Topic: Couple of Samplitude questions

[Brian Skalinder]

Try Convert Audio->Save in Format. If I split an object and do this a second time it remembers that I selected 24-bit last time (and does not revert to 32-bit float).

I recall this option in SAM SE v8, but looks like in SE v9 it's been merged with the Export Audio feature.  I tried what looks like it's replacement, Export Audio | Save in Format, and SAM does, indeed, remember the bit-depth.  But Save in Format doesn't allow me to generate song-tracks at the same time, so I'll have to go through the Export Audio | Wave process again (selecting the proper bit-depth for 24-bit files), anyway.

[ghellquist]

2496 bounce with no dither -> export with no dither -> normalize -> 1644 bounce with dither -> export with no dither

I find two things that annoy me are that the export seems to always pick a odd default for the bit depth, often defaulting to 32bit so that I have to manually set and verify each time and the export always appends "_xx" when I would prefer it append "_txx" or just "xx" when I specify an output file name with "_t".

Hmm. Not sure I am following you here. What are the end results - is it one file per song in several different formats? Personally I never bounce or export anything from Sam except the final results.

I have not used SE, but tha larger Pro version. It might be that there are functions there that closer resemble what you are trying to achieve. Perhaps with a slightly different work flow, but then the end result is probably what counts.

First - export CD. This works by using the CD markers, as if you were burning a CD from inside the program. Set the markers and then name them. Now the command is File / Export Audio / wave. One nice thing is that you can force the file names to be the same as the CD markers. So if you name, say, the first song as "Morning" the file will be called Morning.wav. No suffixes added or removed. I tend to use this to export mp3-s for previews of my work on the home page.

Secondly - batch processing. Here it is indeed possible to set up "jobs" that remember the export formats. One job could be 24bit and one could be 16 bit. Batch processing works against the files in a directory so you would have to export one set first (ie with the CD export function).  This does not change any names unless you want it to.

Have to agree that the manual is really, really difficult to understand. There are a few very nice tutorial vidoes on www.samplitude.com but sadly not of these file operations.

Gunnar

[boojum]

Disclaimer: I am about 80% German and speak, read and write it, poorly.  So, here goes: do you think they are still pissed about losing the war and are getting back with that awful manual they have.  I'll bet the German one is great.  LOL   <-----   joking, folks.     

[Lil Kim Jong-Il]

Hmm. Not sure I am following you here. What are the end results - is it one file per song in several different formats? Personally I never bounce or export anything from Sam except the final results.

The end results I want are about what you would expect: a set of 2496 WAVs broken on the song boundaries and a set of normalized and dithered 1644 WAVs on the song boundaries.

I bounce before saving the tracks because that is what was recommended to me when I could not figure out how to get the output I wanted from 2496 but I also thought you had to bounce to apply offline effects like normalize and filtering before resampling to 1644.  I'll play around with export CD. 

I previously looked into the batch processing and I found what you describe, that it works on the files in the directory.  A practical issue is that I  want to batch operate on a finished track that is a complete set composed of multiple file objects, trimmed, faded and tracked.  For example I just tracked a show that was 3h of continuous playing so there were 4 component files.  They all four have to be pulled into the virtual project and tracked as a contiguous stream before being written out as songs.  I could not figure out any other way than to bounce the project and export the result. 

The other issue I had with batch was that if I did set up a job to process the 2496 tracks into 1644 for CDDA, I was concerned that there would be discontinuities at the track boundaries where the dither window drained and the refilled.  So in my work flow I resample and dither the entire contiguous set to ensure that the window starts and ends  where the set fades in from zero and back out to zero.   I could not figure out how to get the batch processing to do that. 

Any thoughts you can share on these would be welcome.

[Lil Kim Jong-Il]

Try Convert Audio->Save in Format. If I split an object and do this a second time it remembers that I selected 24-bit last time (and does not revert to 32-bit float).

When saving the file you can click on a file that appears in the dialogue and then quickly modify it's name. If you had one that had _txx from the last "convert" you can edit it to be filename_t01, filename_t02, etc. 

I'll try this too.  Thank you.

[ghellquist]

The end results I want are about what you would expect: a set of 2496 WAVs broken on the song boundaries and a set of normalized and dithered 1644 WAVs on the song boundaries.

There are obviously several different ways to do things. So I´ll simply describe how I would work this kind of things. I will keep it simplified here to keep the description short. One thing I often do is to treat applauds differently from the songs, as example turning up the volume of songs and turning down it on applauses.

