24bit vs 16bit, the myth exploded!

[Davesrose]

It is the same with 4k 1080p picture. It actually compresses to a smaller 1080p file than 1080p does, invented recently, 1080p that is recorded at twice the definition (4k) has a cleaner downscale and a smaller lossless algorithm compression due to the wider range of pixel colors, especially when taking HDR into the mix. A case of compressing a more complete information index into a small size rather than not being able to compress files too well because of the

(Genius) earlier who suggested entropy causes a problem. He was wrong and totally missed my point this is a better way to explain my philosophy on compressing audio.

This is utterly false. I was specific about blu-ray and UHD standards in a previous response to you where you didn't understand the difference between an uncompressed audio CD standard vs CD containing compressed mp3s. It seems you're trying to understand video codecs: formats for compressing video to a smaller file size. 1080P blu-rays are encoded in MPEG-4 H.264. A good video encoding program will let you specify properties such as 2 pass encoding (where the software first goes through the video to analyze and then finalize a variable bitrate that's high enough to maintain quality, but small enough to save disc space). If all things were equal with a 1080P resolution video compared to a UHD resolution (and barring frame rate or color depth differences), the 4K video takes up 4 times the data. H.265 was implemented with UHD so that UHD movies could still fit on a blu-ray (with up to 100GB storage). If you compress a 4K movie with the h.264 codecs common with HD, you still wind up with a much larger file size. If you encode a 1080P movie in h.265, you wind up with a much smaller file size than UHD.

 

[KeithPhantom]

quantinization errors make lower sampling rates harder to do lossy and lossless compression without losing information.

It is the same with 4k 1080p picture. It actually compresses to a smaller 1080p file than 1080p does, invented recently, 1080p that is recorded at twice the definition (4k) has a cleaner downscale and a smaller lossless algorithm compression due to the wider range of pixel colors, especially when taking HDR into the mix. A case of compressing a more complete information index into a small size rather than not being able to compress files too well because of the

There is a problem with your analogy. More points above the ones needed by Nyquist are not needed if we account for the accepted limitations of the human ear. 16 bit undithered is more than enough since errors and the noise floor are ~ -90 dBFS at worst from the fundamental. In terms of 4K and 1080p, the differences are apparent when screen size is increased, but we can't do that with audio (we can't zoom with audio. I'malso ignorant in video, so correct me if needed). For me, I would worry more about transducers, with the distortion of the best ones averaging -60 to -70 dB occasionally in the best of the cases. Electronic, and especially, distortion caused by the files and encoding methods are way down in the list of things to take care of.

 

[Davesrose]

There is a problem with your analogy. More points above the ones needed by Nyquist are not needed if we account for the accepted limitations of the human ear. 16 bit undithered is more than enough since errors and the noise floor are ~ -90 dBFS at worst from the fundamental. In terms of 4K and 1080p, the differences are apparent when screen size is increased, but we can't do that with audio (we can't zoom with audio. I'malso ignorant in video, so correct me if needed). For me, I would worry more about transducers, with the distortion of the best ones averaging -60 to -70 dB occasionally in the best of the cases. Electronic, and especially, distortion caused by the files and encoding methods are way down in the list of things to take care of.

I think we should also be specific about audio compression (compressing dynamic range) vs @Lazysnakes miss-use of an analogy with 4K video compression: which is about file compression.

 

[KeithPhantom]

I think we should also be specific about audio compression (compressing dynamic range) vs @Lazysnakes miss-use of an analogy with 4K video compression: which is about file compression.

I would like to get more into video, do you have anything like sound science but for video? It could a forum, but a YouTube channel/video would be perfect.

 

[Davesrose]

I would like to get more into video, do you have anything like sound science but for video? It could a forum, but a YouTube channel/video would be perfect.

When it comes to consumer products, I think AVS is pretty good: they do have forums on firmware updates and settings for TVs, receivers, etc. Rtings is also a great site for TV reviews. I'm involved with 3D graphics and video editing....so I'm used more with sites for professional topics, training! There is pretty good info out there for resolution benefits vs certain viewing distances. Photography has also been a serious hobby with me (starting with the film days). There are sites like Head-Fi for that (and less biased like headphone reviews, as photography is applied recording). They do have more threads now for videography...which can be a separate application than cinema specific camera equipment. There's also YouTube channels devoted to photography news (some get more money with their own beginers guides to photography).

