[jnorris]

Why? How loud do you play your stereo? If you’re getting into the zone between 16 bit and 20 bit, you should be concerned about damaging your hearing.

The difference between 16 and 24 bit is not how loud it gets, but how quiet it gets. It's the number of bits that are used to define the amplitude of the waveform.

 

[RRod]

The thing is, the spec should have been 20-bits/48kHz and then we may not be arguing about this.

Wasn't it going to be 14 bit at first?

But how do you know you don't hear increased resolution? And while you may not hear the extended frequencies, the idea is for the frequencies you do hear to sound better.

For integer formats, resolution and dynamic range are the same coin looked at on either side. If you take as example 16 bits mapped into 24 bits, you can either see that as padding zeros before the 1st bit, where the extra 0s are increased 'detail' or 'resolution', or padding after the 16th bit, where the extra 0s are increased 'dynamic range'. Saying you need one is saying you need the other.

So far, I like the way 96 kilohertz 24-bit sounds. Another possibility with these higher bitrate recordings is that more care is taken mastering them back into digital format than the CD creators.

They're already in digital format... And well cared-for recording/mixing/mastering has always produced good results. The issue with your argumentation is that you are begging the question: you have assumed 24/96 is audibly superior. The whole issue is how you actually prove that in an unbiased way.

 

[71 dB]

So far, I like the way 96 kilohertz 24-bit sounds. Another possibility with these higher bitrate recordings is that more care is taken mastering them back into digital format than the CD creators.

I also love SACDs because they are consistently brilliantly prodused recordings (+ multichannel!), but that doesn't mean 24/96 is perceptually better than 16/44.1. It means good mastering/production is better than bad mastering/production.

 

[danadam]

The difference between 16 and 24 bit is not how loud it gets,

So... are you saying that in the following video you are hearing the sound till the end, at your usual listening volume level? And that you need to hear something even quieter?

 

[71 dB]

The difference between 16 and 24 bit is not how loud it gets, but how quiet it gets. It's the number of bits that are used to define the amplitude of the waveform.

Yes, if you set the MSB to equal loudness. Let's assume your volume setting is such that MSB equals 100 dB. At 16 bit LSB equals 10 dB. That is above the threshold of hearing between about 300 Hz and 7000 Hz and below elsewhere. However, if you listen to music loud, the threshold of hearing goes up a lot temporarily. Also, the noise level in very quiet living room is about 30 dB, already masking your LSB! If that's not enough, that example is true if dither is not used, but dither is used on every CD! So, the "resolution" of CD is perceptually even better than that! At 20 bit LSB would equal -14 dB and at 24 bit LSB -38 dB. Amplifiers alone have such a noise floor that any potential benefit of going beyond 16 bits is easily lost. So, it is a question of how quiet is should get, what is relevant in real life.

 

[gregorio]

[1] Here is the bottom line... Recordings at 96k/24 do, unquestionably have the potential for better sound.
[2] Whether you can hear the difference is entirely up to the ears (and prejudices) of the beholder.

1. "Better", how?
2. Yes, assuming the beholder is not a human being.

[1] The thing is, if there are going to be issues by hitting boundaries, then having a spec that goes past those boundaries you don't have those issues.
[2] Also, in some 24/96 recordings, you do see the sound does go past 20kHz. [2a] Plus, with 24-bit, you have more bits to give more resolution.
[3] But how do you know you don't hear increased resolution?

1. Good, glad we cleared that up, 44.1 it is then.
2. Yes, we do see it but we don't hear it.
2a. More bits does not give more resolution.
3. Because there is no increased resolution to hear! It's a common audiophile myth that more bits = more resolution. To understand what actually happens and why there is no more resolution try this post.

To me its obvious 192 kHz and 24 bit just sounds more full, intimate, and less sharp than 16 bit 44.1.

How do you know that the recording is not supposed to sound more sharp, less intimate/full and that 16/44 is therefore being reproduced more accurately?

G

 

[frodeni]

Because listening tests tell us that. Sounds that we do hear can't sound better (better in what way?), because 16/44.1 already can reproduce them completely.

People who think high res audio sounds better are victims of thinking more must be better, victims of placebo effect and victims of not understanding enough digital audio or human hearing.

The proof for 16/44.1 is made by using sine waves. Beyond that, we do not know very much. You do not know everything ever written, just because you know the alphabet.

Is 16/44.1 enough for vector reproduction? Do you have any proof of that? Is that a headset limitation? How do you know that it is not?

