word level timestamps

[li-henan]

Dear author, could you help me to solve this problem?
I input :whisper p226_007.wav --model tiny --device cpu --word_timestamps True , cause I want to see word level timestamps, but errors occur like this : AssertionError: 3D tensors expect 2 values for padding , is my python version wrong?
Thanks very much for replying

[phineas-pta]

try other audio

[li-henan]

thanks for replying, what kind of audio should I input?Is there any requirements？

[phineas-pta]

your audio may be corrupted

try any other one

[li-henan]

but these wavs can be used in whisper p226_007.wav --model tiny --device cpu ，when not have --word_timestamps True