Project 1: Text Input System
Java-base
As per Specs, Text Input Program Specifications.doc.
?/span>
Project 2: Speech recognition System
?/span>
1) Record single tone voice (eg. A ,e , I , o , u) to .wav file. For each
tone, there are 10 recorded .wav file.
?/span>
2) Feature Extraction - covert all the the windows PCM(.wav) files to
frequency information(frequency Domain) in text file.
Preferred using Fast Fouier Transformation. The .wav file format is ,
single
channel, sampling rate 8000, 8 bit.
3)feed to the frequency information of the 10 .wav files to a back
propagation neural network to "learn" the tone.
4) The neural network should be able to identify the tone corresponding to
the correct phonemes.
?Basically, this project no need to start from scratch as the person already did something (projectsource.zip):
?/span>
SimpleAudioRecorder.java = a textbase sound recorder ...
SoundRecorder.java = a graphical implementation of SimpleAudioRecorder.java
(not working)
FastFourierTransform.java = to convert a wave file
Test.wav = recorded wave file to test FastFourierTransform.java
TrainingSet.java = part of A neural network model (not completed)
[email protected]