Science: Sound, Percption, MP3, and Beyond
February 10, 2011 7:50 PM   Subscribe

I'm in need of graduate level schooling on psychoacoustics and audio compression on the cheap. What are the latest and/or most used textbooks on these topics?

There are two studies involved here: the science of how humans perceive sound and the art of data compression algorithms. My aim is to understand the state of the art in the overlap of the two with an emphasis on psychoacoustics, which may very well require reading up on each topic separately. Regarding psychoacoustics, I'm looking for the most comprehensive text(s) available. Something that discusses and illustrates concepts would be good. Something that specifies the math behind various models and refers to journal publications would be ideal. Regarding compression, math/reference heavy is preferred, and I'm primarily interested in techniques and algorithms that take advantage of psychoacoustic models to exploit frequency/loudness thresholds, phenomena like temporal masking, etc. Learning the various ways to perform related transforms such as FFT, wavelet, etc is also within my sights. The idea is to gain an understanding of, say, the science behind MP3, then and now, plus the challenges/limits involved in furthering similar technology.
posted by waxboy to Science & Nature (3 answers total) 5 users marked this as a favorite
 
Although it may not help directly with MP3 compression algorithms, you might try looking at Analysis, Synthesis, and Perception of Musical Sounds: The Sound of Music (Modern Acoustics and Signal Processing) for the mathematical background for understanding the issues related to perceiving music. It's not a psychoacoustic reference, either, but it has lots of good stuff in it.

I'm having a hard time thinking of the perfect reference for your questions. I'm not so much into psychoacoustics, so I hope someone else will pop in with a great reference for you.
posted by achmorrison at 8:36 PM on February 10, 2011


This thread on books at Hydrogenaudio should help.
posted by Gyan at 8:39 PM on February 10, 2011


The state of the art in psy modeling for mp3 is LAME which happens to be open source, so have a look. Also the transform used most often in audio coding is the MDCT (implementation) not the FFT.
posted by Rhomboid at 6:33 AM on February 11, 2011


« Older I taught myself Python 3.1... ...   |  Good psychiatrist for adult AD... Newer »
This thread is closed to new comments.