fix(deepgram): use punctuated_word for word-level transcripts

[rosetta-livekit-bot]

Summary

• use Deepgram punctuated_word values for word-level transcripts when punctuate or smart_format is enabled
• preserve raw word values when formatting is disabled and fall back to them when punctuated_word is absent or empty
• apply the behavior to both prerecorded and streaming transcription paths

Ports livekit/agents#6699.

Source diff coverage

Source diff coverage

• livekit-plugins/livekit-plugins-deepgram/livekit/plugins/deepgram/stt.py: adapted to plugins/deepgram/src/stt.ts. The target already has corresponding prerecorded and streaming conversion paths, so the Python option forwarding and _word_text selection logic were translated to camelCase TypeScript while preserving defaults and fallback behavior.
• Source tests: not applicable. The authoritative source PR did not add or modify test files, so no target tests were added.

Validation

• pnpm test plugins/deepgram (4 passed, 3 credential-gated tests skipped)
• pnpm build (40 packages passed)
• pnpm --filter @livekit/agents-plugin-deepgram lint (passed with pre-existing warnings)
• pnpm lint (blocked by an unrelated existing Prettier error in plugins/minimax/src/models.ts)
• cue-cli runtime validation was not run because DEEPGRAM_API_KEY is unavailable in the environment

---

Ported from livekit/agents#6699

Original PR description

Summary

SpeechData.text and SpeechData.words are built from different Deepgram fields, so the two disagree on the same object.

text comes from the transcript, which honours the punctuate option. The word list was built from the raw per-word field, which is lowercase and unpunctuated regardless of any option:

text=alt["transcript"],                    # honours `punctuate`
...
words=[
    TimedString(
        text=word.get("word", ""),         # never punctuated

Deepgram returns punctuated_word alongside word, in the very dict that comprehension iterates:

keys : ['confidence', 'end', 'punctuated_word', 'start', 'word']
word            : hello my name is alex
punctuated_word : Hello. My name is Alex.

Field selection across the four cases:

_word_text({"word": "alex", "punctuated_word": "Alex."}, punctuate=True)   # -> "Alex."
_word_text({"word": "alex"},                             punctuate=True)   # -> "alex"
_word_text({"word": "alex", "punctuated_word": ""},      punctuate=True)   # -> "alex"
_word_text({"word": "alex", "punctuated_word": "Alex."}, punctuate=False)  # -> "alex"

[changeset-bot]

🦋 Changeset detected

Latest commit: f5741b4

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 39 packages

Name	Type
@livekit/agents-plugin-deepgram	Patch
@livekit/agents	Patch
@livekit/agents-plugin-anam	Patch
@livekit/agents-plugin-anthropic	Patch
@livekit/agents-plugin-assemblyai	Patch
@livekit/agents-plugin-azure	Patch
@livekit/agents-plugin-baseten	Patch
@livekit/agents-plugin-bey	Patch
@livekit/agents-plugin-cartesia	Patch
@livekit/agents-plugin-cerebras	Patch
@livekit/agents-plugin-did	Patch
@livekit/agents-plugin-elevenlabs	Patch
@livekit/agents-plugin-fishaudio	Patch
@livekit/agents-plugin-google	Patch
@livekit/agents-plugin-hedra	Patch
@livekit/agents-plugin-hume	Patch
@livekit/agents-plugin-inworld	Patch
@livekit/agents-plugin-krisp	Patch
@livekit/agents-plugin-lemonslice	Patch
@livekit/agents-plugin-liveavatar	Patch
@livekit/agents-plugin-livekit	Patch
@livekit/agents-plugin-minimax	Patch
@livekit/agents-plugin-mistral	Patch
@livekit/agents-plugin-mistralai	Patch
@livekit/agents-plugin-neuphonic	Patch
@livekit/agents-plugin-openai	Patch
@livekit/agents-plugin-perplexity	Patch
@livekit/agents-plugin-phonic	Patch
@livekit/agents-plugin-protoface	Patch
@livekit/agents-plugin-resemble	Patch
@livekit/agents-plugin-rime	Patch
@livekit/agents-plugin-runway	Patch
@livekit/agents-plugin-sarvam	Patch
@livekit/agents-plugin-silero	Patch
@livekit/agents-plugin-soniox	Patch
@livekit/agents-plugin-tavus	Patch
@livekit/agents-plugin-trugen	Patch
@livekit/agents-plugin-xai	Patch
@livekit/agents-plugins-test	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no potential bugs to report.

View in Devin Review to see 1 additional finding.

[chenghao-mou]

Port verification: livekit/agents#6699 → this PR

Compared this PR's diff against the merged Python source (livekit-plugins/livekit-plugins-deepgram/livekit/plugins/deepgram/stt.py) line-by-line.

Verdict: Faithful, mechanical port. No correctness issues found. Labeled effort:low — a single small, well-isolated helper function extracted and threaded through two existing call sites with matching default values.

Full analysis

• _word_text (Python) → wordText (JS): identical logic — return raw word when use_punctuated_word is false; otherwise fall back from punctuated_word to word when the punctuated field is absent or empty (or raw / || raw).
• Both call sites thread punctuate or smart_format (Python) / punctuate || smartFormat (JS) into the helper, applied identically to the streaming (live_transcription_to_speech_data / liveTranscriptionToSpeechData) and prerecorded (prerecorded_transcription_to_speech_event / prerecordedTranscriptionToSpeechEvent) paths.
• Default value for the new parameter (use_punctuated_word: bool = True / usePunctuatedWord: boolean = true) matches in both languages, so any caller that doesn't pass the flag explicitly keeps prior (punctuated) behavior.
• The source PR added no tests (test coverage for this behavior was added in a different, previously-merged PR), so the JS port correctly adds none either — matches the "not applicable" note in the PR's own source diff coverage table.
• Verified the JS STTOptions.punctuate/smartFormat fields already existed pre-PR with the same semantics as Python's punctuate/smart_format, so no config drift was introduced by this change.

Not independently re-run: relied on the PR's own reported pnpm test plugins/deepgram result (4 passed, 3 credential-gated tests skipped), since DEEPGRAM_API_KEY isn't available here either.

---

Generated by Claude Code