1 - lets start with a long performance recorded in 24/96. This will come in several split files (my 722 can be set to split at for example 2GB).

2 - I import these into a folder in my PC. Start Samplitude. Create a new VIP, 2 channels, 96K. And load the wave files into the VIP. 

    Note: For those not using SAM, a project is called a VIP in that program.

3 - In the VIP I now search out the songs. They are generally very easy to see on the screen. At the start of the song I generally do a split the object (mark the object in the "track" and press T). It is possible to slide the split forward or backward.

4 - once I have marked them all I add CD track markers with "Set Track Marker on Object Edges". I get extra markers on the file splits, but simply select these and remove them.

   EDIT: Forgot to write that you should name the marker (in marker manager) as this name is used as the file name when you export.

5 - now, in the file menu, export audio. Each CD track in a file. Select 24/96. The versions of Sam is a little different in exactly where you set dither or not. On the other hand, with or without dither on a 24/96 export is totally unimportant as the noise added lays well below the noise from the AD conversion. I generally export to a new folder on the PC, with only files of one type.

6 - now go back to the VIP. Select all objects (Ctrl-A). Offline Effects, Normalize. Notice that normalize here only "remember" how much to increase the volume. I would do Unified Normalize to maybe 100%. But this would of course differ.

7 - I always add a limiter set to something like -1 or -2dB. The reason is that some playback devices start to sound really bad when very close to 0dB.

8 - now export again, but with settings 16/44. This also goes to a new folder.

Gunnar

[ghellquist]

This is a continuation of the previous post about working in SAM.

Under the section 7 is where you can apply various types of effects. The different versions of the program has slightly different set of effects so you should try different ones.

limiter - I suggest, when available, hard limiting with lookahead (have to check what this is called). I set the limiter at the very end of the chain and set it to -1 or -2 dB.

normalizing - notice that Sam-s normalizing works differently from just about all other programs. What it does is not writing a new file, but setting an internal volume (it is in the object). What you can do with normalizing is to normalize each song separately. Or what I tend to do, split out applauds from the songs and normalize them separately. It is possible to normalize to slightly above 0dB when you have a final limiter as that will pull down the occasional spikes going higher. You have to listen carefully though as too much limiter sounds like the crap they send on many radio stations.

EQ - this can be done either in the mixer or on each "object" separately. As it is easy to split a song into several objects, the Ctrl-T command, you can EQ different songs differently. Say EQ-ing an acoustical song differently from the electrical one. Typical things is to cut everything below 20 Hz as that does not come across any speakers anyway. You could also to a certain effect improve on a bad sounding room by EQ-ing out the worst "honkiness".

Compressor - a small amount of compression can be very effective. Most often I use single-band and a ratio of perhaps 1.2. The effect is a bit of "sweetening" of the sound and if done correcly has no hearable side effects. Of course, heavy handed compresion will totally mangle the song. Sometimes I do multiband compression, but that is a tool that I find as easily can make things sounding worse.

Stereo spread - this is above normal and to be used as something extra, sometimes I tend to use the so called "stereo enhancer" tool to reduce the stereo spread below about 100 Hz. The ear cannot hear any stereo effect here anyway, so I might as well make it mono there.

The final works you can do on a song like this is often called mastering. I tend to think of the word meaning a skill that has to be mastered through experience. It also takes a really good chain of listening equipment in order to be well done -- good speakers in an acoustically treated room. Regardless, we all use what we have and instead try a bit more conservative approach.

The purpose of traditional mastering is to make the music playable on a lot of different sound systems. Or you might want to master a special version for "car use" with very little dynamics (compressor), anything is allowed.

General advice though is to keep mastering on the conservative side, a little less is much better than too much.

Gunnar

[live2496]

Once you "convert->save in format" your files in 24/96 format and name them accordingly, you could use R8Brain to resample, reduce the word length to 16-bit (and dither). This can be done as a batch process. filename_t01 will become filename_t01_r8b, etc.

That is what I do if I am exporting individual tracks to go to somebody else. If I need a quick test CD for listening in the car I will burn it straight from the VIP. SE does not do this, but it is convenient. (These tracks can also be re-imported into another VIP project at 16/44.1, but that is a bit risky as burning the CD/extraction could introduce errors.)

