealing is https://juce.com/. Apparently it can create the 'framework' for your algorithm that you can then export to become a VST or other types of plugins or even stand-alone applications for Win/Mac/Linux/Android! There seems to be many examples, tutorials etc.. I guess the learning curve might be steep, as is for the many other options apparently around. I guess for most options some C++ knowledge would be required. This site https://enphnt.github.io/blog/vst-plugins-rust/ could also provide some guidance...

The first step seems to be to get the math that gives the output for each channel, depending on the position of the panpot (0..360) and the inputs. I must admit math isn't my biggest asset...

I am curious why you'd want something like this. I find it hard to imagine what your four inputs might be and what turning the panpot would do to it...

[if_then_else]

For a 4-channel circular panning system with constant power, there are a few existing VST solutions that come close to your requirements:

VST MultiPanner offers 3D panning capabilities with constant power law implementation, supporting multiple channel configurations including custom routing4.

Panstation 2 provides circular panning with phase relationship control between channels, though it's primarily designed for stereo operation5.

However, none of these exactly match your specific need for a dedicated 4-channel circular balance control. The closest professional solution would be the DirectivityShaper by IEM plugins, which allows for multi-channel constant power panning while maintaining consistent perceived loudness across all positions.

[Gutbucket]

Thanks to you both.  If_then- I'll check those out when I can.  Jez- yes I'd figure the math out first and graph the transfer function. 

My use for such a thing is potentially twofold.  I'm sitting on a lot of four-channel baffled L/R/C/B recordings that I've made over the years using microphones facing in all four directions.  I have a few different methods for mixing those down to L/R stereo that I really like, and alternately for playing them back over multichannel playback systems.

One application is as a potential stereo mixdown tool for some of them where the angular orientation of the recording array was not optimally aligned with the orientation of the stage, room, and audience.  For example, in some the Left mic channel faces the "perceptual center" rather than the Center mic channel.  Such a situation would be standing off to the left with the closest PA to my Left and the majority of the audience and room off to my Right.  Fixing that is easy by just reassigning the channel routing as follows: L to C; C to R; R to B; and B to L.  I don't need the circular panner to do that.  And I do have some effective mix methods for correcting the ones which are askew by less than a full 90 degrees, but I've always been curious about alternate, faster, and potentially better ways of doing so.  Having the four cardinal directed channels available provides a much larger degree of freedom in doing this in comparison to having just two L/R channels.  I realize that simple sum-based panning may not work as well as I'd like in doing this, but I'd still love to try it.  Would be great to just "turn the knob" to reorient the 4-directional recording as desired.  A set and forget type of static adjustment.

The other application is essentially a dynamic implementation of the same for a somewhat more esoteric end, in enabling active mix-down of those four channel recordings to headphones while providing a rudimentary form of head-tracking.  Turning the knob in synchronously with my head would "turn the virtual mic array" in the same direction.  The B channel already gets filtering applied to it to perceptually spatialize it so as to sounds more diffuse and "coming from all around behind". Turning the knob would dynamically assign what information gets routed where.  Similarly I can EQ the Left, Right, Center, and Back inputs and outputs differently before and after going through the circular balance control.  Before fixes the recording the same as I would in a typical stereo mix down. Afterward would affects a form of HTRF filtering to better emulate the actual directional perception of sound arrival from different directions to the head, and even more advanced filtering could be applied if needed.  If it works, the obvious next step would be tying the circular pan-knob to a physical head tracker device.  If would enable the ability to virtually turn my head in the recorded performance space and listen in any direction during playback. 

Been thinking about this since starting to make recordings in this way back in 2006 or so.  And especially so after demoting the Smyth Realizer that does this to a far, far more advanced degree via manipulation of binaural convolution, so astoundingly convincing that at first I thought the headphones were turned off and I was hearing the actual sound sources distributed around the room as I turned around to look at them.

[kuba e]

Gut, I think this can be done via script in a Reaper. I was doing simple things in Rea script couple years ago. I know that it is also possible to create own GUI in Rea script. For example, this page explains it very nicely - https://www.admiralbumblebee.com/music/2018/09/22/Reascript-Tutorial.html . Unfortunately, I am very busy now. I will have a lot of work at least a year and a half. I need longer free time in a row to study everything again and to script it. It's an interesting task. If something changes in the future and I have more free time, I write to you.

[Gutbucket]

Thanks and same here.  Was wondering if it might be configured within Reaper. I may shift from Samplitude to Reaper at some point, or at least learn my way around Reaper, but I don't really have the time to commit to doing that at present, and will need to get up to speed before diving that deep!  Now noted for future reference.

[kuba e]

I looked at it and it's not such a complicated task. I tried ChatGPT and it generates some basic scripts. It looks pretty good. If I have one or two free weekends, I can try it. Maybe ChatGPT could generate the whole script if we ask him correctly.

[if_then_else]

The original Audalign project / Python library (https://github.com/benfmiller/audalign) which I had used for my script now supports stereo output out of the box.
I tried it on one project and it worked. (By left-padding the shorter of the two source files with the right amount of silence).

Still, depending on the algorithm it would take a massive amount of RAM to fingerprint and align the tracks.

I'm going to update my own script at some point. With the updated library, the workaround of splitting stereo input files to mono and merging them at after the alignment has become obsolete.