Skip to main content

Topic: Stochastic Processes in Audio Processing (Read 2649 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • TimYork
  • [*]
Stochastic Processes in Audio Processing
I'm in a class called "Advanced Stochastic Processes" this semester, and we have to do a project applying the mathematics of probabilty theory to engineering in some fashion.  I was considering using audio processing for a topic.  If anyone can help me out with info or articles, it would be of great aid.  I'd hate to have to do the project on financial engineering.

  • Jasper
  • [*][*][*]
Stochastic Processes in Audio Processing
Reply #1
Why don't you do it on compression, it has a lot to do with probability (for example, the entropy of a file is calculated based on the probabilities of the different values, more or less), here are two links that deal with compression:
http://www.cs.auckland.ac.nz/~peter-f/
http://www.data-compression.com/lossless.html
But there is lots more on the subject (both online and offline).
If you want to do it about lossless audio compression you might want to have a look at a program I made together with a friend of mine, it allows you to experiment with different methods of prediction and encoding of the residual signal, and there is quite a lot of documentation (in Dutch) for it, as well as some configuration files we used to perform tests, if you need them:
http://sourceforge.net/projects/pws-sces/

  • TimYork
  • [*]
Stochastic Processes in Audio Processing
Reply #2
Thanks for the info.  I'll talk with the instructor tomorrow about it, and see if he thinks it's worthy of consideration.

  • jmvalin
  • [*][*][*][*][*]
  • Developer
Stochastic Processes in Audio Processing
Reply #3
Quote
I'm in a class called "Advanced Stochastic Processes" this semester, and we have to do a project applying the mathematics of probabilty theory to engineering in some fashion.  I was considering using audio processing for a topic.  If anyone can help me out with info or articles, it would be of great aid.  I'd hate to have to do the project on financial engineering.

You can try anything related to speech recognition or speaker verification/identification. It's basically one big statistical model (gaussian mixtures, HMM's, ...)