16-bit limit from CD, 24 bit needs better source?

Audacity 3 Help Forum Windows
1

Since my digital music player has usually been a smartphone, and the source of my music was mainly CD (with some high quality tapes from vinyl still in there, thanks to Audacity) I’ve only just recently moved up to a true DAP which can do justice to 24-bit lossless recordings.

And here’s where I need a reality check. If I’m understanding it correctly, the CD medium uses 16-bit native files, so no choice of FLAC or other format is going to make it sound as good (if my ears can hear the difference) as a 24-bit “DVD” quality file, either downloaded from a 24-bit source, or recorded from one of the “better than a CD” disc sources.

Yes? No? Tempest in a teapot?

Would upsampling in Audacity just change the distortion, without improving anything that could be heard on a DAP then?

2

Standard audio CDs don’t use “files”, they just have a long stream of data, plus a “table of contents” that tells the player where to start and stop for each “track”.
The data is 16-bit PCM, as is a 16-bit WAV file (so CD data is very much like a 16-bit WAV file).
The sample rate of a standard audio CD is 44100 Hz.

The dynamic range of 16-bit PCM (dithered) is around 100 dB, which, if you have a big enough and good enough sound system, can go from inaudible, to jack hammer loudness (but not recommended that you turn it up that loud as you will quickly go deaf).

The frequency range for 44100 Hz sample rate (again, assuming that the sound system is good enough), goes up to around 20 kHz (which is considerably higher than most adults can hear). The lowest frequency is entirely dependent on the sound system.

Super high definition formats (example, 24-bit 192 kHz) can theoretically reproduce higher frequencies, and greater dynamic range than the standard audio CD format, so theoretically they can produce more sound that is beyond the range of human hearing, and dynamics that can make you deaf even faster, but in the real world there really no point.

There is an excellent article on the subject, by one of the lead developers at Xiph.org (the inventors of FLAC, OGG and Opus) here: 24/192 Music Downloads are Very Silly Indeed

3

“better than a CD”

Better than Audio CD. Compact Disks can also be shiny, flat hard drives (Data CD) and you can put any files on there you want.

a 24-bit “DVD” quality

I wonder about that. Last I checked, entertainment DVDs use 48000, 16-bit PCM and Dolby Surround, a compressed format. You can also have a Data DVD, a shiny flat hard drive.

Koz

4

Koz-
In all cases I’m referring to the “Audio” products, not to just media with data files on them.
Just to make that clear.

(Still making my way through the link that Steve provided.)

5

Hope you are enjoying it :slight_smile:

The author of that article also has some very good videos explaining the rudiments of digital audio. If you’re interested, you can find them here:
https://xiph.org/video/

6

Would upsampling in Audacity just change the distortion, without improving anything that could be heard on a DAP then?

Upsampling from 16 to 24 bits does nothing, except make a bigger file with an extra byte (8-bits) full of zeros.* It’s mathematically lossless and completely reversible.

Increasing the sample rate is “mathematically imperfect” but it’s audibly perfect. Of course, you shouldn’t do it unless you have a good reason. For example, if you are converting from CD to DVD, you’ll have to upsample to 48kHz.

Intuitively, it seems “better” to convert between 44.1 and 88.2kHz than between 44.1 and 96kHz, but in practice there is filtering which involves rounding, the DSP is the same, and there’s no advantage to “even numbers”.

And here’s where I need a reality check. If I’m understanding it correctly, the CD medium uses 16-bit native files, so no choice of FLAC or other format is going to make it sound as good (if my ears can hear the difference) as a 24-bit “DVD” quality file, either downloaded from a 24-bit source, or recorded from one of the “better than a CD” disc sources.

The guys who do scientific-blind [u]ABX Tests[/u] have pretty-much demonstrated that “CD quality” is better than human hearing. For example, if you take a high-resolution original and downsample to 16/44.1, you won’t hear a difference.** In fact, a good quality MP3 often sounds identical to the original (in a proper-blind listening test***).

Sometimes when you buy a high-resolution recording you’re getting a different master or different mix, or even a different recording, so it can sometimes sound better or “just different”.

as a 24-bit “DVD” quality file,

DVDs are either 16-bit 48kHz uncompressed stereo (or mono) or Dolby Digital (AC3) which is lossy compression 1-6 channel). Dolby Digital isn’t an integer format so it doesn’t have a bit-depth. Sometimes there is a choice between lossless (LPCM) stereo and Dolby 5.1. I’ll take the lossy surround sound every time! Some of the best sounding music I own is on 5.1 channel concert DVDs.

Technically, DVD supports uncompressed 24-bit audio, but there are some strange rules that say they can’t use it with the CSS copy protection. There is an oddball format called DVD-Audio that supports high resolution lossless audio but the discs are rare and most DVD players won’t play it. Blu-Ray also supports high-resolution lossless.

