search

crouchred / speaker-recognition-py3

Base on MFCC and GMM(基于MFCC和高斯混合模型的语音识别)
Apache License 2.0
245 stars 79 forks source link

model.out #7

Closed abdvlz closed 5 years ago

abdvlz commented 5 years ago

i didnt understand how would multiple wav files of different speakers are compared with only one model.

it always outputs score of 1.0 and arbitrary result.

could you please explain ??

thanks lots

crouchred commented 5 years ago

It does generate multiple models. Maybe you put all wav files in the same folder and then they are identified as one label? Please try put each file in a folder if you want to use a single file as a label. using wildcard like '/tmp/person*', it will use every folder detected as label. W

alonyomtov123 commented 4 years ago

the output for me is also always 1.0. can someone help?