I notice Samsung is really advertising their 8K QLED TVs, but there isn't going to be true 8K film sources for awhile (there have been just a few 8K film restorations, while many 4K...only the most recent big production movies are done in 4K digital intermediates). Like plasma was rated highly before, the 4K OLEDs still score higher than QLEDs because of clearly defined pixels (QLED is brighter and better for brighter rooms, with OLED being better with blacks and great for rooms you can control lighting). I also think 4K standards are great for cinephiles as it reaches the realized limits of 35mm film formats (may be last older film you need to double dip with....and some new Dolby Atmos/DTS:X remixes have been nice).

 

[bfreedma]

When it comes to consumer products, I think AVS is pretty good: they do have forums on firmware updates and settings for TVs, receivers, etc. Rtings is also a great site for TV reviews. I'm involved with 3D graphics and video editing....so I'm used more with sites for professional topics, training! There is pretty good info out there for resolution benefits vs certain viewing distances. Photography has also been a serious hobby with me (starting with the film days). There are sites like Head-Fi for that (and less biased like headphone reviews, as photography is applied recording). They do have more threads now for videography...which can be a separate application than cinema specific camera equipment. There's also YouTube channels devoted to photography news (some get more money with their own beginers guides to photography).

I notice Samsung is really advertising their 8K QLED TVs, but there isn't going to be true 8K film sources for awhile (there have been just a few 8K film restorations, while many 4K...only the most recent big production movies are done in 4K digital intermediates). Like plasma was rated highly before, the 4K OLEDs still score higher than QLEDs because of clearly defined pixels (QLED is brighter and better for brighter rooms, with OLED being better with blacks and great for rooms you can control lighting). I also think 4K standards are great for cinephiles as it reaches the realized limits of 35mm film formats (may be last older film you need to double dip with....and some new Dolby Atmos/DTS:X remixes have been nice).

Agree on all points regarding commercial 8k, but there is some limited but very high quality content out there for anyone looking for 8k native demo content. Check https://drive.google.com/drive/folders/1TSdV36G_npDtjRJze54GEYpdxpBt7nCK. (sorry about the OT, but figured some might be interested...)

Before anyone asks, I would not spend more for an 8k TV this year than a 4k set. The only reason I have an 8k set is that Samsung was having a very hard time moving them late last year and I actually negotiated a better price for a TOTL 8k Samsung than was possible on the 4k equivalent. The store really wanted to move stock..

 

[71 dB]

1. Mmm I may not, but I have no interest in believing in something with no evidence, just physical proof that 24 bit is absolutely better than 16 bit depth, regardless of human abilities to tell the difference.

2. Likewise I am not a high-end hi-fi audiophile, and my interest in pointing out the glaring flaws of the original post is purely scientific.

3. I believe the relative factor is still not of your understanding here. The idea that the change from normal music to the maximum is present in going from 8 to 16bit,
Even there most people never use the range in record broad music, but narrower music interest as well as home and public theater do use difference of this and more. And on recording that comes from rock and such still have a difference that could be widened with 24 bit.

4. The use of other techniques is still sound dampening by some degree, and although I wouldn't have this huge expensive push toward things that are not really making a difference for your hi-fi music, I personally don't see such a ridiculous hostility toward larger bit depth technology and not utilizing it to a further degree.

5. The military and metal detection use sampling above 1gigahertz (1,000,000 hertz) in order to produce accurate measurements of sound waves around 1/10 of that, or 100,000 htz which are used for echo location and for slowing down of the recording. The latter which could actually help a lot with pitch control and redefinition.

6. Also more advanced techniques for algorithm coding to shrink the size of such recording after first making it, including lossy compression of dense music that has as you say "more information than nessicary for human hearing " could be used down the road to keep files the same size as lossless 16 bit and have the same quality, with possibly more relevant information. And even compressed smaller, lossy, and still sound the same because of the information gained from the curves and the algorithmic curve prediction for frequencies of sound being predicted, due to 4 times the recording points. This can not be improved without wider acceptance of 24 bit depth recording and even consumption.

7. Although considered irrelevant by some here, no one has provided evidence that the future is not brighter with 24 bit depth, mark my words, 24 bit will be used to noticeably make audio better. And as of 2021/2020 there is no longer a compelling counter that we shouldn't use it.

8. I was messaged by someone that said it themselves: some recordings are only available in 24 bit depth, there is plenty of storage for the increase, there is less need for dithering and other labor techniques, less mixing nessicary, and a broader range for deliberate manipulation of the sounds.

