Ranking audio formats in terms of sound quality

FMiguelez · Post by **FMiguelez** » Tue Jul 15, 2008 2:58 pm

.

mhschmieder · Post by **mhschmieder** » Tue Jul 15, 2008 4:08 pm

I've always used VBR, and felt really stupid until the first dissenting view arrived

.

I think I was biased by my early experiences with digital technology. In the late 80's and up until two years ago, I owned a Rotel CD player, which was quite affordable but unique in that it used bitstreaming, along the lines of what Korg is doing with their new MR-series recorders. The idea of 1-bit processing is that you should have zero artefacts because you are always maximising your packets to the material at hand, vs. wasting bits where not needed and not having enough where more are needed.

arth · Post by **arth** » Tue Jul 15, 2008 7:08 pm

bongo_x wrote:What they said.

Lossless is just that, lossless, so it should sound the same. I didn't know they used m4a as a suffix, that's weird. MP3, AAC, OGG, etc, are lossy formats, they get rid of information that you can never get back when they are encoded, hence they are smaller.

OGG is a container format, and the compression used with OGG can be either lossy or lossless.
OGG-Flac is an example of lossless OGG, while OGG-Vorbis is an example of lossy OGG.

arth · Post by **arth** » Tue Jul 15, 2008 8:01 pm

kassonica wrote:And don't forget to Rip them with LAME.

There is NO better codec out there, period.

I have seen no double blind or abx tests that shows LAME to be any better than Fraunhofer/Thomson's FhG codec or mp3sencoder.

The only test I've seen where LAME "beat" Fraunhofer was one where a new LAME was tested against a four years old FhG codec, using 128 kbps VBR, and the average bitrate was higher on the LAME samples. LAME came out slightly ahead, but with a ~6% higher bitrate and testing against a very old FhG codec, that's not surprising.

Anyhow, I much prefer FLAC as my codec -- lossless is better than MP3. Period.

bongo_x · Post by **bongo_x** » Tue Jul 15, 2008 8:18 pm

arth wrote: OGG is a container format, and the compression used with OGG can be either lossy or lossless.
OGG-Flac is an example of lossless OGG, while OGG-Vorbis is an example of lossy OGG.

Thanks, I never understood that. I thought OGG was just short for OGG-Vorbis (which is certainly not a catchy name unless you run a comic book store or host RPG's in your parents basement).

bb

arth · Post by **arth** » Tue Jul 15, 2008 9:48 pm

bongo_x wrote:
arth wrote: OGG is a container format, and the compression used with OGG can be either lossy or lossless.
OGG-Flac is an example of lossless OGG, while OGG-Vorbis is an example of lossy OGG.
Thanks, I never understood that. I thought OGG was just short for OGG-Vorbis (which is certainly not a catchy name unless you run a comic book store or host RPG's in your parents basement).

It can be confusing. Much like AIFF, which can refer to both uncompressed audio as well as compressed (which is better called AIFC, but Apple disagrees...).

Or MP4 which can contain anything including AAC, MP3 or even MIDI through SAOL, Structured Audio Orchestra Language. The idea is great - embedding both the instruments and the MIDI score in a single MP4, and even real-time mixer adjustments -- I'd buy that, if all the daws and apps supported it.

chamelion · Post by **chamelion** » Tue Jul 15, 2008 10:10 pm

Dang! This thread has become a veritable fountain of information! Thanks guys.

Cheers,

Geoff

twistedtom · Post by **twistedtom** » Tue Jul 15, 2008 11:43 pm

Yes it has some food for thought chamelion.

When I make DVD's I always do 2 pass high VBR.

Matcher · Post by **Matcher** » Wed Jul 16, 2008 3:10 am

arth wrote:
kassonica wrote:And don't forget to Rip them with LAME.

There is NO better codec out there, period.
I have seen no double blind or abx tests that shows LAME to be any better than Fraunhofer/Thomson's FhG codec or mp3sencoder.

The only test I've seen where LAME "beat" Fraunhofer was one where a new LAME was tested against a four years old FhG codec, using 128 kbps VBR, and the average bitrate was higher on the LAME samples. LAME came out slightly ahead, but with a ~6% higher bitrate and testing against a very old FhG codec, that's not surprising.

Anyhow, I much prefer FLAC as my codec -- lossless is better than MP3. Period.

The difference between codecs is more noticeable at lower bitrates, and using VBR. With higher bitrates (256 and 320kbps) the original audio file and the mp3 become very hard to distinguish from each other, and the codecs as well.

monkey man · Post by **monkey man** » Wed Jul 16, 2008 4:48 am

chamelion wrote:Dang! This thread has become a veritable fountain of information! Thanks guys.
Cheers,
Geoff

Veritable bitrates give me the jitters.

mhschmieder · Post by **mhschmieder** » Wed Jul 16, 2008 12:14 pm

So, then how would you feel about venerable bit rates?

bkshepard · Post by **bkshepard** » Wed Jul 16, 2008 12:33 pm

mhschmieder wrote:So, then how would you feel about venerable bit rates?

Sorry, for a moment there, I thought you said venereal bit rates.

Shooshie · Post by **Shooshie** » Wed Jul 16, 2008 12:34 pm

Venerable bit rates? Are they contagious? Can you get them by listening to 2nd-hand background music?

My vote is for LAME, for a number of reasons. One of those reasons is that it is very, very good. Almost unimpeachable. Another reason is that it's easy to find now. When I first started researching how to encode the best MP3's for web playback, I eventually found LAME. It was hard to figure out back then, since LAME was basically the codec to be installed down in the bowels of UNIX. You still had to learn how to use it. The choices were, talk to it through Terminal, or find a front-end that does that for you. The early front-ends contained a series of fields into which you typed various kinds of information, which it transmitted via UNIX code to the LAME codec. You had to learn the right combinations to get what you wanted, and that required trying all the variables, learning what each extreme did in each variable, then working your way down to that "perfect" sound and size ratio that you were looking for.

Since most people tended to want the same things, the front-ends started making it easier for us. They only showed us the variables that most people wanted to fiddle with. Some made it a graphic interface. All do the same thing, though, which is to talk to the installed codec via UNIX.

Because there was so much variability in the LAME codec, it was possible to get absolutely horrible results, which is why many people think LAME stinks. But by tweaking it until you learned the parameters, you could match CD quality in very little space. VBR was always the best way to go, it seemed. Thus, LAME incorporated something you rarely see in software: user-talent. Some people simply could do a better job than others. That's where the front-ends began interceding and making choices for you. Digital Performer's front-end is an example of that. They do not include all the choices that could lead to a steaming pile of donkey kong.

So, when talking about LAME, there's more to it than meets the eye. I would never want to believe a comparison test unless I was personally programming LAME. Getting the most out of it was never easy, but getting the best compromises has become just another push-button dialog.

Shooshie

LAME is capable of much more than most front ends now reveal.

madboy · Post by **madboy** » Wed Jul 16, 2008 12:45 pm

monkey man wrote:
chamelion wrote:... This forum's all about sharing information and expertise, and you guys have delivered in spades.
Sorry I can't deliver here Geoff.
I have no garden tools to speak of.

At least no sharp ones Metaphorical Monkster.

Careful Simian Semantical Spanking Specialist, don't trip over that Rake you're considering...

twistedtom · Post by **twistedtom** » Wed Jul 16, 2008 3:52 pm

I know this sounds lame but I do not know much about Lame I will have to play with it and see what I can do. Mostly it is for the web right?