SpeechColab / GigaSpeech

Large, modern dataset for speech recognition
Apache License 2.0
649 stars 62 forks source link

"begin_time" is larger then audio length #125

Closed Wonder1905 closed 2 years ago

Wonder1905 commented 2 years ago

Hi, probably I'm missing something there is something, The begin_time is sometimes out of the boundaries of the audio , for example: "~/GigaSpeech/dataset/audio/audiobook/P0013/AUD0000001209.wav" file, one of his segments hold a begin_time of ~4540, I assume it is 4540 second, but there are only 704 seconds in the audio. Some segments are perfectly match for example for audio:"podcast/POD0000000001.wav" the begin_time makes totally since. Any idea what am I missing!? why are the begin_time, in a large portion of audios does not make sense..

Thanks!