Is there any music that actually requires 24 bits for replay?

[oleg87]

If the recording was mastered properly, there should be no distortion due to quantization. Just uncorrelated noise.

 

[paulg1]

Distortion is not the correct really word for this, it's properly just a lack or resolution. 8-bit resolution is 1part in 256, which equates to about .25% NOT following the exact signal, which is about .25% distortion in my math book. It still doesn't change my point, the fact that soft sounds are recorded at much lower resolution than the stated 16-bit maximum resolution, and that's a good reason to use higher bit depth, to increase the resolution of these softer sounds.

 

[paulg1]

Pardon my typos above. This also in a way justifies the trend to master at higher levels, as overall the effective bit-depth increases for the softer sounds.

 

[Blumlein 88]

I've read all the posting here, and I think that there is something important missing: resolution of the SOFTER SOUNDS.

A CD has 16 bits of resolution that 1 part in 65,536, that like .0015% distortion, incredibly low, but that's at 96 dB. Hardly anything is recorded at 96 dB on a CD or even 90 dB. Most of the sounds on a CD are recorded at a much lower than this, about 40 dB down form 96 dB. Those are not recorded with 16-bit resolution, they are recorded with only 8 bits or resolution. That's one-part in 256, or 0.25% distortion, about the same as a vinyl record.

Everything gets worse as the signal level drops. at 50 dB down, the resolution is 4 bits, that's 1 part in 16, or about 6% distortion. Bear in mind that these distortions in digital are not benign, even-order harmonic distortions, they consist of mixed harmonics and mixed products. And good vinyl reproduction still has another 10 dB of dynamic range beyond this. That's not to say that vinyl is superior, Vinyl has a lot of other problems, mechanical resonances being the most troublesome, but the harmonic distortion level does not drop with decreasing volume like it does with PCM digital.

There is a compelling reason for higher bit depth in PCM digital recordings, and that is to increase the resolution of the softer sounds. It is in this realm that capable listeners can identify the difference between 16-bit 44.1 kHz and the higher bit depth recordings. These differences are subtle, because of the low level, but they show up on material with a lot of softer sounds, in particular, orchestral (classic), and music recorded in a live setting, where reverberations and decays, and instantaneous dynamic shifts can be heard.

What is Dithering? Using Dithering to Eliminate Quantization Distortion - Technical Articles

Learn how dithering can be added to a signal to improve the performance of an analog-to-digital conversion system by eliminating quantization error and distortion.

www.allaboutcircuits.com

Yet with dithering the problem is pretty much solved. It reduces quantization noise at low bit levels and reduces harmonic distortion. Noise shaped dithering allows you to get resolution below 16 bits with a 16 bit signal. It can be and has been demonstrated. Low levels are not a problem of resolution at that point. The softer sounds will come through just fine.

 

[paulg1]

Yes, that is very true, but you are still left with low resolution of the softer sounds. How else can you explain certain individual being able to consistently identify 16-bit vs 24 bit versions of the same recording in double-blind testing? Amir has stated that he can do this on certain material. That is how one hears the difference, it's in the softer sounds they are subtle but there. It is very much dependent on the type of recorded music.

 

[Blumlein 88]

Yes, that is very true, but you are still left with low resolution of the softer sounds. How else can you explain certain individual being able to consistently identify 16-bit vs 24 bit versions of the same recording in double-blind testing? Amir has stated that he can do this on certain material. That is how one hears the difference, it's in the softer sounds they are subtle but there. It is very much dependent on the type of recorded music.

PHILIPS RED BOOK

But isn't Amir's point rather straightforward, though? I'm not sure we need neither simulations nor blind tests here. It's a simple question of logic. In acoustic music settings, min to max dynamic range may reach 110-120 db in the presence region (there will always be a higher noise floor in...

www.audiosciencereview.com

You can try this short track I created. Where you can hear an amplified version of -118 db tone with dithered 16 bit.


I don't know of people who detect 16 bit dithered vs 24 bit in double blind testing with music. Maybe with some peculiar test tones, but I'm not sure you could do that either without playing only the very low level signal at extremely high volume. This would mean in a normal context if anything other than the low signal were there you would deafen yourself.

