Add Parakeet (HuggingFace) cloud transcription tab (#307)#393
Merged
Conversation
Third Cloud engine: NVIDIA Parakeet TDT 0.6B v3 via HuggingFace's inference
router, called directly from the browser with the user's own HF token (no
server; endpoint is CORS-open). New js/hyperaudio-lite-editor-parakeet.js does a
single synchronous multipart POST to
router.huggingface.co/together/v1/audio/transcriptions with
response_format=verbose_json + timestamp_granularities[]=word, and parses the
verified { text, words:[{word,start,end}] } shape (seconds -> data-m/data-d).
No diarization. 25-language dropdown + auto-detect.
Reuses the eye show/hide, remember-key opt-out and localStorage persistence from
the AssemblyAI work. Cloud group is now AssemblyAI / Deepgram / Parakeet (HF).
Verified end-to-end against the live endpoint with a real token: the earlier
huggingface.js#2225 "model not supported" concern did not reproduce.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #307.
Adds a third Cloud transcription engine: NVIDIA Parakeet TDT 0.6B v3 served through HuggingFace's inference router, called directly from the browser with the user's own HF token — no server.
Verification (the reason #307 is unblocked)
Re-checked against the live endpoint with a real token:
access-control-allow-origin: *).nvidia/parakeet-tdt-0.6b-v3transcribes viaPOST router.huggingface.co/together/v1/audio/transcriptionswithresponse_format=verbose_json+timestamp_granularities[]=word.{ language, duration, text, words:[{ id, word, start, end }] }with start/end in seconds.Implementation
js/hyperaudio-lite-editor-parakeet.js(parakeet-hf-service): a single synchronous multipart POST (like Deepgram, not AssemblyAI's poll). Parses the word list (seconds →data-m/data-d). No speaker labels (Parakeet doesn't diarize). Clear errors for invalid token (401/403), quota (402), unreadable media/model (400/422), inference down (5xx), and cross-origin media-fetch failures.Notes