Welcome Guest ( Log In | Register )

320K MP3 vs 256K WMA10, Observations
Glenn Gundlach
post Feb 4 2013, 08:19
Post #1

Group: Members
Posts: 380
Joined: 19-April 08
From: LA
Member No.: 52914

There has been quite a bit of MP3 S/N and negative comments about WMA files so I tried some experiments in Adobe Audition version 3.0.1 build 8347.0. I was interested in what kind of artifacts happen on boundaries so I created a series tone bursts starting with a 15 second sweep of 10 Hz to 22000 Hz at 44100. Then I inserted many 0.1 second silences with no 'fades' to smooth the boundaries so the file is now 24.7 seconds and saved as a WAV. Then I converted to 320 Kbit MP3 and a 256K WMA10 file and started looking at the residue at the start and stop of the bursts. WMA is significantly lower - about 10-15 dB on average. Well that's nice but what happens with actual music since that's what this is all about? I took Michael Buble's 'Everything' and lined up to the sample of the FLAC/WAV to the MP3 and WMA. Invert the phase of the FLAC and add to the MP3 and then add to the WMA and save those files. The difference is not subtle. I will no longer be using MP3s in my car. I'm of the opinion that the difference file is the 'distortion' and the lower the better. I do not know if there is a better MP3 encoder than in Audition but it's what I've been using for the last 7 years.

The WMA files tag in MP3Tag exactly like MP3 files and while I don't care about the space used, the WMA files are somewhat smaller.

I've been looking around to find out how to post files and screen shots but I'd be happy to share what I have so far.

Go to the top of the page
+Quote Post
Start new topic
post Feb 4 2013, 15:39
Post #2

Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180

…which is not something that we wish to promote, contrary to the belief of some.

The purpose that might be served is that the OP would rethink the fatally flawed idea that difference signals, spectrograms, or any other visually or measurement-based metrics need have any correlation with perceptual quality, which obviously they do not. Logic aside, such means of measuring are clearly specified as being irrelevant in TOS #8, so this thread has no future if it does not yield a revised approach to the comparison in question. Again, the fact that this was posted not by a clueless new user but by someone who has been registered for years and posts reasonably frequently makes it all the less understandable.
Go to the top of the page
+Quote Post

Posts in this topic

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:


RSS Lo-Fi Version Time is now: 29th November 2015 - 00:34