Converting digital audio files

Hello,

Flac is supposed to be the digital audio format with the highest quality. When you convert an mp3 file (or any other format) to flac, can you really get the best possible audio quality? Isn’t it similar to “converting” a Mazda to a Ferrari?

A similar question could be: “Can you get a better quality mp3 file by converting a 128 mbps file to a 320 mbps file?”

Thanks,

spee

I don’t think you can improve the quality of a file once it has been put into MP3 format. I know that you really lose quality if you try to do anything with MP3 files in Audacity. That is, you can start with WAV file and compress it to MP3, but you can’t go the other way.

MP3 is a compression algorithm, so once it has been compressed, you can’t re-expand it to its original state, much less improve it.

Absolutely not. If anything, it’s going to get worse.

I’d say it’s more like recording your wedding on a camera phone, then “converting” it to cinema quality by playing it back on a TV which you’ve pointed a 35mm movie camera at.

They mean that Flac is better audio quality if you take an uncompressed file and convert it to Flac, than if you take an uncompressed file and convert it to MP3. Both Flac and MP3 are lossy compression formats, so what gets lost in the raw-to-MP3 conversion can’t be recovered. Taking that compromised audio and then losing some more info by converting that to Flac will necessarily result in a worse product.

According to http://flac.sourceforge.net/, the term “FLAC” stands for “Free Lossless Audio Codec”, so it does not appear to be a lossy compression format.

You have lossless formats

FLAC
APE
WavPack
WMA - Lossless

These are the “big” four. They all are lossless which means you can freely convert between them as much as you want and you will never lose quality.

Lossy formats mean that part of the quality is sacrificed for the sake of file size.

While lossless is about 40% to 60% of the uncompressed file, lossy formats can be as little at 10% to 15%.

Lossy formats include

mp3
wma - lossy
mp4 (a4a) [they are the different file names for the same thing]
Once you convert a format from lossless to lossy you lose quality. You can never regain the quality by upconverting it. Once it’s gone it’s gone forever.

Now as I said, lossless is lossless. You can go freely between WavPack, WMA Lossless or FLAC, a hundred times and never lose quality

But there are different types of lossy formats

We’ll use mp3 as an example.

mp3 has three ways to compress. Average, Variable or Constant. It also allows you to choose the bit rate. The lowest typical bitrate you can have and have reasonable sound quality in mp3 is 128. The highest is 320. (You can go above 320 but theirs no added benefit and the file size increases so much you may as well use a lossless format)

If you compress a music track to 320cbr (constant bit rate) this means all bits of that song will take up 320bitrate whether they need it or not.

This isn’t the most efficient eay of doing things. Suppose you have a song and it contains a 5 second silence. Well in 320cbr the 5 seconds of silence is going to take up as much space as music. There’s no need to allot that space, it’s a waste.

Now if you use 320vbr (variable bitrate) that means the compression will use UP TO 320 IF needed. If it’s not needed it won’t use it. So if you set compression to 320vbr you may wind up with 250vbr when the track is finished, because the track didn’t need to use high bit rates for whatever reason.

The debates is whether you can go from higher bit rate to lower bit rate. In otherwords can you go from a 320cbr to 192cbr?

The answer is you probably could do it an not notice anything but in reality you would lose quality because the algorithm to compress music is based on lossless not lossy. So the algorithm would be messed up. In otherwords the alogrithm would be using data it is assuming is lossless and using that to calculate the lossy result. As you can see that would be wrong.

So in reality you would lose quality but whether you’d notice it or not depends, not only on the person, but the type of music. For instance, classical music is much more susceptible to compression issues.

[note: notice windows media has two formats

wma-lossless which is true lossless
wma-lossy which is a compressed format like mp3

BOTH use the file extension wma and you cannot tell if it’s lossless or lossy by the file extension. You’d have to look at the file size and guess.]

I’m skeptical that going from 320 to 192 is going to be any worse than going to 192 directly.

To the OP, what you’d get in your case is basically 128 kbps quality, but 320 kbps file size.

Also ALAC, which is Apple’s lossless format.

Side question: I feel like I’ve asked this before and didn’t get a satisfactory answer but:

Say I rip a CD at 128kbps, then burn those mp3s to a Redbook CD. If I rip it again at a higher bitrate, will it still be in <=128kbps quality while the file says it’s 320kbps or whatever? Will ripping programs inform you of the format? Will the file be a larger size if it goes 128->320 instead of 128->128?

Yes - once you have rewritten the CD, the file is a CD bitstream with no information about compression/decompression. So the CD data (which is 128kbps quality audio information stored in two-channel 16-bit PCM encoding at a 44.1 kHz sampling rate per channel) will be mp3 compressed to your 320kbps data rate, introducing some additional compression artifacts (but not many - MP3 algorithms filter specific frequency ranges from the source, 320kbps just filters less - the 128kbps data will already have filtered those frequencies out).

The ripping program will not know - the raw data from the CD will be two-channel 16-bit PCM encoding at a 44.1 kHz sampling rate per channel - there is no metadata to tell anyone that it once was a 128kbs mp3.

A 320kbs mp3 will always be bigger than a 128kbs mp3, no matter how it was generated.

Si

If you went with a VBR encoding, you’d probably get a smaller end file from the CD->128-CD->VBR file, than from CD->VBR directly, since the compression to 128kbps step is removing information. That would be a pretty easy test to do.