9. Cannons and such sounds, bass boosting without using amplification, and public theater explosion sounds can also accept an improvement. I don't understand all of this yet but I will unfollow this forum because I have reduced the responses from the majority helping my understanding and contributing, to jabs at my reputation and blatant "no you" and "you're wrong" "you knownothing" statements, which besides being unhelpful and far from the truth are damaging to the spirit of education and science.

1. This is not about what is "absolutely better." Infinite sampling rate together infinite bit depth ( ∞ Hz / ∞ bits) is "absolutely" best, but any serious person agrees what counts here is human hearing and what we can hear. What is good enough? What is not?

2. The facts of digital audio don't care about whether you or anyone else is a high-end hi-fi audiophile.

3. What is "normal music"? What is "maximum music"? What are "narrow" and "broad" music? 8 bits is not considered enough for high quality consumer audio. I don't consider it either. It's 5 bits less than what I consider adequate for transparent consumer audio. 8 bit is low quality digital audio. Going from 8 bits to 16 bits means going to the other side of the threshold line. Once you are on the other side, it doesn't matter how far you go.

4. This is not hostility toward larger bit depth technology. This is about calling out the snakeoil bs about 24 bit providing higher fidelity than 16 bit. In consumer audio this is not the case. In music production things are different, there are additional things to consider so in music productio 24 bit is definitely better than 16 bit, but ones the music is produced and scaled optimately adding 16 bit dither and dropping the last 8 bits (most of them probably just noise from various sources) out means zero loss of perceptual fidelity. In other words 24 bit technology in studios allows us to take the full potential out of 16 bit technology.

5. Well, I am not detecting enemy tanks when I listen to music. I am not an expert on "metal detection." If 1 MHz (megahertz, not gigahertz I believe) sampling frequency is needed then that's what is used.

6. Lossy coding is whole another issue. This is about non-data-compressed music. This topic explodes when we start to talk about lossy compression.

7. So music sounds better to you when there's stuff you can't hear? Have you ever listened to the difference of 16 and 24 bit audio (same master?) It's noise. Insanely quiet noise. You can't hear it at any rational listening level and even if you could, the music masks it.

8. 24 bit serves it's purpose. So does 16 bit. They sound the same (if from same master). Use 24 bit all you want, but when you say 16 bit gives less perceptual fidelity you are wrong. Is 24 bit the magic number for you? No need for more? 32 bit? 64 bit? 1348457 bit? How far do you want to go?

9. Considering human hearing and how loud sounds must be mixed to make any sense 16 bit is enough. Public theaters are noisy places. Heck, the audience might not notice if the theater sound was (dithered) 8 bit sound because the noise from pop corn eaters would mask the dither noise.

 

[castleofargh]

Wouldn't you have to change the mixing and thus the sound ratios? Why does 8 bit sound so much worse than 16 bit if the recording is in 16 bit and gets compressed, and why
are go u totally and completely backwards when it comes to that. Because the difference between 8 bit and 16 bit is NOT in the quietest sound, it is the median to loudest values that have the most relevant change, I'm arguing the same goes for 24 bit depth.

Furthermore cannons are well over 150db. And I'd love to be able to actually here one so the point still stands.

The record for human audio is 171db in an automobile recorded early this year, 2020 with a custom setup. In order to achieve the record a human must withstand the audio present and he did.

Little of what you said makes sense. Again, when you go from one bit depth to a smaller one, what we do is truncation. We get rid of the lower bits. It is not a compression!
As to 8bit sounding "so much worse than 16bit", instead of repeating what I already told you in PM that you didn't understand or didn't believe, here is an actual example to download(3.8Mo flac):
https://www.dropbox.com/s/jyzibehg3bhe4ux/The Vampyre Of Time And Memory .flac?dl=1
I took a sample of the song, saved it to 16/44 and 8/44, then I put both into the same 16/44 track, one after the other. So the container is 16bit but the second sample is a signal that only had 8bit of data when it was added to the track. the missing 8bit for that second part can't magically come back from the bit cemetery, so you get to hear the quantization noise at 8bit on one of the versions in the file. Without getting whatever issues people may get because their audio app doesn't know how to play 8bit, or whatever.



"does anyone ever get this right?"

