Skip to content

feat(models): add Roman Hinglish Whisper model - #46

Merged
Mr-Sunglasses merged 2 commits into
mainfrom
feat/hinglish-roman
Sep 3, 2026
Merged

feat(models): add Roman Hinglish Whisper model#46
Mr-Sunglasses merged 2 commits into
mainfrom
feat/hinglish-roman

Conversation

@Mr-Sunglasses

@Mr-Sunglasses Mr-Sunglasses commented Sep 3, 2026

Copy link
Copy Markdown
Member

Problem

VocaPhone PR #218 adds an opt-in Roman Hinglish output mode backed by Oriserve/Whisper-Hindi2Hinglish-Apex. The gateway catalog does not currently expose that model, and passing its hinglish_roman output contract directly to whisper.cpp would use an invalid decoder language token.

Summary

  • Add the pinned 574 MB ggml-apex-hinglish-q5_0.bin GGML model from the PR's immutable third-party conversion, with its SHA-256 pin and Apache-2.0 attribution.
  • Advertise hinglish_roman as a Latin-script output contract and map it to Whisper's hi decoder token while preserving the contract in health/session responses.
  • Accept the exact underscore-bearing output code without broadening language-tag validation to arbitrary underscore values.
  • Reject incompatible explicit language requests and enforce Roman-only output even when a client requests auto, so Devanagari leakage cannot silently reach the cursor.
  • Regenerate the model reference docs and keep the experimental model out of recommendation sets.

Verification

  • just test — lint, WPS, mypy, package/audit, 508 tests, and Compose config passed
  • uv run scripts/verify-model-pins.py --only whisper.cpp:ggml-apex-hinglish-q5_0.bin — 1 pin checked, 0 mismatched
  • uv run scripts/generate_model_docs.py --check
  • Container build — not applicable; no Dockerfile or lockfile changed

Privacy and security

  • No secrets, recordings, transcripts, or private hostnames added
  • No weakening of bearer auth, upload limits, retention, or default binding without discussion

@netlify

netlify Bot commented Sep 3, 2026

Copy link
Copy Markdown

Deploy Preview for voca-gateway canceled.

Name Link
🔨 Latest commit 2c5eada
🔍 Latest deploy log https://app.netlify.com/projects/voca-gateway/deploys/6a99bee55d3a0500081f6df0

@greptile-apps greptile-apps Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Mr-Sunglasses has reached the 50-credit limit for trial accounts. To continue receiving code reviews, upgrade your plan.

@greptile-apps greptile-apps Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Mr-Sunglasses has reached the 50-credit limit for trial accounts. To continue receiving code reviews, upgrade your plan.

@greptile-apps greptile-apps Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Mr-Sunglasses has reached the 50-credit limit for trial accounts. To continue receiving code reviews, upgrade your plan.

@Mr-Sunglasses
Mr-Sunglasses merged commit d6ed5fb into main Sep 3, 2026
7 checks passed
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