No self-introduction
Loading
Profile
Follow
codes (1)
speech recognition
no vote
All files in the current directory of MATLAB are copied, and type & quot; speech recognition & quot; window on the matlab command. After extracting the file correctly, you will find two folders for training and testing, each containing some files named S1. WAV，S2。 WAV,......,; ID of the speaker after each mark. These files are recorded in Microsoft wav format. On Windows system, you can listen to the tape by double clicking into the file. You can use the sound file in the & quot; train & quot; to create a database of known speakers and use the sound file currently in & quot; Test & quot; to evaluate the auto recognizer. Note that it should sound like the same sampling frequency and this value must match the size of the frame window: 256 samples should have a duration of about 30 ms: this uses the sampling frequency to achieve about 12500. If this condition is not met, you must resample the input sound (you can use & quot; resample & quot; Matlab command) or equivalent, change the window parameters to filter. This code is based on the public offering of Amin koohi http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=8802&objectType=FILE The results of distance calculation using advanced metric are improved. In this way, a better recognition rate is obtained. Complete source code, please visit http://www.advancedsourcecode.com/speaker.asp Any questions please email luigi.rosa @ tiscali.it
sarincr 1 0
1
No more~
Follows0
Fans0