Taperssection.com

Gear / Technical Help => Ask The Tapers => Topic started by: dgodwin on December 08, 2017, 03:02:08 PM

Title: 16 bit still relevant?
Post by: dgodwin on December 08, 2017, 03:02:08 PM
So in the past couple weeks I put together an inexpensive rig to tape a few times a year, starting with a Guster show in January.  As neither of the last two laptops I've purchased have had a cd drive, and it seems in general that optical media has gone by the way side, is there any reason to release recordings in both 24bit and 16bit on archive.org?
Title: Re: 16 bit still relevant?
Post by: Jhurlbs81 on December 08, 2017, 03:24:36 PM
This is an opinion question so I'm sure you will get different answers.  Me personally, I record in 24 bit to get the headroom, but only release 16 bit file sets.  If I was recording more dynamic music I would release in 24 bit, but for amplified PA stuff I personally don't think 24 bit is necessary.  One thing I will add is you can release in 24 bit and still allow derivatives for folks who want to stream mp3.  If I had to guess, I would say 90% of people will stream or download the MP3, the other 10% are probably split between people who still burn to CD and those who grab the 24 bit FLACs.
Title: Re: 16 bit still relevant?
Post by: EmRR on December 08, 2017, 03:38:34 PM
50% more data at 24 bit.   Agree that 24 bit release is probably overkill for most show recordings, given the nature of background noise masking.  It's certainly preferred for the recording side of things, leaving more room to change volume and do processing in a less destructive manner.  16 bit release after post processing is probably just about indistinguishable with rock shows, apples to apples.  24 bit for classical, jazz, or Pink Floyd '69-71 with time machine assistance, sure.   

From the recording studio side of things, I do everything at 88K2/24, but clients rarely ever listen to anything other than my 320kbps mp3 reference(!), which I make from the 88K2/24 master, down-converted to 44K1/16, then to mp3.  Most can't tell THAT difference with the playback equipment they own. 
Title: Re: 16 bit still relevant?
Post by: mfrench on December 08, 2017, 03:39:50 PM
A dear friend of mine is a music loving nut, especially live music.  But, his playback system, all flac file playback, is 1644. He doesn't feel a need to upgrade it, and always grabs my 1644 offerings, and not the high-res stuff.
I shrug, and send him 1644 links.
Title: Re: 16 bit still relevant?
Post by: wforwumbo on December 08, 2017, 05:46:22 PM
Yes there is reason.

16/44.1 is still the de facto standard. It’s a way to guarantee that everyone can hear the show.

24-bit is VERY useful for higher fidelity in general - to my ear the difference between the two is very stark. And releasing at the highest possible sample rate is useful for similar fidelity arguments. Another argument I’ll make for higher sample rates is for production/post: I use digital EQs, and the higher the sample rate they are provided the lower the filter error. For compression, EQ, and analog modeling of any sort plugins work SIGNIFICANTLY better at higher sample rates - the difference is night and day.

I can post two sound clips on Sunday: one with a signal using a filter at 24/96, and another at 16/48, with the same quantity of audio and with the same filter settings, and you can listen to see if there’s a difference for yourself. I’ll allow you to make the final decision for yourself after the fact.
Title: Re: 16 bit still relevant?
Post by: vanark on December 08, 2017, 05:58:09 PM
I've started moving to only posting my 24 bit recordings. Reason? I don't way to take the computer time to dither and resample. Simple as that. If someone wants 16 bit files, there are plenty of tools out there to do that.
Title: Re: 16 bit still relevant?
Post by: thatjackelliott on December 08, 2017, 06:53:42 PM
For the recordist of live unamplified music, 24 bits does offer headroom in case the drummer smacks the snare especially hard. If taping from a sound board or from mics in front of a stack, the sound system has peak limiting built in so you can push things pretty close to 0dBFS without concern that an unusually loud event will clip.
Title: Re: 16 bit still relevant?
Post by: CorFit Chris on December 09, 2017, 01:12:07 AM
I've started moving to only posting my 24 bit recordings. Reason? I don't way to take the computer time to dither and resample. Simple as that. If someone wants 16 bit files, there are plenty of tools out there to do that.

