Modeling speech signals and recognizing a speaker
ReadWavMakeFrame Make model DataArray.wav filePointer to frames shift Frame duration MFCC’ s model Pre emphasis Windowing Powers spectrum Mel spectrum Mel cepustal Pointer to Frame Feature vector
The problem includes writing C++ programs for three stages: Modeling : Modeling speech signals Training : Creating a database of models for different speakers Testing : Identifying a speaker
Modeling : It will have four different modules/functions : 1. ReadWave Input : Input to this function is the.wav files obtained from different speakers. Output : Output of this function is an array containing the data from the.wav file. 2. MakeFrame Input : (i) Array obtained from the function ReadWave. (ii) Duration of a frame. (iii) Shift between two consecutive frames. Output : Pointer to arrays containing the frames.
3. CalucalateMFCC Input : Pointer to an array containing a frame. Output : An array containing the MFCC’s (feature vector.) 4. CreateModel: Input : The feature vectors Output : The model. Eg Mean and Co-variance of feature vectors.