 

[antcollinet]

I've read all the posting here, and I think that there is something important missing: resolution of the SOFTER SOUNDS.

A CD has 16 bits of resolution that 1 part in 65,536, that like .0015% distortion, incredibly low, but that's at 96 dB. Hardly anything is recorded at 96 dB on a CD or even 90 dB. Most of the sounds on a CD are recorded at a much lower than this, about 40 dB down form 96 dB. Those are not recorded with 16-bit resolution, they are recorded with only 8 bits or resolution. That's one-part in 256, or 0.25% distortion, about the same as a vinyl record.

Everything gets worse as the signal level drops. at 50 dB down, the resolution is 4 bits, that's 1 part in 16, or about 6% distortion. Bear in mind that these distortions in digital are not benign, even-order harmonic distortions, they consist of mixed harmonics and mixed products. And good vinyl reproduction still has another 10 dB of dynamic range beyond this. That's not to say that vinyl is superior, Vinyl has a lot of other problems, mechanical resonances being the most troublesome, but the harmonic distortion level does not drop with decreasing volume like it does with PCM digital.

There is a compelling reason for higher bit depth in PCM digital recordings, and that is to increase the resolution of the softer sounds. It is in this realm that capable listeners can identify the difference between 16-bit 44.1 kHz and the higher bit depth recordings. These differences are subtle, because of the low level, but they show up on material with a lot of softer sounds, in particular, orchestral (classic), and music recorded in a live setting, where reverberations and decays, and instantaneous dynamic shifts can be heard.

Ever heard of dither? That gives you around 120dB effective dynamic range.
Have a read here:

24/192 Music Downloads are Very Silly Indeed

people.xiph.org

Particularly the section "The dynamic range of 16 bits"

Plus most stuff isn't recorded 40dB down. Most stuff (at least that I have analysed) is around -15dB average, with peaks going close to 0dB (and that's for the stuff that hasn't been loudness war'd to death.

 

[antcollinet]

Yes, that is very true, but you are still left with low resolution of the softer sounds. How else can you explain certain individual being able to consistently identify 16-bit vs 24 bit versions of the same recording in double-blind testing? Amir has stated that he can do this on certain material. That is how one hears the difference, it's in the softer sounds they are subtle but there. It is very much dependent on the type of recorded music.

Amir has stated he did this by cranking the volume up during the silence between tracks, and listening for the noise.

 

[paulg1]

Amir has stated somewhere on ASR that he passed double-blind testing of this type. Let's ask Amir.

This has been studied and measured. Early scientific testing showed that most people cannot repeatable recognize the difference between 16-bit 44 Khz and 24-bit 96 kHz. But later on, other studies confirmed that a small number of people can consistently here these differences. I looked quickly, but at the moment I cannot locate these studies. I will find those studies.

 

[Blumlein 88]

Amir has stated somewhere on ASR that he passed double-blind testing of this type. Let's ask Amir.

This has been studied and measured. Early scientific testing showed that most people cannot repeatable recognize the difference between 16-bit 44 Khz and 24-bit 96 kHz. But later on, other studies confirmed that a small number of people can consistently here these differences. I looked quickly, but at the moment I cannot locate these studies. I will find those studies.

antcollinet explained how Amir did that. Listening to only a silent portion of the music at extremely high volume levels amped to higher than normal to hear a difference in the noise floor. Had he then played music he would have blown out his phones/speakers and ears. At a level you can listen to music it would not be heard.

 

[raif71]

This particular song sounds better with 24 bits

 

[DVDdoug]

Distortion is not the correct really word for this, it's properly just a lack or resolution.

I think it can be considered distortion (it is distortion of the waveform). It's normally called quantization noise and it sounds like noise or "fuzz" riding on top of the signal. Like regular analog noise it's most noticeable with quiet parts but unlike analog noise it goes-away completely with digital silence.

If you make an 8-bit file in Audacity you can hear it. (Disable dither because it changes the nature of the noise and you can't get pure digital silence.)

Amir has stated he did this by cranking the volume up during the silence between tracks, and listening for the noise.