 

[bigshot]

The difference between 16 and 24 bit is not how loud it gets, but how quiet it gets.

Absolutely correct. But assuming the music you're listening to is normalized, a CD has a dynamic range of almost 100dB. Your body has a noise floor of about 15dB. That's why when people are in caves underground, they can hear their own blood flow through their ears. The quietest room you would ever find has a noise floor of 30dB. So assuming you're wearing headphones, you probably have a noise floor of 20dB or so. Therefore, in order to hear all the way down to the lower limit of 16 bits, you need to boost the volume 20dB. That makes the peaks 120 dB, which is the threshold of pain and the point beyond which you incur hearing damage in short term exposure.

Do you follow the logic there? In order to hear all the bits in a CD, you have to incur hearing damage.

It doesn't matter anyway, because recorded music rarely has a dynamic range wider than 45dB. And your own ears can only hear around 45dB of dynamics at one time. In order to hear louder or quieter, they need a few minutes to acclimate. That's why LP records sound fine. They have around 35 to 40dB on the average.

 

[frodeni]

1. "Better", how?
2. Yes, assuming the beholder is not a human being.



1. Good, glad we cleared that up, 44.1 it is then.
2. Yes, we do see it but we don't hear it.
2a. More bits does not give more resolution.
2b. No you don't, a bit error is a bit error.
3. Because there is no increased resolution to hear! It's a common audiophile myth that more bits = more resolution. To understand what actually happens and why there is no more resolution try this post.



How do you know that the recording is not supposed to sound more sharp, less intimate/full and that 16/44 is therefore being reproduced more accurately?

G

1. Better at soundstage, particularly for vector reproduction. Better for articulation. That could be the case, I just cannot get to work for my part, due to loss of signal quality, using USB.
2. This is about inferring complex sound, not sine waves. If you got any scientific papers on complex sound, please share.

As for when the boundaries are hit, we got a fairly good understanding of this, for sine waves. As for S/N for music reproduction, it is really hard to tell sounds apart at great dynamic range. At least I struggle. But for complex sounds, it is hardly understood how that is experienced, and why some people got some gifts, while lacking others.

Until people got any form of proof for vector sound and complex issues on timing, by not using sine waves, nothing is even close to a given. Even in the nature sciences.

 

[bigshot]

1. Better at soundstage, particularly for vector reproduction. Better for articulation.
2. This is about inferring complex sound, not sine waves. If you got any scientific papers on complex sound, please share.

1) Soundstage and sound location is a function of the mix and the acoustics between the transducers and the listener, not the recording medium. Usually when people say "better soundstage" they mean "better expectation bias". Articulation would be covered under distortion. All digital formats have inaudible levels of distortion.

2) All audible sounds, both simple and complex are *perfectly* reconstructed with 16/44.1 and can be represented as sine waves. All of them. Again, not being able to accurately reconstruct sign waves would fall under the category of distortion. See #1.

See how interesting it is to hang out in Sound Science! You learn something new every day!

 

[RRod]

Until people got any form of proof for vector sound and complex issues on timing, by not using sine waves, nothing is even close to a given. Even in the nature sciences.

You're going to need to define 'vector sound'. Also, the sine-wavy aspect of all this is due to Fourier analysis, and I task you to get a voltage function out of a mic that doesn't have a Fourier decomposition...

 

[bigshot]

I believe by vector sound he means sound location within the soundstage.

 

[RRod]

I believe by vector sound he means sound location within the soundstage.

So wanting to get better location by adding bits/samples rather than channels (or using headphone virtualization)? L'sigh.

 

[bigshot]

Yeah... maybe he assumes that bigger numbers means better channel separation of something.

 

[71 dB]

The proof for 16/44.1 is made by using sine waves.

All possible signals are a sum of sine waves, so if you prove something for sine waves and the system is linear, the proof is valid for sum of sine waves too.

Beyond that, we do not know very much.

Well, clearly you don't know much, so you are partly correct.

You do not know everything ever written, just because you know the alphabet.

No, but if I can prove I can write every alphabet, it's easy to show I could write anything if I wanted.

Is 16/44.1 enough for vector reproduction? Do you have any proof of that? Is that a headset limitation? How do you know that it is not?

Vector reproduction? Is that a term coined by highres advocates? Music is signals consisting of frequencies between 20 and 20 000 Hz. 16/44.1 is enough for that. Don't believe me, do some ABX tests to see it yourself.