Question about fine-tuning
#27
by
jz703
- opened
Hi, I'd like to fine tuning on a dataset that the target output should be the phoneme of the words.(e.g. "examination and testimony" should be "ɪɡzæmənˈeɪʃən ˈænd tˈɛstɪmˌoʊni") I just want to know if this is possible with whisper if I build the vocab carefully. And is this more like a transcription task or translation. Thank you!