The truth about bit-depth (and digital audio ‘resolution’)

How to do this, that and the other. Share, learn, teach. How did X do that? How can I sound like Y?
Post Reply New Topic
RELATED
PRODUCTS

Post

jochicago wrote: Mon Feb 11, 2019 12:17 am44.1k/16 bits is not just arguably inferior to better formats, it is documented as audibly inferior to better formats:
http://www.aes.org/e-lib/browse.cfm?elib=18296
That meta-study is so full of disclaimers that it is practically worthless. Most of the studies they used were not music-related but involved test frequencies only. Several were bone-conduction tests and the rest are the subject of big disclaimers, like "this does not necessarily imply that high resolution audio is consciously, or even subconsciously, distinguished".
As an artist/producer I don't want it to be 44.1k/16bits. If there's a chance someone could tell the difference (or take less enjoyment out of it even subconsciously).
So you would never release anything on CD or put it on Soundcloud or Bandcamp or Spotify, even though you know as well as I do that 99.99% of people will be listening to it as streaming, compressed mp3 on shitty earbuds? When I master something for CD, it is exactly the way I want it to sound. EXACTLY. Not "good enough for 44.1/16" but EXACTLY the way I want it to sound. Just because someone with perfect hearing might, under controlled laboratory conditions, be able to hear a difference between that and the 24 bit master before I downsample it is of less than no consequence to me. OTOH, I would never be happy if anything of ours was released on vinyl because it is nothing more than catering to people who don't know any better and I wouldn't want to take advantage of them like that.
And even if subtle and borderline unnoticeable I know it's there, while just as easily I could use a format of the same file that meets a higher standard of demonstrable quality. I want to produce at a resolution high enough at which it truly won't matter, and IMO, from what I've seen/read, 44.1k/16Bits is not high enough to meet that standard.
But you said it yourself - "subtle and borderline unnoticeable". That means if you are listening to it in the car it is completely unnoticeable. If you listen to it on your multi-room Sonos set-up at home, completely unnoticeable. Listening to it on your favourite headphones on the bus or train, completely unnoticeable. And the real clincher is that only you know about it, nobody else has any point of comparison.

You're carrying on as though someone will sit, perfectly positioned, in an anechoic chamber with a $100,000 hi-fi, hear your CD and go "OMG, that sounds bloody awful!". Not going to happen. In fact, they aren't even going to say "it could be a bit crisper in the top end" because they won't have anything to reference it to.
Now, ask me if I personally can hear a difference between high quality mp3 and cd format. :hihi:
(I guess I have to go test this now)
You will. It is very obvious to me, even in my car, but only on my own stuff, not for anything else. That's because when I put on some music to listen to, I am not listening to it critically, but whenever I listen to my own stuff, I am being super-critical, so I notice stuff like that. And that's the thing - I don't want to make music for critics, I want to make music for normal people who enjoy what we do (although people who do enjoy our music tend not to be normal, almost be definition).
NOVAkILL : Asus RoG Flow Z13, Core i9, 16GB RAM, Win11 | EVO 16 | Studio One | bx_oberhausen, GR-8, JP6K, Union, Hexeract, Olga, TRK-01, SEM, BA-1, Thorn, Prestige, Spire, Legend-HZ, ANA-2, VG Iron 2 | Uno Pro, Rocket.

Post

This is what I'm getting from you:

Personal Opinion: I can't hear the difference, so:

Fact: The difference is entirely unnoticeable

That second one is wrong.

You don't like the research. Fine. I'm not going to push this thread deeper by trying to tell you anecdotes, things I've heard, other's opinions, because you didn't like actual studies. Take what you will. Others will as well. I'm not muddying the waters, I'm presenting the facts as I understand them and what has guided my choices.

For me, the bottom line is simple. Music does work at 44.1/16bits. But we know for a fact that's not the limit of the human ear. If you want to have a format that covers human-hearing to its full capacity, you need something deeper. And that's just as an export format. We haven't really explored what happens to 44.1k/16 when you use it as source files and apply 15 effects to it.

Personally, whenever I have the option, I'll try to work with better than 44.1/16bits. That means tracking, mixing and exporting higher definition. I also publish stuff to share online, I often use mp3 and that goes 16bits. But when I can, given the option, there's no need to drop the quality of the format just 'cause it is hard to hear a difference.

And that's all I have to say about that. :wink:

Post

. We haven't really explored what happens to 44.1k/16 when you use it as source files and apply 15 effects to it.
Sorry, but yes we have. As long as the intermediate format between these 15 effects is not also 16bits (and the VST standard dictates it is not) it will be absolutely fine.
We are the KVR collective. Resistance is futile. You will be assimilated. Image
My MusicCalc is served over https!!