Since my digital music player has usually been a smartphone, and the source of my music was mainly CD (with some high quality tapes from vinyl still in there, thanks to Audacity)

Of course, analog vinyl & tape is nowhere near CD quality. The “resolution” is limited by the noise floor. There are usually frequency-response variations, and sometimes with records there is distortion.

I’ve only just recently moved up to a true DAP which can do justice to 24-bit lossless recordings.

Most phones are quite good (flat frequency response, no audible distortion, and no audible noise). But, sometimes headphone impedance can “interact” with the phone’s output impedance giving frequency response variations (this doesn’t happen if you plug your phone into your stereo system or powered speakers) and sometimes a DAP can go louder without distortion than the phone.



\

  • In Audacity, you have to make sure [u]dither[/u] is turned-off if you want the audio “untouched”.

**The ABX software is available so you can scientifically compare different file formats yourself if you wan to. An ABX test comparing your phone to your DAP isn’t as simple. The devices have to level-matched, someone else has to the switching so you can be “blind”, and the switching has to be done without any switching-noise that might give-away which device is being used.

*** It also turns-out that you don’t need a super high-end stereo to hear compression artifacts. If you hear compression artifacts it’s more-related to the program material (some sounds are easier to compress than others) and your ability to hear compression artifacts.

7

Doug-
"Of course, analog vinyl & tape is nowhere near CD quality. The “resolution” is limited by the noise floor. There are usually frequency-response variations, and sometimes with records there is distortion. "
No no no. That’s only true for mass-market stuff.
I had a Nakamichi 582. When used with metal tape, and a “direct master” analog pressing (they were pricey, not the department store stuff) or if I used a CD as the source for that, the tapes made by the Nak582 could not be distinguished from the originals. High end cassette systems, and there were many of them, could do a 20-20k tone range just as well as most open reel decks, and better than some.
Yes, there can be distortions. When mine was new, everything sounded off, just a little too slow. And the tech at Nakamichi ran some tests and then looked up at me, because I had said “It sounds like the difference between a European middle C and an American middle C” (440 Hz versus 444 Hz) and said “You know, it IS almost 1% off”. The deck was quickly recalibrated.
It couldn’t improve the source, but it could readily match it, when maintained properly.

What bugs me the most about phones, is that they all have a 16-click volume control. Would anyone buy a TV or hifi that only had 16 volume levels? HellNo. This one is too loud, that one is too soft, and there’s nothing in between. So, I’ve been looking at DAPs. Which can also generally carry a larger library, and not run down my phone over the weekend. And handle more than MP3.

On different masters…I’m very aware of that. Led Zep keeps remixing rereleases, damned if I know what’s up on which. CSNY remixed from lp to cd, and there are whole new instruments visible. Which is not necessarily a good thing. And then, god bless their little money-grubbing souls, the “record” industry has been splitting tracks and entering metadata in a manner that only trained baboons could be doing. My old LP’s, manually track cut during conversion, are cut in the right places. Lots of new CDs? Simply cut wrong. Group names spelled wrong, amateur night from the production groups these days.

8

but the dynamic range, even of a well maintained Nakamichi, is substantially less than 16-bit PCM (around 30 dB less, which works out as about 30 times less than what 16-bit PCM is capable of :open_mouth: ) That is not knocking Nakamichi, it is saying just how good the CD format is (though the dB range that is actually used in commercial recordings is typically far less than the available dynamic range of CD or a good Nakamichi).

The main thing that makes Nakamichi sound so darned good, is that the analogue electronics and mechanical build quality is so good. Sadly I never owned a Nakamichi, but I did own a rather nice Sony reel to reel (in fact, two of them!). In one of my past lives I was an electronics engineer, and I took the opportunity of testing the pre-amps in the Sony (all discrete components), and found (to the limits of the test equipment) that it was dead flat from 10 Hz to 25 kHz!

9

a good quality MP3 often sounds identical to the original

Perfectly correct as an end-user.

Perfect Performance > Compressed Format > Listening Enjoyment.

We should not take this as unlimited license to convert between formats. You still can’t make an MP3 from an MP3 without increased sound damage. The compressed formats work by carefully hiding the effect. Anything that disrupts the magic is going to cause problems.

Koz

10

You have posted this question on a technically biased forum and as a result have received a storm of technical responses and I am not going to add to them or offer criticism. But I strongly advise you to trust your own ears and compare the formats for yourself on the best audio equipment you can afford within reason. Your own musical enjoyment is all that matters.

