Human transcriptions in pt-br

#7
by lucashueda - opened

Hey, I've checked some samples of the human transcriptions in pt-br while listening to audios, and constantly the audios were saying only the middle part of the human_transcription column text.

for example if the human transcription was "ola tudo bem com voce?", in the audio itself the person was saying only "tudo bem com". Any idea why is this happening? maybe some windowing that is larger than the actual audio while slicing samples?

Sign up or log in to comment