I try to write a program that could test pronunciation of user.
This program will play a wave sound, say an English sentence, and user will try to pronounce that and record his voice, and program will check the similarity of that waveforms and will give point to the user's pronunciation.
The question is, is that possible to write this program using .net? And if yes, which libraries may I use? Any guide to start?
mesut!
Assuming you are using C#.NET, I may have an answer for you. A simple Google search led me to this YouTube video, which shows you how to "Plot Audio Waveforms": http://www.youtube.com/watch?v=ZnFoVuOVrUQ
Good luck, and the rest is up to you!
Music recognition programs usually represent sound in a spectral domain (as to capture pitch).
From some spectral representation, a distance measure (Euclidean or other) may be used.
For speech, a time dimension is necessary.
In a spatial domain, warping measures that account for time-deformations (slow vs fast) seem like a good choice as well.
Template matching algorithms could be good candidates as well like in OCR algorithms, but now you would be dealing with a sound wave.
A usual consideration with sound wave algorithms are sliding windows or segmentation.
