Music files: 16bit vs 24 bit

[Gregory13]

Does anyone find a big difference between 16 bit files and 24 bit files? I have a lot of music files in 24 bit and have done the comparison myself but wanted everyone one else's opinion.

[theoldnic]

Would say the 34 bit (3 8 bit bytes) would be the better of the two for clarity. More information storage, makes no difference whether a 32 or 64 bit file system or any other 'NIX system.

[davewantsmoore]

On 08/08/2019 at 5:30 PM, Gregory13 said:

Does anyone find a big difference between 16 bit files and 24 bit files? I have a lot of music files in 24 bit and have done the comparison myself but wanted everyone one else's opinion.

The difference is not due to 16bit vs 24bit .... it is due to the audio on the 24bit release is either (re)mastered differently to the 16bit (CD) release .... or it is the same audio, but has been through less resampling .... or some other "human error".

 

 

Otherwise..... if you take the same audio .... and you encode it with 16bit .... and separately encode it with 24bit .... all else being equal ....... they will not be audibly different.

 

3 hours ago, theoldnic said:

would be the better of the two for clarity. More information storage

No.

 

A 16bit audio format is capable of storing (approximately) 96dB below the peak loudness.

A 24bit audio format is capable of storing (approximately) 144dB below the peak loudness.

 

Where the two overlap .... ie for peak loudness all the way down to 96dB.... for the 16bit and the 24bit formats, they store the information identically.

 

 

In all sane practical situations, you cannot hear 96dB below the peak loudness.    For example.   If you set the peak output to be 120dB  (this will really hurt your ears crazy loud) .....  96dB below that is 24dB.      The background ambient (silent room) noise in my listening room is 30dB.     So 24dB is inaudible.... and that's not even before you begin to account for loud sounds masking the VERY quiet sound in question here.

 

 

You will see this sort of thing   (ie. people saying => more information storage, means 16bit stores a lower quality representation than 24bit)   written all over the internet.    Often by people who really should know better.    It is complete and utter garbage.....  and shows a dramatic lack of understanding of how digital audio formats work.

 

 

Alot of the confusion is perpetuated  (in addition to "bigger numbers must be just-better syndrome")  by people comparing 16 and 24bit versions of audio .... where either the audio in each isn't identical ..... or the playback equipment has quirks.

[a.dent]

On 08/08/2019 at 3:30 PM, Gregory13 said:

 

Does anyone find a big difference between 16 bit files and 24 bit files? I have a lot of music files in 24 bit and have done the comparison myself but wanted everyone one else's opinion.

 

No.

I agree completely with @davewantsmoore post above.

[davewantsmoore]

On 08/08/2019 at 5:30 PM, Gregory13 said:

Does anyone find a big difference between 16 bit files and 24 bit files?

I think it's important to restate......  that if you DO hear a difference.    There are many possible reasons.

 

BUT... those reasons do not involve the generalisation that  "24bit contains more information than 16, so it uses that data to represent the music better".   That logic seems very very intuitive.... but it is wrong.
 

What lots of people do is they say:

"I heard a difference.... and 24 sounded better ..... therefore the logic is correct that 24bit formats have a higher quality representation of the audio.... and anybody who explains otherwise is wrong, because I heard it with my own ears...  There must be some sort of magic going on that the engineers don't understand"

[Irek]

I agree with that statement:

"If you take the same audio .... and you encode it with 16bit .... and separately encode it with 24bit .... all else being equal ....... they will not be audibly different." 

[Guest]

Unless your music has more than 96dB of dynamic range, which it very likely doesn't, then there is no difference.

 

24 bit most definitely has its uses in the studio for 'overhead' during mastering, but in 'real life' we only need 16 for good reproduction.

 

Of course as we all know, it does come down to the quality of the master.

[Gregory13]

Unless your music has more than 96dB of dynamic range, which it very likely doesn't, then there is no difference.
 
24 bit most definitely has its uses in the studio for 'overhead' during mastering, but in 'real life' we only need 16 for good reproduction.
 
Of course as we all know, it does come down to the quality of the master.

Yes, I think that is true but when doing comparisons between a 16 bit file and 24 bit file (of the same song) I do notice a 5% or so difference in clarity. It sounds slightly clearer, especially with vocals. Have you ever noticed any difference at all?

[LogicprObe]

6 hours ago, TDK said:

Unless your music has more than 96dB of dynamic range, which it very likely doesn't, then there is no difference.

 

24 bit most definitely has its uses in the studio for 'overhead' during mastering, but in 'real life' we only need 16 for good reproduction.

 

Of course as we all know, it does come down to the quality of the master.

 

A lot of the processing done in studios is 32 bit.

[a.dent]

7 hours ago, Gregory13 said:

10 hours ago, TDK said:

Unless your music has more than 96dB of dynamic range, which it very likely doesn't, then there is no difference.
 
24 bit most definitely has its uses in the studio for 'overhead' during mastering, but in 'real life' we only need 16 for good reproduction.
 
Of course as we all know, it does come down to the quality of the master.

Yes, I think that is true but when doing comparisons between a 16 bit file and 24 bit file (of the same song) I do notice a 5% or so difference in clarity. It sounds slightly clearer, especially with vocals. Have you ever noticed any difference at all?

If you notice a 5% difference, which is a significant amount, then I would suggest you are listening to different masters.

Mid-range “clarity” is not something that will change with bit depth change. 

[buddyev]

If, as seems to be the consensus in this thread, there is no practical difference/improvement in 24 bit over 16 bit is there any point paying the premium for 24 bit files?

Edited August 11, 2019 by buddyev

[georgehifi]

On 08/08/2019 at 5:30 PM, Gregory13 said:

Does anyone find a big difference between 16 bit files and 24 bit files? I have a lot of music files in 24 bit and have done the comparison myself but wanted everyone one else's opinion.

I hear a big difference for the better between PCM 24/96 and 16/44 using https://referencerecordings.com/ CD's.

But this is with a dac that is discrete R2R multibit 24/96, I listened to the same with a YBA dac that is DS and the difference wasn't as obvious.

 

Cheers George

[a.dent]

1 hour ago, buddyev said:

If, as seems to be the consensus in this thread, there is no practical difference/improvement in 24 bit over 16 bit is there any point paying the premium for 24 bit files?

My feeling is it all depends.

The same master at 16 v 24 bit will sound the same.

Often a 24 bit release will be be a new remaster and consequently sound different.

[Guest]

1 hour ago, buddyev said:

If, as seems to be the consensus in this thread, there is no practical difference/improvement in 24 bit over 16 bit is there any point paying the premium for 24 bit files?

 

Unless they started off with a better master (which they often do), there isn't.

Of course this will be debated forever, as is everything in the audiophile game, but the maths and the science doesn't lie. 

 

24 bit and 32 bit in the studio makes sense for technical reasons related to the mastering process itself, but the books were closed on the playback process quite some time ago.

 

24 bit / 384 Khz sells DACs and FLACs because in most peoples minds more is better.

 

 

[theoldnic]

On 10/08/2019 at 8:43 PM, davewantsmoore said:

......

You will see this sort of thing   (ie. people saying => more information storage, means 16bit stores a lower quality representation than 24bit)   written all over the internet.    Often by people who really should know better.    It is complete and utter garbage.....  and shows a dramatic lack of understanding of how digital audio formats work.

 

 

Alot of the confusion is perpetuated  (in addition to "bigger numbers must be just-better syndrome")  by people comparing 16 and 24bit versions of audio .... where either the audio in each isn't identical ..... or the playback equipment has quirks.

 

Then given that, 24 bit is a waste of filing space ! ( Which came first 24 bit or 16 bit audio ?)

As for clarity, one person was saying that the encoding "system makes the difference in them", which is 90% right , because alike "JPEG quantization tables" (image compression codec) ,  "THE specific audio codec"  is largely the accuracy of the sound reproduction and much of the point of a codec IS the efficiency of its information accuracy in reconstruction coupled with compression size (metaphorically akin to engine power to weight ratio of aircraft along with lift area efficiency).