This for me too.  I think most people simply stream the mo3 version.  I simply save to 24 bit Flaco and upload.  Want something different, learn to convert it.  I sometimes save an mo3 version and upload to my SoundCloud account because everyone is streaming it there.
Title: Re: 16 bit still relevant?
Post by: aaronji on December 09, 2017, 09:37:35 AM
For playback, as long as the dynamic range of the recording "fits" into 16 bits, which is probably true for virtually all live recordings, there won't be an improvement with 24 bit.  Extra bits of noise.  Since most people are recording in 24 bit, though, I can see the convenience factor of just releasing in that format.  As for frequency, as long as you hit the Nyquist frequency, there is nothing to be gained from higher rates.  You can always upsample for post.
Title: Re: 16 bit still relevant?
Post by: EmRR on December 09, 2017, 10:22:45 AM
Sample rate decisions should be made after listening to the quality of the converter being used.  Some sound noticeably better at specific rates, others do not.
Title: Re: 16 bit still relevant?
Post by: wforwumbo on December 09, 2017, 11:15:32 AM
For playback, as long as the dynamic range of the recording "fits" into 16 bits, which is probably true for virtually all live recordings, there won't be an improvement with 24 bit.  Extra bits of noise.  Since most people are recording in 24 bit, though, I can see the convenience factor of just releasing in that format.  As for frequency, as long as you hit the Nyquist frequency, there is nothing to be gained from higher rates.  You can always upsample for post.

This isn’t really how discretization of audio signals (or any signal, for that matter) works.

Increasing bit depth isn’t adding more bits below a certain threshold. Rather, it’s subdividing the range between 0.0 and 1.0 into more “bins” if that makes sense. So you can more accurately without error represent numbers in the time domain. Audio converters also don’t do this linearly anymore - they use an encoding process called delta-sigma encoding, whose details are outlined here: https://en.m.wikipedia.org/wiki/Delta-sigma_modulation I won’t go through all the nuts and bolts of how it works, but its practical application and effect to audio lies in being able to more accurately represent data logarithmically, which is of use given we hear pressure (and thus dynamic range) on a logarithmic scale.

A second benefit of delta-sigma lies in post production. Basically, plugins can operate at a higher degree of precision, yielding fewer errors when they perform their calculations. So filter calculations are more accurate here.

Likewise, increasing sample rate isn’t the same as recording and playing back natively at the higher resolution. Assuming a perfect up/downsample ratio of, say, 2:1 (so, going from 96k to 48k or 48k to 96k), if you downsample you’re just throwing out every other sample, which is still an accurate representation of what occurred, distortion-free. Upsampling from 48k to 96k means adding samples in between where two exist - you’re adding information that wasn’t there before. Regardless of the process used to do this, the end result of adding in a sample will induce some error - and thus distortion - to the signal. The degree of this distortion is up for debate, and it can be arguably minimized or even below the threshold of perception. But in The end you are still trying to add info that wasn’t previously there. The argument gets compounded if you’re upsampling to a non-perfect ratio (44.1 to 96, for example) as the two signals share samples less frequently.

Sample rate decisions should be made after listening to the quality of the converter being used.  Some sound noticeably better at specific rates, others do not.

This is particularly clever, and a methodology I actually agree with. The differences in the sound of gear as a function of sample rate are partially a result of what I’ve mentioned above (I.e. how accurately the original signal is actively being rendered), and partially a function of jitter - or, how accurate the master clock controlling the converter is. Effectively, increased jitter = larger deviations from when a signal is “supposed” to be captured = larger error of when, on playback, the DAC is expecting a new batch of info and not getting it within its own specs = (arguably, as far as perception) distortion. The lower the jitter, the better the signal will sound. Most clocks on audio gear samplers are optimized to run at one specific sample rate with the lowest jitter; it is preferable to use this sample rate for playback.
Title: Re: 16 bit still relevant?
Post by: u2_fly_2 on December 09, 2017, 11:17:29 AM
24-Bit if available.
Title: Re: 16 bit still relevant?
Post by: rippleish20 on December 09, 2017, 12:50:55 PM
I only post 24 bit
Title: Re: 16 bit still relevant?
Post by: achalsey on December 09, 2017, 12:51:13 PM
If anything, 24 bit isn't particularly relevant.