You don't need to set the output as low as -1 or -2. Most mastering guys (by convention) set the output ceiling to -0.3 . The reason that it's not set to 0dbfs is that some older converters in cd players don't handle this well.

BTW, Voxengo Elephant is a very transparent compressor/limiter. They now have a version 3 just out. Use it as a VST plugin in the object editor or on the Samplitude mix buss. You can set the out gain to -0.3 and then experiment with the in gain control.

[boojum]

live2496 -  I just read your post re R8TBrain and Elefant.  I am not sure that there would be any gain in going outside SAM to work w/R8TBrain.  And, inside SAM there are three compressor options I have in SAM 10 Master:

[live2496]

Hi Boojum,
Well, I use whatever tool I think that works best for a particular track. I love Samplitude, but sometimes I use other tools that might offer a slightly different sound. Elephant does a good job of handling transients in a transparent way. I use the multiband compressor in Samp a lot too! I also use R8Brain.

Like a contractor building a house, it's good to have a wide variety of tools and know how and when to use each one.

VST plugins offer a certain diversity to an already great software like Samplitude.

[Brian Skalinder]

I tried to use R8Brain in conjunction with SAM, but the way SAM SE v8/9 deals with WAV files > 2 GB (using the "continue chunk") became an issue for me with R8Brain.  The workaround proved too much of a PITA for me to stick with it.

Before I head over to the SAM forums, do any of you know offhand if newer / different "package" versions of SAM support WAV files > 2 GB without the "continue chunk"?

[live2496]

Not that I have heard of.

[boojum]

If I knew what "continue chunk" meant I might be able to help.

The current SAM 10 Master can handle way huger than 4 GB.  Some humonguous number; bigger than three.      So it is huge.  The manual is upstairs I would tell you.  A trip to the site would do it for you.  I guess you knew that.   

[ghellquist]

I tried to use R8Brain in conjunction with SAM, but the way SAM SE v8/9 deals with WAV files > 2 GB (using the "continue chunk") became an issue for me with R8Brain.  The workaround proved too much of a PITA for me to stick with it.

Before I head over to the SAM forums, do any of you know offhand if newer / different "package" versions of SAM support WAV files > 2 GB without the "continue chunk"?

Reading that thread I cannot but believe that if you did things in a slightly different order you would not have a problem. The order I suggest is:

1 - Edit
2 - Generate tracks (each is probably smaller than 2GB)
3 - SRC and dither in R8Brain

Sam 10 handles so called RIFF64 files which can be larger than 4GB. This is a realatively new format and I am not sure if r8brain supports it. This might your solution otherwise.

Gunnar

[Lil Kim Jong-Il]

<the following was edited for brevity>

I´ll simply describe how I would work this kind of things. I will keep it simplified here to keep the description short.

1 - lets start with a long performance recorded in 24/96. This will come in several split files
2 - load the wave files into the VIP. 
3 - In the VIP I now search out the songs. do a split the object
4 - add CD track markers with "Set Track Marker on Object Edges". I get extra markers on file splits, remove them.
5 - now, in the file menu, export audio. Each CD track in a file. Select 24/96.
6 - now go back to the VIP. Select all objects (Ctrl-A). Offline Effects, Normalize.
7 - I always add a limiter set to something like -1 or -2dB.
8 - now export again, but with settings 16/44.

Gunnar,

thanks for the description.  I use a similar work flow with these exceptions:

3,4 - instead of splitting objects, I just set CD markers directly without splitting objects
5    - I bounce before this so I can probably just eliminate the bounce step
6,7 - I normalize to -0.5db
8    - I bounce before export

Questions:

1) Why do you use object split in addition to setting CD markers?

2) In step 8, I bounce first because it was suggested in a past thread that bouncing produced better results than performing the dither as part of the export process.  I don't recall why but I just went with that.  Is there no difference?

3) I have another reason why i bounce.  There are times when I need to shape levels using a volume curve to hide my my gain adjustment in the field or because of mid show PA adjustments.  Usually this is to lower the volume of a section so I want to normalize before I do a 16-bit export.  When I normalize a multi-object track that has an applied volume curve, samplitude analyzes the original non-shaped track and reports the peak level of the original recording.  By bouncing the shaped track into a new WAV I can then normalize using the peaks of the smoothed track.  Is there a better way to do this?

VST plugins offer a certain diversity to an already great software like Samplitude.

Is it better to use the r8b standalone or the VST plug-in?  Will the r8b plug-in do a batch conversion of tracks?