Using Sorenson Squeeze 3.5 for Windows, I converted several wav to mp4 using AAC at bitrate 128 CBR. When I play back these mp4 with Winamp 2.90 using latest FAAD2 in_mp4 plugin, the bitrate is shown as "0". Anybody else have this problem? Checking the mp4 with mp4ui shows bitrate=0 also. Otherwise I must say I like this encoder - the quality is much better than Quicktime 6.1 on tough samples like fatboy.wav, on a par with Nero, I think.

Same thing happens here. Some bug in the way they write audio streams.

The encoder is FhG professional AAC, that's the reason it sounds good.

I plan to have an AAC listening test set up and ready in some time. Probably this weekend.

I found this out by accident the other day after I used the encoder in Sorenson to try making a few files. When WinAmp said that the bitrate was zero, I loaded up Foobar2000 and selected a file then chose "Show file info" by right-clicking on the file. I then hit update and closed it. After that the bitrate seemed to show in both Winamp and foobar and quiting and restarting the programs .

I can confirm, nykobing's fix works. Thanks for the help guys!

lazyn00b ,
i thought QT was better than both Psytel/Nero and FhG ?! (according to Guruboolez)

That is really subjective opinion. For example, I prefer FhG's professional over Dolby's consumer codec in most samples, but it is not always the case.

Some people are more sensitive to frequency-domain artifacts, and in those cases Dolby consumer might be better (it rarely uses so called short blocks) , while FhG has much better pre-echo control.

Anyway, this kind of testing requires lot of trained listeners to give something which is credible - and even in that case it might not correlate with your hearing

Please do not misunderstand me, I really think that Guruboolez has extraordinary hearing - but I just want to clarify that it could differ from opinion of some other expirienced listener.

I just happen to have testing results of Dolby Consumer vs. Professional codec done during development of Dolby's codec (AES paper) and IIRC - some listeners really prefered consumer codec vs . pro one - but others didn't.

I really think that Guruboolez has extraordinary hearing.



Yes for sure !!
Even he says it's just a question of training !!

Thank you for your reply Ivan !

Certainly not. A real batman can't be satisfied by Fhg Radium CBR 128 encodings ; I was pleased by some times ago by Fhg Radium CBR 128 encodings ; So, I can't be a batman....



This may be true. Maybe more a question of taste, subjectivity (I really hate some artifacts), and not really hearing abilities. Don't forget that I'm judging with headphone, good soundcard, and on small part of a sample. This is a critical, uncommon situation.

I think it has something to do with the subjective measure of artifact "badness" - it is not directly related to the NMR (noise-mask-ratio), but it certainly has something to do with the advanced neural processing.

All tests done with these codecs (Professional, Consumer...) were done in the same way - very good headphones (they use $4K ones in MPEG listening tests and material transfered to tape with excellent DA convertors - so I don't think way of listening affects results.


Back to them, FhG Professional codec usually uses a lot more short blocks, and psychoacoustic model is more-general "model II" type - while Dolby consumer uses more long blocks (bigger frequency resolution), more TNS and psychoacoustic which is much more similar to AC-3 psychoacoustic model - so, there are quite obvious differences between these two, and artifacts they generate are different.

Interesting though, is that I remember that on some samples Dolby was really better by a large margin - but overall, complete results were little bit worse that prof. codec.

But are MPEG LT based on very short samples ? This is really helpful, at least for me. I haven't any synthetic hearing abilities ; only analytic, and on a very short period. My memory isn't trained at all. On a ten seconds sample, it's really hard for me to hear what I can detect on a 2-3 second one (this is maybe trivial, but maybe not - I don't know).

Usually, tests are conducted on short clips:

si02 - 7.7 sec
es01 - 10.4 sec
es02 - 8.6 sec
sm02 - 10.4 sec
sc01 - 10.1 sec

etc..

Thanks for the precision.
Now, I have to be employed by the MPEG Commity

France Telecom is in charge of SBR formal listening tests, you might want to try

This company is totaly ruined... I want money, not 20 free hours of communication and a Nokia phone