search

MontrealCorpusTools / Montreal-Forced-Aligner

Command line utility for forced alignment using Kaldi
https://montrealcorpustools.github.io/Montreal-Forced-Aligner/
MIT License
1.27k stars 242 forks source link

The default behavior of mfa align #680

Closed huangruizhe closed 10 months ago

huangruizhe commented 11 months ago

Hello! Just wondered if I use this command to obtain alignment, what's the default behavior? Say, will the speaker adaptation or fine-tuning (fine-tune the pretrained model on the target data) happen?

mfa align --overwrite --clean --single_speaker ${data} english_mfa english_mfa ${data}/alignment

Just realized that --single_speaker is not related to speaker adaptation at all, while on the other hand the speaker adaptation option --uses_speaker_adaptation false is just fixed recently. BTW, this option is not mentioned in the doc. So I hope to double check.

mmcauliffe commented 10 months ago

As of 3.0.0a5, --single_speaker will imply --uses_speaker_adaptation false, since speaker adaptation doesn't add anything in this case. I've updated the docs to reflect this as well.