PDA

View Full Version : How to know if 24 bits/96 khz true or not



vince33
2010-10-16, 09:06
Hi,

I am new on this forum and have a newbie question. I wish to know how can I know if a file(wav or flac) with a given resolution(24 bits/96 khz for instance) is a true 24 bits/96 khz file and not an upsampled file from a 16 bits/44.1 khz file by adding zeros to missing bits.

Thanks

vince33
2010-10-17, 12:47
no reply is a reply for me. I will keep on playing my old CD's with my old CD player.

TerryS
2010-10-18, 10:32
Hi,

I am new on this forum and have a newbie question. I wish to know how can I know if a file(wav or flac) with a given resolution(24 bits/96 khz for instance) is a true 24 bits/96 khz file and not an upsampled file from a 16 bits/44.1 khz file by adding zeros to missing bits.

Thanks

The only way I know to be sure would be to look at the samples with an editor that allows you to zoom in vertically enough to see the actual quantization levels. I have done this with Goldwave. I think Adobe Audition can also do this, but Audacity can't (at least the version I tried, which was about a year ago).

I doubt that any software could accurately report the number. It wouldn't be able to tell if it was actually 24 bits, or 16 bits padded to 24 bits like Audacity does when it records on WIN machines.

Terry

Phil Leigh
2010-10-18, 11:56
Hi,

I am new on this forum and have a newbie question. I wish to know how can I know if a file(wav or flac) with a given resolution(24 bits/96 khz for instance) is a true 24 bits/96 khz file and not an upsampled file from a 16 bits/44.1 khz file by adding zeros to missing bits.

Thanks

1) if the file contains frequencies greater than 22kHz, it is not upsampled.
2) "adding zeroes" is only about converting 16 to 24 bit, nothing to do with sample rates.

vince33
2010-10-20, 13:27
1) if the file contains frequencies greater than 22kHz, it is not upsampled.
2) "adding zeroes" is only about converting 16 to 24 bit, nothing to do with sample rates.

I think in the 1) you mean "less" than 22khz instead of "greater" than...
for the 2) you're right I made a confusion.

To resume it's not possible to tell that a file claimed to be a HD file is true or fake. That's I wanted to know.

Thanks to both of you for replying.

Vince

aubuti
2010-10-20, 14:10
I think in the 1) you mean "less" than 22khz instead of "greater" than...
Well, I'm not an expert like Phil, but I'm pretty sure he did indeed mean greater than. Look at it this way: every file has frequencies less than 22kHz. At least, every file that humans can hear. So "having frequencies less than 22kHz" would be a lousy criterion to distinguish one track from another.

If the original source material was 44.1, then it won't have frequencies greater than 22kHz, and upsampling won't magically add sound at those frequencies.

TerryS
2010-10-20, 14:40
I think in the 1) you mean "less" than 22khz instead of "greater" than...
for the 2) you're right I made a confusion.

To resume it's not possible to tell that a file claimed to be a HD file is true or fake. That's I wanted to know.

Thanks to both of you for replying.

Vince

I wouldn't say it is not possible. I can easily tell using a software package like Goldwave that lets me look at the samples. There are other software packages that can do the same.
For instance, there is a bug in Audacity (when used on a windows platform) that makes it record in 16 bits and pad up to 24 bits even though it looks like it is set to record in 24 bits. But with Goldwave, I can easily see the data is only 16 bits with zero padding.

Terry

Phil Leigh
2010-10-21, 00:06
Well, I'm not an expert like Phil, but I'm pretty sure he did indeed mean greater than. Look at it this way: every file has frequencies less than 22kHz. At least, every file that humans can hear. So "having frequencies less than 22kHz" would be a lousy criterion to distinguish one track from another.

If the original source material was 44.1, then it won't have frequencies greater than 22kHz, and upsampling won't magically add sound at those frequencies.

Sorry - what I meant was if you look at a 24/96 file and it contains frequencies > 22k, then it cannot be an upsample of a 44.1 original, as the original could not have contained those frequencies...

vince33
2010-10-21, 04:02
Sorry - what I meant was if you look at a 24/96 file and it contains frequencies > 22k, then it cannot be an upsample of a 44.1 original, as the original could not have contained those frequencies...

@ Phil, OK understood now.

@TeryS, I will try to find Goldwave and have a try. Yhanks for your info.

lrossouw
2010-10-21, 05:22
@ Phil, OK understood now.

@TeryS, I will try to find Goldwave and have a try. Yhanks for your info.


