Skip to content

Support Chatterbox multilingual frontends#375

Merged
ivan-digital merged 2 commits into
mainfrom
fix/chatterbox-multilingual-frontends
Jul 5, 2026
Merged

Support Chatterbox multilingual frontends#375
ivan-digital merged 2 commits into
mainfrom
fix/chatterbox-multilingual-frontends

Conversation

@ivan-digital

@ivan-digital ivan-digital commented Jul 5, 2026

Copy link
Copy Markdown
Member

Summary

  • Add runtime Chatterbox frontends for Chinese, Japanese, and Korean, and expand the enabled language set to 23 languages.
  • Add safe Hebrew support: pre-diacritized Hebrew with niqqud is accepted, while raw Hebrew fails early with a clear error until automatic Dicta ONNX diacritization is bundled.
  • Require the published Cangjie map in the Chatterbox bundle so Chinese does not silently fall back to raw [UNK]-heavy tokenization.
  • Default OmniVoice high-level loading to the fp16 bundle while leaving explicit int8 loading covered by tests.
  • Update Chatterbox/OmniVoice docs and translated READMEs to match runtime behavior.

Test plan

  • swift test --filter MTLTokenizerTests — 9 passed.
  • swift test --filter E2ERoundtripTests — 3 executed, 1 expected skip for missing /tmp/cbx_ve.* golden files, 0 failures.
  • swift test --skip E2E — 1,278 executed, 28 skipped, 0 failures.
  • git diff --check — passed.

@ivan-digital ivan-digital merged commit 4392fbd into main Jul 5, 2026
1 check passed
@ivan-digital ivan-digital deleted the fix/chatterbox-multilingual-frontends branch July 6, 2026 18:53
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