 

 

 

[davewantsmoore]

 

Quote

Then given that, 24 bit is a waste of filing space

Yes, for the most part .....the are (perhaps you could say, obscure) reasons why you might ....   but in general 16bit is sufficient.

 

Quote

JPEG

 

Is exactly the wrong metaphor.

 

16 bit can capture a certain "space".

24bit can capture a larger "space"

 

Where these spaces overlap, they capture the audio IDENTICALLY.

 

You only need the 16bit space..... the extra audio (they are quieter sounds) in the 24 bit space is inaudible in all sane playback use-cases.

 

Edited August 12, 2019 by davewantsmoore

[davewantsmoore]

2 hours ago, TDK said:

24 bit / 384 Khz sells DACs and FLACs because in most peoples minds more is better.

Because if you had a 24bit audio file .... and you wanted to play it on a DAC which could not understand 24bit .....  then you may be leaving it up to some hamfisted (low quality) process to convert the audio from 24 to 16 bit.

 

 

It is (usually) better to have hardware that can natively understand the audio format ..... even if that audio format is not necessary.

[davewantsmoore]

2 hours ago, a.dent said:

My feeling is it all depends.

The same master at 16 v 24 bit will sound the same.

Often a 24 bit release will be be a new remaster and consequently sound different.

 

100%

[Guest]

25 minutes ago, davewantsmoore said:

Because if you had a 24bit audio file .... and you wanted to play it on a DAC which could not understand 24bit .....  then you may be leaving it up to some hamfisted (low quality) process to convert the audio from 24 to 16 bit.

 

 

It is (usually) better to have hardware that can natively understand the audio format ..... even if that audio format is not necessary.

 

Right, so next comes 48 bit audio and a whole round of new DAC's and Downloads.

 

Meanwhile, us with the equivalent of 14 bit audio (vinyl) quite often still have superior sound and most definitely do not suffer from bit depth envy. 

 

Don't misunderstand me though, if the music in question is indeed recorded with more than 16 bits of dynamic range, then go for it. Not much, if any is though - and even it if were, your ears could not cope with 120dB differences in sound range anyway.

 

This is why 16/44 is all that's needed for good quality reproduction. There's something about bigger numbers that affects the psyche and makes us feel we are hearing something we're not.

 

Now in respects to sampling rates (as opposed to bit depth), there are those out there who will argue that high frequency reproduction 'feeds' into lower frequencies... They are of course, wrong. This 'feeding' is called 'harmonic distortion' and was never in the original recording anyway.

 

Of course if you like the sound of harmonic distortion, all the more power to you (an I do mean that, genuinely, as each persons preferences are different) but I find it funny that the purists who call for 24 bit/96Khz audio and badmouth vinyl are the same lot who champion for distortion to make their music sound better. I wonder how many use valve amps

 

Edited August 12, 2019 by Guest

[davewantsmoore]

5 minutes ago, TDK said:

Right, so next comes 48 bit audio and a whole round of new DAC's and Downloads.

No.

 

24bit audio takes the format to the edge of what analogue electronic is (ever going to be) capable of..... ie. down to things like the johnson noise of a resistor.

 

32bit audio exists because sometimes it's "just easier" to do it that way inside a computer.

 

5 minutes ago, TDK said:

Meanwhile, us with the equivalent of 14 bit audio (vinyl) quite often still have superior sound and most definitely do not suffer from bit depth envy. 

Not a lot of bits are required to represent most audio .... if they are used sensibly.

 

For example, people are all up in arms that MQA is only using X bits to encode the audio....  there is no reason for alarm.

 

 

5 minutes ago, TDK said:

Don't misunderstand me though, if the music in question is indeed recorded with more than 16 bits of dynamic range, then go for it. Not much, if any is though - and even it if were, your ears could not cope with 120dB differences in sound range anyway.

Yes.... not even 96dB of difference.

 

5 minutes ago, TDK said:

This is why 16/44 is all that's needed for good quality reproduction.

