Jocelyn
Jocelyn
Have you run the `create_manifests_and_textfiles.py` script with the `--save_transcripts_in_txt` flag yet? I repro'd this and it turns out nothing errors and yet nothing gets written if MFA can't find the...
@ekmb Thanks! I've updated the installation in this notebook, and I'll make a PR to update it for all other TTS notebooks shortly.
> @redoctopus @blisc this PR will be closed soon. Are we planning to merge to NeMo? I bet this is the final internship work. Yes--it's a large PR, so we...
Could you provide more of the error context? Does it give a line number where it fails? The TTSDataset class does expect that each entry have a `text` field that...
A few possibilities off the top of my head: Have you updated the pitch statistics `pitch_mean` and `pitch_std` to reflect the speaker data? MixerTTS is meant for single-speaker training, are...
Looks like there are some conflicts since some of the segmenting changes had already been integrated previously. Could you fix those?