Converting audio files.

[jujufactory]

When converting MP3 of WMV files to WAV for editing does it matter which conversion software I use? At the same sampling rate will all software deliver the same sound? Just wondering.

[Jeff Hayat]

Originally Posted by jujufactory

When converting MP3 of WMV files to WAV for editing does it matter which conversion software I use? At the same sampling rate will all software deliver the same sound? Just wondering.

Interesting Q. Theoretically, they should be the same - the same way all DAWs all produce the same results b/c it's just math, all sw conversion apps should all produce the same results. But I wouldn't bet the farm on that without first doing a null test. Has anyone done any testing?

[jujufactory]

I'll take that as a NO.

[Dayo]

No they are not all the same - very far from it

[jujufactory]

Davo could you be more specific? I currently have the option of converting mp3 and wmv audio files to 16 bit 48hz wav using either:

the Cubase 5 timeline and exporting

OR

using XILISOFT Audio Maker Audio Maker - Audio Converter, Audio CD Burner, CD to MP3 Ripper to convert.

Which one would you recommend?

[Jeff Hayat]

Originally Posted by jujufactory

I'll take that as a NO.

Why would you take that as a NO? It's simple - if the results from app A and the results from app B completely null, then the answer is yes. If not, then it's no.

Cheers.

[jujufactory]

"if the results from app A and the results from app B completely null, then the answer is yes".

How do I test that?

[fastlanestoner]

Try the conversion through different software and whichever you like best...use.

Why is that so difficult to grasp?

[Sigma]

Originally Posted by fastlanestoner

Try the conversion through different software and whichever you like best...use.

Why is that so difficult to grasp?

lol us philly people.... coarse, gruff and to the point LOLZ!!!!!!

[theblue1]

Originally Posted by Jeff Hayat

Interesting Q. Theoretically, they should be the same - the same way all DAWs all produce the same results b/c it's just math, all sw conversion apps should all produce the same results. But I wouldn't bet the farm on that without first doing a null test. Has anyone done any testing?

Actually, it's math and a lot more.

They call them perceptual encoding algorithms because the concept is to remove as much data as possible while effecting the perceived sound as little as possible.

And different algorithms have somewhat different approaches that can make a very real difference from codec (encoder/decoder) to codec.


But what a surprising number of us 'engineers' seem to not get is that these codecs tend to be highly configurable.

For instance, here is the page that documents all the command line switches for the LAME encoder:

http://lame.cvs.sourceforge.net/viewvc/lame/lame/USAGE

... and that is a whole lot of configurability that will potentially change much about the sound and whether a given bitrate sounds good on specific program material -- or not.


But most folks -- even many of us RE types -- just use a so-called wrapper program that dumbs down all those options, usually based on a number of probably reasonable but not necessarily universally applicable presumptions.

[theblue1]

Originally Posted by Jeff Hayat

Interesting Q. Theoretically, they should be the same - the same way all DAWs all produce the same results b/c it's just math, all sw conversion apps should all produce the same results. But I wouldn't bet the farm on that without first doing a null test. Has anyone done any testing?

Null testing is best when confined to simply checking indenticality of two digital signal streams that were derived from the same ADC clocking. (If the timing isn't locked, the potential for nulling is removed.)

Comparing two files resulting from two different perceptual testing algorithms via null testing would be unlikely to provide much valuable information, particularly with regard to which algo and settings produced the best results for a given piece of program material -- although it's understood that the question was whether different algorithms (presumably set for the same target bitrate) would produce the same results.

It's highly unlikely that the results would be indentical -- but, depending on the algorithms, the target bitrates, as well, crucially, as all the other operational parameters, you might get two files that untrained listeners might not be able to reliably tell apart.


Like the Philly guys say, give them all a try (after coming to grips with whether or not you want to deal with configuration issues or just want to trust the default options in a wrapper) and go with the one you like best for the kind of production/music you do.

[Jeff Hayat]

Originally Posted by jujufactory

"if the results from app A and the results from app B completely null, then the answer is yes".

How do I test that?

You take the file from app a, and the file from app b, line them up in a sequencer, flip the phase on one channel, and see if they nill.

theblue1 > So you are saying that when it comes to conversion software, it's not just a matter of 0s and 1s, but algos that different aps use? Didn't know that - thanks for the info.

Cheers.

[jujufactory]

I think this thread would have been much more productive without the sarcasm. There is nothing obvious about the original question so I don't see why some had to be sarcastic.

At least now I know that all software will not yield the same conversion results. I will play it safe and do the conversion using Cubase 5 which at least is a professional program.