Which means if he didn't "cheat" with the volume control he might get distortion during the normal-volume parts, or blow-out his speakers, or blow-out his his ears and get temporary hearing loss and he wouldn't hear it.*

* rafi71 beat me to it.

 

[dlaloum]

Take a 24-bit track. Convert to 16 bits. Convert back to 24 bits. Use DeltaWave to compare original 24-bit track with converted track.

I'm getting results such as the following:

View attachment 416729

I understand that it might be useful sticking to 24 bits in the studio (for further processing), but what's the point of using 24-bit tracks for replay?

FWIW, good ol' redbook has always sounded perfectly fine to me.

Mani.

At the recording and mastering stages, 32 bits would be nice, but we have to live with 24bits

At the playback, post mastering stage, 16bits is ample!!

Having said that - the 16bit standard was set when everything was stereo (mostly... quad had already faded, movie surround was still small)... and there was little or no processing done at the consumer end of the pipeline.

But today, many of us take the 16bit signal (often multichannel), we process it for Room EQ, we apply personal preference target curves, and then we finally send it to the DAC to feed our speakers (whether the DAC is external, in an AVR or in the speaker itself!).

That means we are doing a lot of processing, very similar to the recording and mastering process, in our homes at the point of listening/consumption.

And what some are experiencing is that we need to take care with gain and dynamics to ensure that we don't boost things outside the fixed limitations of our 16bits....

A lot of AVR's do their processing at 24bit/48kHz.... possibly to provide the required processing headroom?

 

[Old_School_Brad]

Distortion is not the correct really word for this, it's properly just a lack or resolution. 8-bit resolution is 1part in 256, which equates to about .25% NOT following the exact signal, which is about .25% distortion in my math book. It still doesn't change my point, the fact that soft sounds are recorded at much lower resolution than the stated 16-bit maximum resolution, and that's a good reason to use higher bit depth, to increase the resolution of these softer sounds.

Here are a few things I think are worth looking into -if nothing else, just to better understand your own limitations. I’ve found that recognizing my own limitations can be surprisingly enlightening.

Dithering explaines - With listening tests

What is audible? - Video explanation

And finally, the most shared link on ASR:
Klippel Distortion Test

 

[paulg1]

Thanks, Old_School_Brad. I appreciate the opportunity to learn. BTW, I know what dithering is. I read the article and tried the listening tests. I took the Kippel test also with two different music, tracks.

I definitely agree that dithering is an effective technique to reduce harmonic distortion of the softer sounds. But I honestly do not think that 16-bit audio is "good enough". Music reproduction is still distanced compared to live, unamplified music and voices, even on the best systems I've ever heard. Audio should be able to improve, we shouldn't be thinking that CD is the end of the road.

I know it's heresy to say this in this forum, but here goes: Despite its serious flaws, I still find that the best examples of vinyl playback reveal low-level details that seem to be missing in CD quality digital. Higher resolution recording is available and capable of reaching better overall quality and can most likely fill that gap. Dithering is remarkable, but it won't turn a sow's ear into a silk purse.

 

[Old_School_Brad]

Thanks, Old_School_Brad. I appreciate the opportunity to learn. BTW, I know what dithering is. I read the article and tried the listening tests. I took the Kippel test also with two different music, tracks.

I definitely agree that dithering is an effective technique to reduce harmonic distortion of the softer sounds. But I honestly do not think that 16-bit audio is "good enough". Music reproduction is still distanced compared to live, unamplified music and voices, even on the best systems I've ever heard. Audio should be able to improve, we shouldn't be thinking that CD is the end of the road.

I know it's heresy to say this in this forum, but here goes: Despite its serious flaws, I still find that the best examples of vinyl playback reveal low-level details that seem to be missing in CD quality digital. Higher resolution recording is available and capable of reaching better overall quality and can most likely fill that gap. Dithering is remarkable, but it won't turn a sow's ear into a silk purse.

On one hand, you seem to acknowledge the science and the thresholds involved in hearing.
On the other hand, you circle right back to making claims that aren’t backed by any evidence.

Why is that? And why not just provide examples and evidence that support your opinion-based argument?