11

Technically biased forum…Yes, by longstanding choice. The psychoacoustics and other “gray matter” weren’t my concern here.(G)

Steve-
My Nak has been a most embarrassing paperweight in a box for a long time now. After a long period of not being used, it developed the famous “won’t play, only rewinds” syndrome. Sent it out for repair, two years later it did that again. And given the infuriating fact that it must be “something simple stupid!” like a belt, which I can’t seem to access despite having the repair manual somewhere, I’ve left it to sulk until and unless I find a Nak genius who doesn’t live on caviar.
But whatever the numbers are for dynamic range, I used to A/B that with a decent set of headphones or a good stereo (Sansui AU-919 driving Yamahopper NS-500’s) and I have to say, whether it was the Telearc 1812 (where the canon fire can and did shake my kitchen dishes) or the Berlin Philharmonic doing Grand Canyon Suite (and there’s a very clean and precise triangle “ding” after the storms) the metal tapes were always able to totally duplicate the source, analog or CD, to the best of my ears. Which used to hear ultrasonic gizmos, no problem with highs or lows.
If human ears are capable of hearing larger ranges, the human perception system is capable of pretending it can deal with that.

Which kinda comes back to what Koz was saying.

I’ve been mainly listening to VBR-2 MP3’s. When I was getting everything digital, I went for three libraries figuring I’d need backup anyway. FLAC, WAV, and VBR-2 because as much as I strained, I couldn’t be sure if I could hear anything wrong with VBR-4. So I figured, if I can’t be sure that -4 is any real loss, I can settle for VBR-2 as a tradeoff against lossless file size. Which was a bigger issue ten years ago. (No really, who had a terabyte postage stamp? Let alone hard drive?)

Gotta say, to my aging ears (I’ve got some loss over 13,000 Hz) the VBR-2 MP3s still sound damn good in the real world, with real background noise and other issues. I’m wondering if it makes sense to use the FLAC, because in reality…you know, those MP3’s may be as golden, but if I’m condemned to the 16-bit source material, the FLAC files don’t seem to really be bringing a lot into the game.

My SONY cans may be a POS (high end consumer is still not high end) and my Sure 215 earbuds are the low end of their line. The Eponymous we won’t talk about either…except to say they beat all hell out of $10 smartphone buds or the old Phillips CarryCorder.(G) They serve a purpose. But I can’t help wondering…Maybe it is time to say the VBR-2 MP3’s are better than what’s left of my ears?

12

It absolutely makes sense to keep a backup copy in a lossless format (such as WAV, FLAC or AIFF), in case you ever need to edit or process the files in the future. The FLAC format has the advantages that it has good, fully specified support for metadata, and FLAC files are about half the size of equivalent WAV or AIFF files.

Lossy compressed formats (such as MP3, AC3 or Ogg) can have excellent sound quality, but the encoding process always discards some of the information. Anything other than very basic editing requires that compressed formats are uncompressed first, so editing an MP3 file in Audacity and exporting as MP3 loses information twice - once when the MP3 was first encoded, and again when it is exported from Audacity. This is a bit like making a photocopy of a photocopy; the first photocopy may look identical to the original, but after a few copies of copies, the quality progressively deteriorates.

If you have a backup / archive copy in a lossless format, then if you ever need to edit or processes the recording, you can use that copy each time (don’t overwrite the archive copy), so there will only ever be encoding losses once.

FLAC was originally developed as a format for archiving audio recordings, and is well suited to the task.

13

Agreed on all points. What I meant abusing “use the FLAC” was to use them in a player, as opposed to listening to the MP3’s, because the sound of either is so close. (VBR-2/vs/FLAC/vs ambient noises). I had originally made both FLAC and WAV as dual archives, since FLAC support used to be kinda scarce, and I figured wtf, I’ll keep the dual archives in two formats just in case, down the line.

Now I’m wondering if I should just make that two FLAC archives. I have noticed that even when one CD is ripped to multiple formats (using dbPoweramp typically) that the metadata it tags the audio files with differs. Sometimes, even the simple things like artist title, album title, vary from one to the other, indicating something is being handled differently with the metadata.

I’ve seen “Grateful Dead, The” and “The Grateful Dead” come off the same CD, just depending on which doorway it walked through. Never had that problem with Audacity!

14

Audio CDs do not usually contain metadata, as it is not part of the orange book specification. Information about the contents of audio CDs is usually drawn from the Internet, based on a unique ID for the disk. (https://en.wikipedia.org/wiki/CDDB)

It is possible for audio CDs to contain text information, but it is a departure from the original standards. In the case of CD-Rs, it is in the form of “CD-Text” https://en.wikipedia.org/wiki/CD-Text

15

It also depends on the size of the room. Speakers can’t generate a wave longer than the dimensions of the room.