Post

BONES wrote: Sun Feb 10, 2019 11:38 pm No, you don't. It's a mechanical process with so many points where degradation can occur that it is simply not possible, even on the first listen. By the tenth listen, it will have degraded noticeably, even with a good turntable, tone-arm, cartridge and stylus. You have a sharp piece of diamond running through soft plastic, it's a recipe for disaster. OTOH, a CD will contain exactly the same digital information as the mastered source material, making it absolutely identical in every respect. With a CD, it is like copying a file from one folder on your hard-drive to another - it is a perfect copy, exactly as the artist/producer wanted it to be, and will remain that way forever.
It's also shown grooves return if you let it rest a bit. I saw some test they did 10 plays in a row, and looked at deformation - it was severe. They compared various shapes of stylys, elliptical, round, and semi something.

But vinyl is a clever material in being alive. Let it rest for hours and it pretty much return to original.

But in the sense of perfect copies forever, this is advantage of digital. But 44/16 seems to be ok for consumer level. Even thinking that mp3 is enough for many - so it depends how you use music, as musak in background or dedicated moments just listening and being swept away. I'm more like the latter, it's dedicated time for me.

Stylys are worn, and records themselves - but this is life. Everything alive is worn and part of the charm. Some people like antique furniture and stuff because of this. Some are even into 78 rpm records capturing incredible performances of artist early of the 20th century.

All this digital stuff, never destroyed - it will soon need another planet to store it all. The giant companies look in the northern parts of sweden for server storage having mostly woods and no citizens, kind of. But turns out the electrical grid is not up for the job everywhere to support these monster server locations.

Things being alive - also involves being built in destruction. Digital does not do that. Books, films everything is turned digital to be able to store it - forever.

I saw a comment by Neil Young in the film about Soundcity studio to the meaning:
- they say digital is so good, you can copy forever and forever without loss
- but you already lost everything before doing that, so how good is that

So this was why he was so involved in the Pono project and think they went all the way to 192 KHz or something as max. Later reading his autobiography he told the fuller story looking for funding of the project.

So the compromise 44/16 for musak - extending that is probably the way to go for more serious listening. Just as vinyl made a comeback, maybe SACD will too. I saw players now support DSD if that is format for future. some daws support it for import too.

Everything being better does not win the commercial success, but still. The Pono player was available, but competition is providing music in a fashion consumer friendly is hard. Smartphones came up while Pono was being introduced.

Same goes for SACD. There was one place in hongkong that had a couple of thousand titles for SACD, but very few I was interested in. And since standard was no players allowed the full quality on spdif to my own dac, I did not go for it. Maybe there are cracked firmware to fix this on some units, don't know.

I'm going to do some more tests with 96k/24 all the way since I can play back that in my own dac from a pc. If to go 96k all the way.

I did since some years ago even, doing upsampling to 96k of all VST instruments inside 48k project - and what you gain is clearly audible improvement. But now using mostly external gear I want to record 96k.

If you can notice differences, it adds up for a full project. We have these incredible tools available today, why not use them to the fullest. I gave up in the 80's not affording to go at least to 16track Fostex reel-to-reel at the time. Now being retired and doing this full time I can go for what I feel is doing the job.

If to put all the work into the music, why not go for as good technical quality as possible as well. If everything can sound smooth as silk, why not.