(a tangent and I'm feeling a little nihilistic...)

The vast majority of people listening have no idea about bit rate, or sample rate, or compression rates, or types of compression, or file types.  None of this matters.  No one has the equipment to truly accurately compare different samples.  No one is really carefully, critically listening (in a quiet room with great acoustics on an amazing system) to anything that gets posted.  They're streaming; on phones; on shitty ear buds; in the car; walking around; in public; doing something else.... No one is having a listening party.

We're mostly recording sub-par PAs cranked to 11 in terrible box-y rooms FULL of people from far further than our mics are meant to be from the source.

That's not to necessarily take away from trying to get as high quality product as possible, but all things considered in the grand scheme the extra storage data is really just not worth it as an end product.  The internet isn't quick enough; there aren't TB SD cards; most phones don't even have removable storage; most services don't have unlimited data plans for streaming.  I mostly mention phones since they've taken over the 'portable music storage' scene.  I can't imagine most people have computers hooked up to a personal home stereo system.

Do what you like, listen to whatever sounds best to you, but just remember we live in a bubble.  No one cares about 24 bit stuff.  Ask 10 random people what the difference in visual quality is between Blu-Ray and DVD.  Ask if they know if they're even two different things.  Follow up with what audio rate their TV plays.  Audiophiles are such a small niche.  If you're recording for just the audiophiles: go for it.  Put out 24/96 only.
Title: Re: 16 bit still relevant?
Post by: bvaz on December 09, 2017, 04:40:46 PM
I actually encounter people each year that still burn CDs.  I started burning CDs for the car this year again (but I can always just do that for myself and not post a 16 bit version).

I still go by the concept that media is cheap and so many devices have micro SD card slots now that I don't assume everyone is streaming mp3s, although most are.

there is no wrong answer, but it just seems easy to offer both options for me.  I want to keep the 24 bit for myself as my archive and I think more people grab a 16 bit vs a 24 bit (torrent trends seem to indicate that) so I just offer both and let people decide for themselves.
Title: Re: 16 bit still relevant?
Post by: willndmb on December 10, 2017, 04:41:10 PM
I record 24 and release 16
Tapers are the onlY ones I know who even know what 24 is. Most I encounter don't even know that 16 is cd standard. It's just much easier for the masses to enjoy at 16
Title: Re: 16 bit still relevant
Post by: aaronji on December 10, 2017, 06:06:40 PM
This isn’t really how discretization of audio signals (or any signal, for that matter) works.

I think I have a pretty decent (layman's) understanding of the process in audio (and deal with discretization in different environments in my work); I probably should have been a little more careful with my phrasing. Maybe, "In effect, extra bits of noise." I guess I also should have mentioned that the extra dynamic range (extra bit, halving of quantization error, 6 dB reduction in digital noise floor with commensurate 6 dB increase in dynamic range) contains noise that is almost certainly inaudible with most playback systems (and at non-deleterious volumes). In any event, I don't think playback at 24 bit is an improvement over 16 bit, particularly given the dynamic range of most music. Maybe I am missing something here, but I have also read many papers that see no benefit in higher bit rate playback. YMMV and whatnot...

With respect to upsampling for post work, that was something posted by ~Jon Stoppable and involved bandwidth limiting to a final target rate. I will try to find the posts when I have some time.

