CS691 MACHINE LEARNING
PROJECT PROGRESS
By: Jun Li
Project: Speech and Music discrimination
Stage: Reading Paper
Paper in Reading
1.
Speech/music
Discrimination Based On Posterior Probability
2.
SPEECH/MUSIC
DISCRIMINATION FOR MULTIMEDIA APPLICATIONS
3.
Construction
and Evaluation of a Robust Multifeature Speech/Music
4.
SPEECH,
MUSIC AND SONGS DISCRIMINATION IN THE CONTEXT OF HANDSETS
5. Speech and Music Discrimination based on Signal Modulation
6.
RHYTHM DETECTION FOR SPEECH-MUSIC
DISCRIMINATION IN MPEG
7.
ROBUST HMM-BASED SPEECH/MUSIC
SEGMENTATION
8.
Speech-Music Discrimination
from MPEG-1 Bitstream
9.
ROBUST
SPEECH / MUSIC CLASSIFICATION IN AUDIO DOCUMENTS
10.
Audio Signal
Classification: An Overview
11.
The
Method of an Audio Data Classification and Segmentation
12.
A comparison
of features for speech, music discrimination
To find
Real Time discrimination of broadcast speech/music, Proc IEEE Int. Conf. On Acoustics, Speech, Signal Processing
2004.04.17
Installed fftw3.0 and libsndfile. Read document; try to use code from Marcel, and library of fftw and libsndfile to find a way to process audio file
Failed to link with library of fftw and libsndfile.
2004.04 26
With Marcel's help, libsndfile and fftw compile succeed
2004.4.27
Read source code of FFTW and libsndfile. In the meantime looking for referrence on audio processing. Tyr to modify the code to calculate frequency, energy, zero-crossing from FFT transform
2004.5.30
Download and install software for audio processing :"Maaate.3.1 Source Tarball" http://www.cmis.csiro.au/maaate/docs/download.html. Trying to use Maaate's API and modify the code to calculate frequency, energy, zero-crossing from FFT transform
2004.06 22
By using Maaate.3.1's API, I finish procedure for audio processing and calculated audio features: energy and magnitude
Working with Weka and read help document of weka
2004.07.12
Coding finished. Will do test and data collecting in the following several days. Documentation and post will be done soon
I am working on documentation: please click here to see detail