Post

lfm wrote: Mon Feb 11, 2019 7:31 amBut 44/16 seems to be ok for consumer level. Even thinking that mp3 is enough for many - so it depends how you use music, as musak in background or dedicated moments just listening and being swept away. I'm more like the latter, it's dedicated time for me.
For me too, but it's not about the sound quality, it's about the song. A good song will be enjoyable if someone farts it out of their arse. If all you have to hang your hat on is pristine audio quality, then you need to reassess your priorities.
Stylys are worn, and records themselves - but this is life. Everything alive is worn and part of the charm.
Really? Would you rather sleep with a 20 year old model or her grandmother?
Some people like antique furniture and stuff because of this. Some are even into 78 rpm records capturing incredible performances of artist early of the 20th century.
Lamest excuse ever.
All this digital stuff, never destroyed - it will soon need another planet to store it all.
Right, whereas vinyl is the most efficient storage medium of all time, which is why a collection that's a tenth the size of my complete collection takes up the whole back wall of my garage.
Things being alive - also involves being built in destruction. Digital does not do that. Books, films everything is turned digital to be able to store it - forever.
Well, somethings are worth storing forever, other things are infinitely disposable. I suppose it comes down to how much you care about a particular thing.
I saw a comment by Neil Young in the film about Soundcity studio to the meaning:
- they say digital is so good, you can copy forever and forever without loss
- but you already lost everything before doing that, so how good is that
Yeah, but it's Neil Young, who sounds like shit at the best of times.
Everything being better does not win the commercial success, but still.
That's because it only has to be good enough, not as good as humanly possible. The law of diminishing returns kicks in way before you need to go beyond 44.1/16. That was the whole point of choosing that standard.
I did since some years ago even, doing upsampling to 96k of all VST instruments inside 48k project - and what you gain is clearly audible improvement. But now using mostly external gear I want to record 96k.
This proves that you have no idea what you are talking about. You cannot add quality that's not there in the first place. Simply not possible. If you up-res from 16 bit to 24 bit, all you do is copy the same value to 256 samples in the 24 bit file.
If to put all the work into the music, why not go for as good technical quality as possible as well.
Because only idiots care and I'm not in the business of catering to idiots. If 99.99% of people are happy with CD quality, why would I expend even the tiniest effort into providing anything more?
NOVAkILL : Asus RoG Flow Z13, Core i9, 16GB RAM, Win11 | EVO 16 | Studio One | bx_oberhausen, GR-8, JP6K, Union, Hexeract, Olga, TRK-01, SEM, BA-1, Thorn, Prestige, Spire, Legend-HZ, ANA-2, VG Iron 2 | Uno Pro, Rocket.

Post

shit, this again.

let me put it this way:
how is
1111 1111 0101 1010 0010 1010 in any way different than:
____ ____ 0101 1010 0010 1010
or
0101 1010 0010 1010 0000 0000

?
I can add those 00000000 or 11111111 for you, i can even and 16 of them and give you 32bit ""resolution"".
(btw this is practically how 32bit FP works)

talking about end result of course...

edit: unless you absolutely NEED that dither resolution which is 30dB below any sensible analogue path. :D
Image

Post

can i get the model, her mum and grandmother?

3g session is on my bucket list.

Post

BONES wrote: Mon Feb 11, 2019 12:24 pm This proves that you have no idea what you are talking about. You cannot add quality that's not there in the first place. Simply not possible. If you up-res from 16 bit to 24 bit, all you do is copy the same value to 256 samples in the 24 bit file.
No, sir, that is not what is happening - that would be useless.
But could be a matter of confusion on terminology.
Oversampling is another term for this I am talking about.
This stuff below is implied in what I was talking about.

This is what happends
a) resampling notes to various pitches create aliasing artifacts - and less so the higher sample rate used.

b) math will do interpolation between each sample, or it would be pointless as you mentioned just duplicating same sample and then back down again. When we talk about synths it will generate these extra samples according to it's own algos - host tells to run in 96k, and out it comes.

So done properly it does improve perceived quality.
Especially on more complex waveforms like strings it become clearly audible.
And also depending on quality of sampler library you use - if every notes is sampled and recorded, with layers and all - or if same sample is resampled for many pitches.

You mention "good enough" and that is correct of course. But when you are fed good enough - long enough - and stumble into "better than enough" you get hooked.

This is what we see on analog synth department nowadays. The digital stuff is good enough - but something about the analog that bring something extra. We want that sound but not the hassle with retuning every couple of minutes.

Same with analog hardware emulating plugins - it just brings something smooth to it all. Enforcing some even harmonics on existing audio will create some "warmth". Companies like Softube was built entirely on finding math to do that well.

If there is a demand - there is a market - and new products can arise that satisfy this. Oldest sales trick in the book
- this is better

I used one of your synthedit based synths way back now(I think at least 10 years) - and you supplied me with that panel since I thought it was just too dark to be used that way. It was good enough back then - but hardly today. So many synths sounding the same, using the same modules just about. I looked around a bit and found other oscillators but stopped at some point. And it did not work well in x64 environment either anything synthedit.

When we stop improving we are dead on so many levels. It's through experimenting we learn new things. If there is potential to improve audio quality without empty all bank accounts - why not do it.

As a child we might think "twinkle twinkle little star" is the best song there is - then we move on finding new music - forever.

Post

i won't even bother commenting in this thread. i'm tired of clueless people repeating the same debunked bullshit.

here's a test you can do. take your favorite 384 KHz/128bit recording, and downsample/truncate it to 44.1/16bit, then upsample it back to the source format, then do a blind test. i guarantee you - under controlled conditions, you won't hear any differences.

you hear a difference between vinyl and CD masters? congratulations: you can hear a difference between two different masters, have a medal. they're not supposed to sound the same!

regarding dynamic range of 16-bit audio, -96dB of dynamic range doesn't mean anything below -96dB turns into noise. this is -96dB average signal strength. you can get down to ~-110dB with dithering, which most audio interfaces will do.

know your math, people.
Last edited by Burillo on Mon Feb 11, 2019 1:29 pm, edited 1 time in total.
I don't know what to write here that won't be censored, as I can only speak in profanity.

