search

Garfield-kh / TM2D

[ICCV 2023] TM2D: Bimodality Driven 3D Dance Generation via Music-Text Integration
75 stars 3 forks source link

Audio tokenizer #9

Closed YvonneLiuym closed 4 months ago

YvonneLiuym commented 6 months ago

Hi, congrats on your great work! In dataset.py, there is opt.tokenizer_name_audio. I wonder if you can share how you obtained tokenized audio in the aistpp dataset. Thanks!

Garfield-kh commented 6 months ago

hi, thank you for the interest! We do try tokenize the music in audio2dance task, however, it seems like it is not perform as good as audio vector, so this was ignored in the later experiment.