I record 24 and release 16
Tapers are the onlY ones I know who even know what 24 is. Most I encounter don't even know that 16 is cd standard. It's just much easier for the masses to enjoy at 16

Me too. I archive the 24 bit raw stuff, but, even for myself, I listen at 16...
Title: Re: 16 bit still relevant?
Post by: tim in jersey on December 11, 2017, 02:03:27 AM
24 bit for mastering. 16 bit for distribution.

Love being able to set levels conservatively @ festivals with multiple stages and just let the rigs run. Boost 'em in post. And long as I'm at/near 16 bit resolution my ears are happy. Love having head-room for the "unexpected-drum-thwack" or the "bass bomb". 

I'm running quality gear. Mics, pres, recorders. Noise floor from the gear is of little concern to me. 

Esepecially true when I go .007... I record music, not audiences... I can always correct/squash audience applause after the fact in post. It's a pain in the ass, but I can do it....

Not particularly well, but I can do it...


Title: Re: 16 bit still relevant?
Post by: noahbickart on December 11, 2017, 08:20:42 PM
I record at 24bit, 48kHz. But not because I can of any quality difference between 48 and 96, but because and the files are so much smaller at 48 kHz. They copy to the computer fast. They render much faster.

I record at 24 bit for added headroom, But not because of any musical content with over 16bits of dynamic range, but becauseI can run levels a little lower for safety, and know that I'm not going to clip- and that I'll normalize in post.

I release two versions, one at 24bit / 48kHz and one at 16bit / 44.1kHz.

The former is a stereo-balanced and normalized version of the raw data, for Audiophiles, purists, and people doing MTX with released SBD versions. The latter is a Redbook compatible, SBE fixed source. This gets EQ, Dynamics work, Stereo processing, and some **very** subtle "analog" saturation like one might get from transformers or tubes. It sounds better to most people without fancy stereos or headphones. That's also the version that goes on my phone for the subway or the car.

This strikes me as the best of both worlds, an "audiophile," hi-fi version, to make me feel good about myself and my gear; and a version which should sound really good, even to people who grew up in loudness wartime, and who are listening over earbuds or computer speakers.

Could I pick out the difference between the 24/48 and a straight 16/44.1 conversion in a ABX test? I doubt it.
Title: Re: 16 bit still relevant?
Post by: bombdiggity on December 11, 2017, 08:34:30 PM

I release two versions, one at 24bit / 48kHz and one at 16bit / 44.1kHz.

The former is a stereo-balanced and normalized version of the raw data, for Audiophiles, purists, and people doing MTX with released SBD versions. The latter is a Redbook compatible, SBE fixed source. This gets EQ, Dynamics work, Stereo processing, and some **very** subtle "analog" saturation like one might get from transformers or tubes. It sounds better to most people without fancy stereos or headphones. That's also the version that goes on my phone for the subway or the car.


Useful to know.  I don't necessarily think I'd realized that distinction but had picked up your classical ones when I see them (in 16bit).  Those probably benefit a lot from.some careful work to even things out. 

Given a choice of 16 or 24 as a listener I usually take the 16 unless I'm getting someone's raw files or think I might work on them myself at some point.   

When I post it is almost always 16bit and edited.  A lot of musicians just want MP3.  Many don't know what a flac is and I'm not sure any want to edit their own stuff.  They trust me to put a good polish on things. 

Recording should always be 24bit though mostly for safety and the better starting point for processing. 
Title: Re: 16 bit still relevant?
Post by: DATBoy on December 13, 2017, 01:09:10 PM
maybe my ears are shot after listening to so much loud music for so many years, but to be honest I have never heard the improvement over 24 bit audio vs 16 bit audio. So I do 16 bit 44.1 khz nearly all of the time - with exception to DVD videos, where I go for the DVD PCM specs of 16 bit 48 khz. On occasion, I ask my friends what sample rates they want - if they know what all that is. So pretty much the bare minimum whenever possible for CD audio and DVDs. Reasons why I record 16 bit is because it is smaller in file size, sounds good enough for me, and I don't do a whole bunch of "remastering" to my audio after I'm done. Less is more IMO.
Title: Re: 16 bit still relevant?
Post by: goodcooker on December 13, 2017, 05:23:26 PM