Should be able to do it in audacity. I've done it, but you may need to convert the file to wav first, as I think there is a bug with opening these kind of flac files in audacity directly.

TerryS
2010-10-21, 08:12
Should be able to do it in audacity. I've done it, but you may need to convert the file to wav first, as I think there is a bug with opening these kind of flac files in audacity directly.

I tried to do it with Audacity and couldn't find a way to do it, but that was about a year ago, so maybe something has changed. It is easy to do with Goldwave. You just zoom in on some low level samples and see if they lie exactly on 1/2^16 values. If it is a true 24 bit file, then the samples will almost always lie between 16 bit values. If they seem to all fall nicely on the 16 bit values, you know the data has been padded.

Terry

vince33
2010-10-22, 16:43
I have Nero 10 and a wave editor is available but I never used it before. Do you know if Nero wave editor can do the job as Goldwave?

I tried with Nero and obtain this figure. I guest that the upper scale is the time line, but on the left side what is this scale (amplitude, frequency, or ?). It's in French if you need translation please let me know.

Phil Leigh
2010-10-23, 00:55
It's trivial with Audacity - but you need to work with WAV files - there are a myriad of bugs with FLAC in Audacity.

1) open file
2) select about 10 seconds of a track
3) Plot Spectrum
4) export (as a spectrum.txt)
5) look at the values in spectrum.txt

If there are frequencies higher than 22k and levels lower than -108dB (don't ask me why it's -108 in Audacity and not -96) then it's a genuine hi-res file.

It has frequencies that could not have come from a 44.1 recording and levels that are too small to capture in 16 bits.

TerryS
2010-10-24, 09:29
I have Nero 10 and a wave editor is available but I never used it before. Do you know if Nero wave editor can do the job as Goldwave?

I tried with Nero and obtain this figure. I guest that the upper scale is the time line, but on the left side what is this scale (amplitude, frequency, or ?). It's in French if you need translation please let me know.

I don't know anything about Nero, but the values along left side (the vertical scale)are almost certainly the sample values. It looks like it is probably normalized to 16 bit values, which all the programs seem to do, even when displaying 24 bit values. So the maximum number that can be displayed is 32,767 which is about half of 65,535, which is the largest number that can be represented with 16 bits. It is half because it will represent it as a signed number, so very loud samples will be at 32767 or -32767, and very small samples will be close to zero. To tell if you have 24 bit numbers, you would have to zoom in vertically (this is usually easiest to do on very quiet signals, near zero) and see if they are reported as "integer" values, like 1, 2, 3, -1, -2 ,-3 and so on or fractional numbers, like 1.23, -2.6 and such. Fractional numbers would indicate more than 16 bits. Integer numbers mean the data is 16 bits deep.

Terry

bhaagensen
2010-10-24, 11:21
Now I've become curious :)

And the curiousness is simply, why [do you want to do/check this]?

TerryS
2010-10-24, 12:34
Just because you are paranoid, doesn't mean they're not out to get you. ;)

I only know about it because I was using Audacity to do needledrops (record albums to a digital format). It turns out there is a bug in Audacity when run on Windows platforms that makes it look like you are recording at 24 bits, but the recording is actually only 16 bits, with zeros added to make it 24 bits. I wasted many hours recording and editing the files before I found out about the bug.
So it might not be a bad idea if you are undertaking a big project to make sure the process is really giving you the correct bit depth. Maybe the difference between 16 and 24 bits doesn't matter to you, but at least you should be getting what you think you are.

Terry

bhaagensen
2010-10-24, 13:04
Bummer with the bug. Certainly would have annoyed me too. Anyway, do carry on with the thread - not directly useful to me, but definitely nice-to-know-stuff :)

magiccarpetride
2010-11-03, 11:08
Hi,

I am new on this forum and have a newbie question. I wish to know how can I know if a file(wav or flac) with a given resolution(24 bits/96 khz for instance) is a true 24 bits/96 khz file and not an upsampled file from a 16 bits/44.1 khz file by adding zeros to missing bits.

Thanks

Your ears will tell you. If you can't tell by listening, then what's the point of even asking this question?

Phil Leigh
2010-11-03, 11:40
Your ears will tell you. If you can't tell by listening, then what's the point of even asking this question?

Except that it can be incredibly hard to tell between a true 24/96 master recording @24/96, a 16/44.1 downsample of that same 24/96 master recording... or even a 24/96 upsample of a 16/44.1 downsample of the same 24/96 master recording!
Fun isn't it?