Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Diffsinger phonemizers: G2p results add langcode by default; Check if phoneme is supported by duration model #1280

Open
wants to merge 1 commit into
base: master
Choose a base branch
from

Conversation

oxygen-dioxide
Copy link
Contributor

@oxygen-dioxide oxygen-dioxide commented Sep 10, 2024

This change is made to solve strange bugs caused by G2p, where the phonemizer thought the phoneme is supported but actually it isn't.

  • For DiffSinger G2p phonemizers, if the duration uses multi-dict, replacements from bare g2p phoneme to <langcode>/<phoneme> will be added by default. No need to have {from: b, to: en/b} replacements.
  • When parsing phonetic hints, diffsinger phonemizer will check if the phoneme is supported by duration model before it desides whether to apply language prefix. (this bug made it unable to use DIFFS JA with phonetic hint without prefix such as [h a])
  • After running G2p, phonemes unsupported by duration model will be filtered out, which limits the error inside one note before it goes into duration model and affect the whole sentence.

@oxygen-dioxide oxygen-dioxide changed the title Diffsinger phonemizers: Check if phoneme is supported by duration model Diffsinger phonemizers: G2p results add langcode by default; Check if phoneme is supported by duration model Sep 10, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant