ATTENTION REAL AUDIO COMPRESSORS

[beaniboy67]

guys.Im a 5th year audio technology student at Glasgow Calewdonian university.For my honours project i done a subjective comparison on Real audio and wma codecs.The tests that were done were listening tests using the only recognised testing methodology.This method was called MUSHRA and is currently being standardised by the ITU(international telecommunications union.As part of the project i also done a small experiment on the pre preperation of audio.This entailed analysing the cut off frequencies of audio files once they had been compressed with white noise.It can also be done with audio.This will show you where the cut off frequencies of the audio files are.Once you find the cut off frequencies of the codec you are analysing you can automatically chop anything above this frequency with cool edit or sound forge.There is spectram anlysis in these packages aswell for analysing.All you would do is then put youre mix into sound forge and chop of the content above the cut off point that will not be encoded anyway.I compared these edited audio pieces to the un edited versions of the same piece and the results showed that pre preperation in this way does offer a little extra on the quality side.Many different types of music were used in the tests and many off the codecs did respond well to this trick.The encoder (real or wm)has less work to do and manages to spend more time analysing and encoding important content instead of the stuff that will be thrown away anyway.Give it a try and you might find that it works.

[Bold Soul]

It's wonderful that you can learn this at university. Most of us in the workforce had to gain experience in these subjects by trial and error. Think of how technological innovation in the next decade will explode!

Keep up the good work!

[ May 04, 2003, 04:27 PM: Message edited by: Bold Soul ]

[imported_Gman]

If I am for example encoding real audio 32Kbs stereo files, which I think have a high frequency cut off of 10k herz, then I should use the equalizer in sound forge to chop off all the frequencies above 10K herz and then encode it. You are saying the resultant file should sound better ?

-G

[beaniboy67]

G man.When i done the tests i was using small excerpts of differing types of music.For example i would firstly encode white noise with 32kbps real media codec and then analyse it for the codecs cut off frequency.When i analysed the frequency content or RM codec at 32kbps it was a cut off point of around 14khz or 14.5khz.Basically i would then encode jazz music,soul,clasical,voice over audio,voice alone and broadcasting type pieces(commentary ect)cutting off all frequencies above 14.5khz.This was too see how each codec performed against each other at matching bit rates(32wm and 32rm) for different types of music but also at the same time comparing the filtered files of the same pieces with the un filtered versions.The results showed that for each piece the codecs did perform slightly better for certain types of music.The one thing that is definate is that u do not lose any quality whatsoever by doing this pre preperation as the content is thrown away anyway.For some pieces of music you do get better results.If doing a mix encoding session it would be very difficult to analyse the whole mix as there is so much to listen back to and it may be difficlut to pinpoint differences in general.It may be that at certain points of the mix there is quality differences.During encoding the codec may have loads of audio content to process very quickly and it may deal with the content that will be heard a lot easieri f the frequencies above 14.5khz are dismissed.