Post

lfm wrote: Mon Feb 11, 2019 1:19 pm
When we stop improving we are dead on so many levels. It's through experimenting we learn new things. If there is potential to improve audio quality without empty all bank accounts - why not do it.

As a child we might think "twinkle twinkle little star" is the best song there is - then we move on finding new music - forever.
the point is increasing bit depth is not necessarily improvement. Right now, bit depth is the least of problems with the music industry...

also increasing bitdepth is not oversampling. sheesh, stop just shelling out opinion and read what people wrote.

edit: nobody is propagating to record at 16bit to be clear, just that if end result is squished to -0.0dBFS you don't gain anything is last 12 bits are 1111 1111 1111 at all times. There's no added resolution, because that's not how binary works.
Image

Post

Ploki wrote: Mon Feb 11, 2019 1:28 pm
the point is increasing bit depth is not necessarily improvement. Right now, bit depth is the least of problems with the music industry...

also increasing bitdepth is not oversampling. sheesh, stop just shelling out opinion and read what people wrote.

edit: nobody is propagating to record at 16bit to be clear, just that if end result is squished to -0.0dBFS you don't gain anything is last 12 bits are 1111 1111 1111 at all times. There's no added resolution, because that's not how binary works.
Increased bitdepth as well as oversampling is about increasing resolution - and that is subject "The truth about bit-depth (and digital audio ‘resolution’)".

And I objected to what OP said in video - all you loose is noise floor.

You have your opinion and others have theirs. What you say is basically we should not care what food taste - since it becoming poop in the end anyway.

In the near future there will be an aftermath of mp3 and 16-bit audio - that is my forecast. Since first step was more and more turning to vinyl again - it is an natural development digital will supply something for those wanting improvements.

Especially since there is now full generation that might not heard anything better than mp3 - thinking this is what music sound like.

Industry growing really strong is providing new experiences like for tourists or otherwise.

Home theater is developing 4k, bigger screens and hdr and all kinds of improvements to bring something new - and better.

I think audio is as large part of experience of film and video part. New types of surround is part of that. And my forecast audio resolutions also will be.

So even bigger reason to stay in as good quality resolution as possible recording audio as well. The know-how of this will be in even greater demand in future- again my forecast.

Post

lfm wrote: Mon Feb 11, 2019 3:45 pm
You have your opinion and others have theirs. What you say is basically we should not care what food taste - since it becoming poop in the end anyway.
No see, this is where are you are wrong. In digital audio, opinions don't count, only knowledge counts.
In the near future there will be an aftermath of mp3 and 16-bit audio - that is my forecast. Since first step was more and more turning to vinyl again - it is an natural development digital will supply something for those wanting improvements.

Especially since there is now full generation that might not heard anything better than mp3 - thinking this is what music sound like.

Industry growing really strong is providing new experiences like for tourists or otherwise.

Home theater is developing 4k, bigger screens and hdr and all kinds of improvements to bring something new - and better.

I think audio is as large part of experience of film and video part. New types of surround is part of that. And my forecast audio resolutions also will be.

So even bigger reason to stay in as good quality resolution as possible recording audio as well. The know-how of this will be in even greater demand in future- again my forecast.
And i'm not saying we shouldn't care what food taste, i'm implying that food (music) is more concerning right now that it's carrier - the f**king plate.
You're more concerned with the plate than the food itself tho.

Turning to vinyl has much more to do with nostalgia, hipsters and music approach than actual quality - because compared to digital, its quality is garbage, on average turntable with an average pressing even more so.

Gen X and millennial generation grew up on f**king cassettes, and yet it was them (us) that caused the resurgence in vinyl and made CD popular... so today's kids growing up on MP3s doesn't mean anything at all. We thought music sounds like when we made cheap bootlegs on cassettes...

New types of surround have nothing to do with bit depth and sampling rate, it's all about 48/24 anyway, new types of surround have to do with acoustics and post production approaches. And i welcome that.

"As good as possible" makes no sense if there are absolutely no benefits to it.

4K is great yes sure, i'm typing this on a 4K 22" screen because I enjoy seeing no pixels on my screen. But increasing resolution becomes irrelevant when said resolution exceeds human eyesight. Screens still aren't nowhere near that point, not in color nor resolution.