 

[71 dB]

1. Using higher sampling allows better definition of the curve. With practice, and a genius interested in music with an immeasurable IQ (none such genius has existed in known history, unfortunately), I suspect it is easier to compress 24 bit to a smaller size than 16 bit, or to be more specific, 196k htz to be compressed to 44k htz with wave forms that make music sound better because the quantinization errors make lower sampling rates harder to do lossy and lossless compression without losing information.

2. It is the same with 4k 1080p picture. It actually compresses to a smaller 1080p file than 1080p does, invented recently, 1080p that is recorded at twice the definition (4k) has a cleaner downscale and a smaller lossless algorithm compression due to the wider range of pixel colors, especially when taking HDR into the mix. A case of compressing a more complete information index into a small size rather than not being able to compress files too well because of the

(Genius) earlier who suggested entropy causes a problem. He was wrong and totally missed my point this is a better way to explain my philosophy on compressing audio.

1. No no no!! Higher sampling frequency allows HIGHER FREQUENCIES. Properly bandlimited signal are represented (within the dynamic range) with 100 % accuracy. NOTHING can be better definition than 100 %. When the signal is bandlimited (for example all frequencies above 20 kHz are filtered away because humans can't hear above that) and we use the sampling frequency on 44.1 kHz (at least twice the highest frequencies in the signal) the signal can't make any kind of curves BETWEEN the sample points. In fact there is only ONE solution for the signal to have it's curvy shape while going through the sample points. That is the WHOLE point of sampling theorem, something you clearly have not studied. Sampling is not about taking samples of the signal to get a somewhat accurate picture of what the signal does. It is about capturing 100 % of the information about what the signal is doing. Because the sample points must be quantized to certain amount of bits, noise gets added to the signal. If that noise is quiet enough so you don't hear it the signal is perceptually 100 % perfect. Higher sampling rates allows music for bats if that's something important to you.

The real thing that makes music sound better is to produce it better. Compose better, play better, mix better, etc. That's what makes better sound, not ridiculous bitrates.

2. In video heavy compression is a must, because without it the raw video feed is MASSIVE. With audio you don't really need data compression. Video compression is different. It hapens spatially and temporarily. Eyes are different from ears. So these comparisons are dangerous.

 

[tansand]

"Because the sample points must be quantized to certain amount of bits, noise gets added to the signal. If that noise is quiet enough so you don't hear it the signal is perceptually 100 % perfect."

Isn't quantization noise just a fancy word for distortion though?

 

[danadam]

Isn't quantization noise just a fancy word for distortion though?

AFAIK, if quantization was done without dither, producing quantization error correlated to the input signal, this would be called a distortion. When quantization is done with dither, producing quantization error not correlated to the input signal, it is called noise.

The difference is most audible with quiet signals. Listen to the attached files:

• "02_8bit_no_dither.flac" sounds distorted when compared to the "01_16bit_orig.flac",
• "03_8bit_dither.flac" sounds the same, only with added noise, when compared to the "01_16bit_orig.flac",
• "04_8bit_dither_shaped.flac" sounds the same, with only a little noise, when compared to the "01_16bit_orig.flac".

(and why would you ever want to do quantization without dither is anyone's guess)

 

[old tech]

Here's another good video explaining bit depths and the effect on listening. The demonstration should be enough to convince the most ardent skeptics unless they just don't want to know.

The difference between 8bits and 16bits is noise, the signal is all there, nothing is missing. With dither, most people wouldn't be able to tell the difference between 8bits and 16bits for most commercial music, let alone 16bits compared with 24bits!

 

[old tech]

3. What is "normal music"? What is "maximum music"? What are "narrow" and "broad" music? 8 bits is not considered enough for high quality consumer audio. I don't consider it either. It's 5 bits less than what I consider adequate for transparent consumer audio. 8 bit is low quality digital audio. Going from 8 bits to 16 bits means going to the other side of the threshold line. Once you are on the other side, it doesn't matter how far you go.

High bias cassette tape is equivalent to around 5bits digital and LP records around 10-11bits. The truth is even 8 bits is enough for most commercial music and for the bulk of listeners. After all, pre-recorded cassettes were the best selling music format from the late 1970s to the early 1990s.

 

[tansand]

AFAIK, if quantization was done without dither, producing quantization error correlated to the input signal, this would be called a distortion. When quantization is done with dither, producing quantization error not correlated to the input signal, it is called noise.

Thank you for this, I appreciate it.

 

[bigshot]

16/44.1 is overkill. The fact that people argue it isn’t nearly enough makes me shake my head. I think it’s the “more is always better theory”.