All 24 for me. My several year old Samsung phone will play 24 bit FLACs in the native player.
You can burn a CD from 24 bit FLACs it just takes a few minutes longer so I guess I don't see the point in dithering...
Title: Re: 16 bit still relevant?
Post by: nak700s on December 14, 2017, 03:33:03 PM
For me, I will always download the 24bit files.  That is also how I record, of course.  If I know I want to burn the show to a CD, I will grab the 24bit and the 16bit, or just let the burning program knock the 24bit down to 16bit on its own.  I don't do mp3, and do my best to prevent people from converting my recordings to mp3 and circulating them that way.
Title: Re: 16 bit still relevant?
Post by: Gutbucket on December 14, 2017, 05:01:55 PM
I like Noah's general approach as a way to balance quality, economy, and file management.

Although I'm not an uploader, I've long planned to do something similar in making the move from focusing almost exclusively on recording towards doing the processing I've intended and making eventual wider sharing possible of formats which make sense.  Caveat- I personally record everything in multichannel (usually some format which I can manipulate for surround-reproduction) even when the eventual release target is probably only going to be 2-channel stereo.  I want to have both available without a major file management hassle or massive storage burden.  The outline below is also informed by my own listening tests, preserving quality where necessary without excess.

Generally, I plan to produce and store 3 to 4 variants-
>The raw archival files, perhaps with non-destructive corrections applied (the things which can't be done better later as Moore's Law, my skills and available software advance - sync, trimming, major glitch repairs, normalization, inter-channel balance)
>The feast (the no holds-barred full-resolution multichannel surround mix version)
>The standard fare (two-channel down-mix of the above, preferably matrix surround encoded to provide for straight 2ch or matrix-decode surround playback, probably 16/44.1 SBE fixed Redbook compatible to remain CD burnable and to optimize file storage size)
>The snack (lossy-compressed version of the same, probably MP3)

Most of the effort go into assembling the feast.  I can derive the standard fare and snack from the full feast without too much additional effort.
Title: Re: 16 bit still relevant?
Post by: opsopcopolis on December 17, 2017, 08:27:16 AM
^ that sounds like way too much work  ;D

I generally record and distribute my AUD tapes in 16/44.1. Occasionally 24/48 is necessary, but I do it on a situational basis. On average I probably spend 20 minutes 'mastering' the recording (EQ and light limiting.)

That being said, in my studio work I never track in 16/44.1. Always 24/48 or higher, although I usually reserve 96k for acoustic music. Never worked at 192. File size is prohibitive 
Title: Re: 16 bit still relevant?
Post by: Brian Skalinder on December 17, 2017, 12:40:28 PM
I approach it similar to Noah, but simpler still:
That's it.  Still relevant, and if I still had to record at 16-bit it wouldn't have much, if any, impact on my recordings -- I'd just sweat the levels more.
Title: Re: 16 bit still relevant?
Post by: live2cd on December 29, 2017, 04:12:09 PM
Recorded in 16bit/48kHz from 2003 through 2011. (1998>2002 I recorded exclusively on MD).
I had friends that filmed shows on mini-DV and always offered them up the 16bit 48kHz files for DVD authoring purposes. I then dithered to 44.1kHz for myself, dime and the archive. In late 2011 when I realized hard drives were coming down in price, I decided to dip my toe in the 24bit pool. I now record everything in 24bit 48kHz. I end up creating 2 filesets, 16bit for mass distribution and the 24bit for me personally. I always upload the 24bit to the archive as there is the space, why not and offer 24bit in the source text via email contact on dime and other torrent sites. I also post every show I record on my personal Facebook and offer it there, as well. Its rare I get a request for the 24bit files, but Im happy to oblige if someone loves the tape.

I never listen in 16bit anymore. I upgraded my home computer audio with the Audioengine A2's and the Audioengine D1 DAC and the sound is out of this world. I realize most people stream, but I record these shows for myself first and the public 2nd. I want to make sure it sounds the best on my system and whatever anyone wants to do with the files after I upload is up to them.
Title: Re: 16 bit still relevant?
Post by: furburger on December 29, 2017, 06:35:15 PM
24 bit is the "Betamax" of taping.

wait for 32.
Title: Re: 16 bit still relevant?
Post by: Sloan Simpson on December 30, 2017, 09:12:14 AM
24 bit is the "Betamax" of taping.

wait for 32.

 :facepalm:
Title: Re: 16 bit still relevant?
Post by: DSatz on January 01, 2018, 09:59:05 PM
I don't know whether you're being serious or not; no converter (neither A/D nor D/A) has ever been available that is capable of full 24-bit performance.
Title: Re: 16 bit still relevant?
Post by: furburger on January 07, 2018, 01:19:03 PM
I don't know whether you're being serious or not; no converter (neither A/D nor D/A) has ever been available that is capable of full 24-bit performance.

exactly my point.


should they ever make a 32-bit, I may look into it.


beyond that, it's an obsolete, unnecessary format.

video games always doubled the bits when they touted an improvement. (8 bit, 16 bit, 32 bit, 64 bit, etc...)


50% of the 'power' (used loosely) of the previous format.....not terribly appealing.

the amount of snatches for each 24 vs. 16 shows I'm far from the only one who feels this way.

24 bit is just a modified DATtitude, nothing more.
Title: Re: 16 bit still relevant?
Post by: yug du nord on January 07, 2018, 02:12:26 PM
^I'm not aware of any dedicated recorder that could handle 32 bit...  so I imagine that you'd have to use a computer as a recorder in order to record "32 bit"...  but................

https://mytekdigital.com/brooklyn-adc/ (https://mytekdigital.com/brooklyn-adc/)

The Stereo192 is fantastic...  and this Brooklyn ADC looks like something that digital dreams are made from!!
No more "waiting" furburger.....  go get it!   8)


Title: Re: 16 bit still relevant?
Post by: wforwumbo on January 07, 2018, 04:16:05 PM
It seems some of you guys are confusing recording bit depth with computational load and processing power.

Modern converters very much so can actually capture a full 24 bits of dynamic range as signed integers. The current cutting edge of converters can handle a full 32-bit floating point conversion. This is slightly distinct from 32-bit and 64-bit chipset architectures associated with computer processing power.

I recommend you generate some plots of 24-bit signed integer recordings to see how much info/data is actually stored in a word length of audio data as an exercise in factual scientific data.
Title: Re: 16 bit still relevant?
Post by: yug du nord on January 08, 2018, 12:20:43 AM
^I am probably confused.....  but I do think that this Mytek Brooklyn ADC claims to output 32 bit.  Not just a floating 32 bit for processing.
I'm also not sure about how many ADC's that can be used in the field actually output a true 24 bit.  I have used ADC's that have been in the 23 bit range I think..  but I honestly don't think I've ever "managed" to get a full 24 bit.  So either my recorder hasn't been capable of capturing a full 24 bit...  or the stand alone ADC's that I've used have not output a full 24 bit.  I've never generated any "plots" though.
Maybe some studio, rack mounted ADC's can output a full 24 bit...  but most here on TS need a field ADC/recorder.
But I very well could be wrong on all of this??...

Can you mention any portable, field ADC's that you've generated plots for that output a full 24 bit?
I am not a scientific recording type of person..  for better or for worse, I tend to trust my ears. 
But I do like to consider "facts" if they are indeed facts. 
Thanks!
Title: Re: 16 bit still relevant?
Post by: WiFiJeff on January 08, 2018, 09:34:44 AM
The Mytek manual claims 128 dB of dynamic range, which is better than the 120 dB of many mid-price modern recorders but still only a tad more than 21 bits as I count.
Title: Re: 16 bit still relevant?
Post by: morst on January 08, 2018, 11:41:25 AM
I use 24/48 for everything.   :headphones:

Yeah, I know it's bigger files but hard drives are really cheap, and besides, I FLAC everything except the master WAVs anyhow, so that helps with download bandwidth.