 

[Blumlein 88]

Thanks, Old_School_Brad. I appreciate the opportunity to learn. BTW, I know what dithering is. I read the article and tried the listening tests. I took the Kippel test also with two different music, tracks.

I definitely agree that dithering is an effective technique to reduce harmonic distortion of the softer sounds. But I honestly do not think that 16-bit audio is "good enough". Music reproduction is still distanced compared to live, unamplified music and voices, even on the best systems I've ever heard. Audio should be able to improve, we shouldn't be thinking that CD is the end of the road.

I know it's heresy to say this in this forum, but here goes: Despite its serious flaws, I still find that the best examples of vinyl playback reveal low-level details that seem to be missing in CD quality digital. Higher resolution recording is available and capable of reaching better overall quality and can most likely fill that gap. Dithering is remarkable, but it won't turn a sow's ear into a silk purse.

So how do you square the LP thing with the fact digital recordings of LPs are indistinguishable from the LP itself? Clearly if the medium can do that it isn't lacking resolution vs LP. And you cannot do any better with higher resolution so I don't know why you think that fills any gap.

The reason CD is not equal to live, un-amplified music and voices is due to many factors none of which have to do with deficiencies in digital audio. It has to do with microphones not hearing like ears, with stereo not being capable of recreating that live sound, and what you are hearing more than 99% of the time has been subjected to some processing that makes a fully lifelike result an impossibility.

 

[Frank Dernie]

I still find that the best examples of vinyl playback reveal low-level details that seem to be missing in CD quality digital.

Back when I was involved it was normal to boost low levels of music for LP cuts to keep the quiet bits out of the noise.
Could this be what you are hearing?
Some early CDs (16 bit is plenty enough dynamic range not to need this) were issued without this adjustment, so comparing CD and LP would certainly result in the quieter bits being more clearly audible.
I am not sure how things are today with loudness wars etc

 

[Frank Dernie]

So how do you square the LP thing with the fact digital recordings of LPs are indistinguishable from the LP itself? Clearly if the medium can do that it isn't lacking resolution vs LP. And you cannot do any better with higher resolution so I don't know why you think that fills any gap.

The reason CD is not equal to live, un-amplified music and voices is due to many factors none of which have to do with deficiencies in digital audio. It has to do with microphones not hearing like ears, with stereo not being capable of recreating that live sound, and what you are hearing more than 99% of the time has been subjected to some processing that makes a fully lifelike result an impossibility.

I 100% agree with every word of this.

 

[Tell]

I've read all the posting here, and I think that there is something important missing: resolution of the SOFTER SOUNDS.

A CD has 16 bits of resolution that 1 part in 65,536, that like .0015% distortion, incredibly low, but that's at 96 dB. Hardly anything is recorded at 96 dB on a CD or even 90 dB. Most of the sounds on a CD are recorded at a much lower than this, about 40 dB down form 96 dB. Those are not recorded with 16-bit resolution, they are recorded with only 8 bits or resolution. That's one-part in 256, or 0.25% distortion, about the same as a vinyl record.

Everything gets worse as the signal level drops. at 50 dB down, the resolution is 4 bits, that's 1 part in 16, or about 6% distortion. Bear in mind that these distortions in digital are not benign, even-order harmonic distortions, they consist of mixed harmonics and mixed products. And good vinyl reproduction still has another 10 dB of dynamic range beyond this. That's not to say that vinyl is superior, Vinyl has a lot of other problems, mechanical resonances being the most troublesome, but the harmonic distortion level does not drop with decreasing volume like it does with PCM digital.

There is a compelling reason for higher bit depth in PCM digital recordings, and that is to increase the resolution of the softer sounds. It is in this realm that capable listeners can identify the difference between 16-bit 44.1 kHz and the higher bit depth recordings. These differences are subtle, because of the low level, but they show up on material with a lot of softer sounds, in particular, orchestral (classic), and music recorded in a live setting, where reverberations and decays, and instantaneous dynamic shifts can be heard.

So when was the last time you heard the dither noise from a 16bit file? I don't think I've ever heard it even running my IEM at max volume listening to reverb trails or whatever higher dynamic range stuff.