Audio is pretty much there, because it's a computationally and data-wise far less complicated medium than video is, so your analogy is kind of false.
When you cannot discern pixels anymore, there's enough resolution. This happens at different resolution depending on screen size and your viewing distance.
It makes sense to store video at greatest possible resolution an eye can discern, or better, greatest resolution that a LENSE CAN EVEN PRODUCE, however that is problematic because it's not as clear cut as audio. For colors, it's already clear what we need.

Increased bit depth when recording video is used not because it can be displayed, but because you can over/under expose and because you are safer from clipping highlight.

For audio, it's pretty clear too. We can hear from roughly 20hz-20khz and we have are capable of consuming of roughly 120dB of dynamic range.

As i already said, having 64 bits of dynamic range, where bottom 48 bits are at full 1 all the time is the same as having 16bits of dynamic range. There's no increase in resolution. Can you please explain to me, from the case i presented a little bit higher, how can something have more resolution if 8 bits are full 0 or full 1 at all times?

Please tell me a practical application of 32bit integer audio which will be quantifiably better than 24bit, considering analog circuits are already limited at 123dB due to physics.
Image

Post

let me put it this way:
Whatever resides of audio signal from -12 to 0dB FS will occupy 2 bits, regardless of whether you use 8bits, 16, 24 or 64bit depth.

edit:
also please refer to this:
https://en.wikipedia.org/wiki/Quantizat ... rocessing)
The higher resolution corresponds directly to lower quantization error which corresponds directly to lower noise floor and that's all there is to it.

The funny joke is that even full 24bit system is not really playable on any analog system anyway... so dreaming about 32 and 64 bits as target mediums is just not know anything about digital audio.

i;m not arguing against using higher bit rates during mixing/manipulation stages, obviously. (but there, 32bit FP is fine)
Image

Post

Ploki wrote: Mon Feb 11, 2019 4:17 pm let me put it this way:
you can put it in any way you can think of, but it's futile. you're not going to change a person's opinion with facts if his opinion wasn't based on facts to begin with.
I don't know what to write here that won't be censored, as I can only speak in profanity.

Post

Ploki wrote: Mon Feb 11, 2019 4:09 pm
As i already said, having 64 bits of dynamic range, where bottom 48 bits are at full 1 all the time is the same as having 16bits of dynamic range. There's no increase in resolution. Can you please explain to me, from the case i presented a little bit higher, how can something have more resolution if 8 bits are full 0 or full 1 at all times?

Please tell me a practical application of 32bit integer audio which will be quantifiably better than 24bit, considering analog circuits are already limited at 123dB due to physics.
You are looking too much at the peak meters and like if music were dead current.

You are thinking like it's about volume and loudness war or something.

You peak at -12 dBFs and you suddenly don't care about two bits?????
- I can just go down to 14 bits and hear no difference?

Signal is constantly jumping up and down between +/- what ever bitdepth you are at - or there would be no music. Dead current is 0 Hz frequency to put it that way, music is alternating current.

I'm thinking of representing every little nuance of the signal all the way from -8 mil bit steps to +8 mil bit steps that constantly is happening.

The true sound of an instrument, is fundamental frequency and then all the odd and even harmonics on that sound defining that a violin sounds like a violin and an acoustic guitar like acoustic guitar, or a piano like a piano or anything of that sort.

And also representing these harmonics, each with lesser amplitude than fundamental frequency of the note played - is what makes it sound true to our ears.

And also representing every harmonic in true phase as original.

So the complex musical signal for a full mix - and yet another level that is to be truly represented in coming out from speakers when we listen.

So that many feel CD quality sound sterile and pale - and less true than vinyl in some case - is this maybe about this?
That our ears sense much more than theory of Nyquist theorem maybe.

So AD converters sampling the true event of music played in a room is maybe about something more than than these +/- 32767 levels at 44100 samples/s. And also big difference in DA converters as I told earlier - one is not like the other. Just changing capacitors as I had in that story. I also heard a shootout here at KVR some years ago over AD converters, it was rather interesting as well.

And since you clearly can hear the aliasing introduced at 44100 Hz compared to 96k or whatever higher - that is also part of the picture.

Anybody spending enough time mixing or listening to music a lot will hear these things.

So this is my game in this thread. So think about it a bit and it will become clear that every step counts.

- you can't hear this
- it's all in your head

Yeah, it is - and our brain is just beyond incredible in what it does interpreting audio for us to enjoy.

And even more incredible what we can obtain in skills challenging anything like an athlet or virtuoso soloist of some sort.

Post Reply

Return to “Production Techniques”