I feel like 48kHz is the standard for video for many years now, and anyone with footage of something I record will have an easier time syncing to my 48k output.

I used to ask folks to post 16-bit because I didn't want to download the 24, now I don't even post 16! Times change!

The archive has lossy derivations so I don't feel like I'm excluding dialup users.

I don't know whether you're being serious or not; no converter (neither A/D nor D/A) has ever been available that is capable of full 24-bit performance.

And no rock and roller has either!
Title: Re: 16 bit still relevant?
Post by: wforwumbo on January 08, 2018, 12:21:07 PM
The Mytek manual claims 128 dB of dynamic range, which is better than the 120 dB of many mid-price modern recorders but still only a tad more than 21 bits as I count.

Ah, so THIS explains the confusion over "full use of 24/32-bit dynamic range" conundrums some here seem to be facing.

The 6 dB-per-bit "rule" is NOT actually a rule. Rather it's a general trend that holds true for bit depths well below 24 bits (really ~6 to ~18 bits is where this rule holds true), based on a theoretical limit. Really, it's a carryover simplification from back in the day, when going past 16 bits seemed like an impossibility at the time.

The absolute BEST 24-bit converters are capable of producing ~132 dB of dynamic range, and the top-of-the-line-super-expensive 32-bit float converters can stretch that to ~140 dB. Obviously this is mostly in highly-controlled rack mount gear, and it all costs an arm and a leg plus sometimes your firstborn.

Edit: I also want to mention, it is worthwhile to discuss to what level of detail we can distinguish between 16 or 24 (and beyond) bits of dynamic range. At the end of the day, I could understand an argument that we wouldn't take full advantage of 24 bits, and that 16 bits is for the most part "good enough" to pull a solid tape especially when taping live rock music. However, my counterargument to that argument lies with people who do post work; having a larger bit depth will assist in lower errors particularly when using equalizers. As a studio guy, I fully support higher bit depths, with this reason being a strong one.
Title: Re: 16 bit still relevant?
Post by: Gutbucket on January 08, 2018, 01:19:07 PM
However, my counterargument to that argument lies with people who do post work; having a larger bit depth will assist in lower errors particularly when using equalizers. As a studio guy, I fully support higher bit depths, with this reason being a strong one.

Along those lines, what about use of editing software which works internally at higher bit depths?  This has to be the common methodology of DAWs and editing software these days.  I use Samplitude, which regardless of the native bit-depth of the source files, does all its mathematical manipulations at the 32-bit float level, then dithers the the internal 32-bit representations back down to whatever is specified upon export.  Call it digital head-room avoiding rounding errors or whatever, equalization math is occuring in a 32 bit space "within the machine" regardless of the bit depth of the source files, which simply get padded with zeros as necessary below their original LSB (least significant bit) to bring the word-length of all sources up to 32bits prior to any manipulation.

That's the way I understand it anyway, which would seem to counter the "important for post-production" argument. 
Title: Re: 16 bit still relevant?
Post by: wforwumbo on January 09, 2018, 01:41:09 PM
However, my counterargument to that argument lies with people who do post work; having a larger bit depth will assist in lower errors particularly when using equalizers. As a studio guy, I fully support higher bit depths, with this reason being a strong one.

Along those lines, what about use of editing software which works internally at higher bit depths?  This has to be the common methodology of DAWs and editing software these days.  I use Samplitude, which regardless of the native bit-depth of the source files, does all its mathematical manipulations at the 32-bit float level, then dithers the the internal 32-bit representations back down to whatever is specified upon export.  Call it digital head-room avoiding rounding errors or whatever, equalization math is occuring in a 32 bit space "within the machine" regardless of the bit depth of the source files, which simply get padded with zeros as necessary below their original LSB (least significant bit) to bring the word-length of all sources up to 32bits prior to any manipulation.