Yes, in theory.

 

Higher sampling rates can be very handy in practise ......  but not because we can hear higher frequencies than 20khz.

 

[theoldnic]

Just now, davewantsmoore said:

 

Yes, for the most part .....the are (perhaps you could say, obscure) reasons why you might ....   but in general 16bit is sufficient.

 

 

Is exactly the wrong metaphor.

 

16 bit can capture a certain "space".

24bit can capture a larger "space"

 

Where these spaces overlap, they capture the audio IDENTICALLY.

 

You only need the 16bit space..... the extra audio (they are quieter sounds) in the 24 bit space is inaudible in all sane playback use-cases.

 

 

With images, the compression process is the maths, after compression by a coefficient (codec) the final number p/segment (8 , 16,32,64 bit chunk) as choice is only a file until re-expanded, so the actual processor systems you are saying in the Multi-media section of the CPU core does not (or in effect have never  been made to) "operate" on music data to produce sound with any more than 16 bits (2x  8 bit bytes or 4x 8bit bytes 24 bits and split).

In short the CPU architecture circuitry cannot (shuffle through an expanded data set for multi media "sound" reproduction/generation) use any more than 24 bits ?!?

Or more succinctly , are there any other multi media sound architectures in CPU use re-expanded sound data , or are all sound data core sections 16 bit maximum info??

Edited August 13, 2019 by theoldnic

[Ittaku]

On 12/08/2019 at 2:11 PM, TDK said:

 

Of course if you like the sound of harmonic distortion, all the more power to you (an I do mean that, genuinely, as each persons preferences are different) but I find it funny that the purists who call for 24 bit/96Khz audio and badmouth vinyl are the same lot who champion for distortion to make their music sound better. I wonder how many use valve amps

 

I use high res audio almost exclusively. I can't hear the difference between 24 and 16 properly dithered bits, but I can hear downsampled audio to 44/48. Either higher res, or well upsampled regular res routinely sounds better to me. And just to put a spanner in your assertion, I never badmouth vinyl, and have valve pre and power amps.

[Ittaku]

As for 24 bits, the important part is having 24 bits at playback time in order to tolerate DSP and volume trimming without audible deterioration. This doesn't mean you need to start with 24 bits, though.

[theoldnic]

In the following article the "sample rate" of 24 bit audio rates around the value of an AM broadcast radio carrier !

https://www.tested.com/tech/1905-the-real-differences-between-16-bit-and-24-bit-audio/

 

The main point to remember is the "samples" from Audio are obviously in the human hearing range for  listening of KHz , whereas the CPU can "sample"  with a BUS (binary unit speed) alike an FM radio station in an old Pentium 3 8086 type CPU of 100Mhz or more BUS speed. (The CPU speed of around 450 to 500 MHz is the beat clarifier to ho;d all the "data and instruction BUS frequencies" harmonically in sync)

 

Actual "files stored on disk" as data should not be ambiguated to the information after the file is opened and re-expanded from its storage compression, as that point is a "codec" is simply a compression calculation for storage, when a file is re-expanded they all play the same by the multi-media section of the CPU as "data" as is found on a CD.

Edited August 14, 2019 by theoldnic

[MLXXX]

 

20 hours ago, Ittaku said:

I can't hear the difference between 24 and 16 properly dithered bits, but I can hear downsampled audio to 44/48.

 

Did you take the precaution of resampling the downsampled audio back up to the same SR as the original file, in order to present the DAC with the same sample rate (so it would use the same reconstruction filtering parameters for its analogue output)?

 

I find Audacity can do downsampling and resampling fine. And foobar2000 with ABX plug-in can then be used to compare the original and twice resampled file.

 

If you can ABX a difference I'd be interested in whether you find that only occurring with a downsampling to 44.1kHz rather than 48.0kHz. I'd also be interested in how "profound" or "minor" the differences are for your ears.

 

(As for myself, I can no longer do meaningful tests even at 44.1kHz, as the upper limit of my hearing peters out at a little above 10kHz.)

Edited August 14, 2019 by MLXXX