I apologize if this issue was already posted here.

I'm currently doing a listening test involving AAC among other encoders, and I need a reliable tool to get very precisely the real bitrate of AAC file. I already noticed in the past that foobar2000 didn't make a good job with AAC (and only AAC, at least wrapped in MP4 container). It could easily be noticed with Apple's AAC VBR mode: at 128 kbps VBR, each encoding are reported as 128 kbps, never more, never less... A bit strange for VBR (BTW, Audition plugin based on faad2 always report a very round 128.000 bit: it's either a miracle or something really suspicious).
I don't know how MP4 or AAC works, but I can imagine that there's somewhere a technical 'field' (tag, metadata, header...) which stores a value computed by the encoder, and this value may be blindly reported by tools like foobar2000, Audition plug-in, etc... Correct me if I'm wrong.

Yesterday, I tried to build a bitrate table based on short samples. I don't like doing it. On short samples the average bitrate is often distorted by the weight of additionnal datas (it could be tags, but also header, container, chunks...). I prefer full tracks to build such tables. But I need it. I tried different tools and then to compute myself the real bitrate, based on the exact number of samples (by sample, I mean 1/44100 sec) and the exact filesize. I never know if I should consider 1 kbit as 1000 bit or as 1024 bit; that's why I used both value.
foobar2000 (0.83 and 0.9b10) values are very different from mine and from values computed by MrQuestionMan 0.7. The latter offers two values: the first one is a raw approach (bitrate based on filsize) I suppose and the second seems to consider the audio stream only (again, correct me if I'm wrong).



As you can see, foobar2000, MrQuestionMan, [Encspot] and manual computation compute nearly the same values for Vorbis and MP3. MrQuestionMan & manual computation are comparable for AAC's bitrate, but foobar2000 is far. I noticed that foobar2000 could report a bitrate corresponding to 120 kbps in the file's properties, but when I'm looking on bitrate variation on playback, all frames are superior to 120 kbps. Weird...

Is this a bug? Or is it intended?


N.B. I "optimized" the MP4 layout with foobar2000 - this is really needed for short samples encoded with iTunes (for 150 samples, the total size droped from 35 MB to 25 MB!), which adds at least 50 kb of unecessary data per file.

Does someone know why such difference between foobar2000 and other bitrate tool or formula exists?

The problem is that reported value is based on "average bitrate" field written by encoder, which is apparently inaccurate.
I've changed it to use audio track size / duration instead.

Thanks for your answer, Peter

It's corrected with beta 12. Thank you again