TIMIT is a classic corpus founded by MIT in 1993 and is suitable for speech recognition, speaker classification, etc., but matlab cannot directly process the audio in TIMIT, because it is a SPHERE file format, so a certain format conversion must be performed first.
code show as below:
function wavdata=readTIMITwav(filepath) % author: % Zheng Rui % date: % 2013-12-29 % function: % Implement the function of reading audio files in the TIMIT corpus % filepath: % Path to audio file in TIMIT corpus % wavdata: % audio file clc; clear all; fidin=fopen(filepath,'r'); % Open the file and specify the operation mode of the file as read-only A = fread(fidin,inf,'int16'); % Read all the data of the file in binary form, the precision is 16-bit integer wavdata=A./(2^15); % SPHERE file header 1024 bytes, converted to wav format fclose(fidin); % close the file