That's the way I understand it anyway, which would seem to counter the "important for post-production" argument.

This still provides the argument for 24-bit; you're effectively capturing more bits for the software to work with, and therefore the output computation is more accurate.

For a simple example, let's say I had a system that wanted to add two numbers. The numbers are 1571 and 0448. The result for any human adding these two numbers together is 2019. The effect of a 16-bit system would be having the two numbers as 1500 and 0400, which adds up to 1900. SOME 16-bit systems can try and adjust this during recording as 1600 and 0400, which gives us 2000 and is a bit closer, but that requires some additional processing power on the A/D converter, most of which doesn't occur except in more expensive 16-bit recorders. If your converter is capable of 24-bit or 32-bit float and you're recording in 16-bit mode or manually converting to 16-bit in post, you're probably getting the former. A 24-bit system is the same as going to 1570 and 440, which adds up to 2010 - a lot closer than 1900.Your Samplitdue system, at 32-bit float and importing in 24-bit files, would be similar to having 1570.00 and 0440.00, which allows for decimal rounding at the end after processing - not of note during recording, but DEFINITELY useful in post, particularly pre-bit reduction and dither, as the equalizers can come MUCH MUCH closer to what the real values should be.

This is obviously an imperfect example, and a touch exaggerated. But it gets the point across.
Title: Re: 16 bit still relevant?
Post by: Gutbucket on January 09, 2018, 04:31:08 PM
I see the argument above as essentially being- "the entire system is more accurate", which is different than what I'm getting at.  I'm talking about "preserving the degree of accuracy we start with".

Let's assume someone is fine with direct 16-bit recording without any additional processing.  The relevant question then becomes "what is required to preserve the accuracy of that 16 bit source during processing"

Given a 16 bit input*, I grant that processing in a 16-bit workspace will not provide sufficient calculation headroom due to rounding errors as you describe above.  Yet processing that 16 bit input in a "mathematical workspace" large enough to avoid rounding errors during the calculations does preserve accuracy within that workspace.  In the course of the calculations, the word-length will increase somewhat (more decimal places are needed to prevent rounding errors).  We can then output a file that has sufficient bit-depth to retain that increase, say writing a 24bit output file, or we can reduce the word-length back to 16 bits again.  In doing that we lose the additional decimal places generated during the processing, but it avoids introducing error during proccessing (except for that final word-reduction step, which is arguably not part of the processing we are concerned with).

In that way, more bits are needed during processing, but if one is fine with the 16 bit original to start with, there should be no concern about processing that.

I remain to be convinced this is not the case.
Title: Re: 16 bit still relevant?
Post by: Gutbucket on January 09, 2018, 04:40:20 PM
^ That's my "mathematical" argument. 


*In a real world situation, the dynamic range of what tapers are recording will most-likely fit easily within 16-bits.  The noise-floor of a taper's recording is typically pretty high, with lots of room at the bottom where the least-significant-bits will be representing randomized noise, and at least a few of the most significant bits will hopefully be vacant unless the recording clipped the ADC.  So there is room to spare within the digital representation envelope represented by 16-bits.  Let's say the meaningful recorded signal uses more or less 11 bits, and we include a few bits of noise included at the bottom for 14 bits total so as not to loose detail beneath the noise-floor.   A cleverly designed DAW constrained to work within a 16 bit workspace could shift that 14bit content up or down within that workspace to provide calculation space needed prior to doing the calculations which produce a 15 or 16 bit result, normalizing the signal prior to doing the calculation.  I've no idea if that's how things used to work back in the 16-bit processing days, but that kind of work-around is rendered moot by increased processing power using sufficient floating-point internal calculation space that can easily contain the accuracy of the source including any calculations done on it. 

The bottle-necks are the input and output bit depths to the editing system, not the calculation space any longer.