Feature/stt refine

[jundoring]

Summary by CodeRabbit

• 새로운 기능

  • STT 오류를 보정하는 기능이 도입되어 음성 입력의 정정된 텍스트를 활용합니다.

• 개선사항

  • 여러 최종 트랜스크립트를 누적해 더 정확한 쿼리 구성을 지원합니다.
  • 의도 분류 로직이 개선되어 더 안정적이고 풍부한 분류 결과를 제공합니다.

• 잡무(Chores)

  • 더 이상 사용되지 않는 음성 파일 변환 스크립트를 제거했습니다.

[coderabbitai]

Warning

Rate limit exceeded

@jundoring has exceeded the limit for the number of commits that can be reviewed per hour. Please wait 3 minutes and 52 seconds before requesting another review.

Your organization is not enrolled in usage-based pricing. Contact your admin to enable usage-based pricing to continue reviews beyond the rate limit, or try again in 3 minutes and 52 seconds.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 052b54e2-bcaa-4459-8581-df49c5df868e

📥 Commits

Reviewing files that changed from the base of the PR and between 0ec8fec and 61200d7.

📒 Files selected for processing (1)

• app/graph/nodes/classify/intent_gate.py

Walkthrough

내부 QA 실행 경로를 그래프 직접 호출에서 추적 가능한 래퍼(traced_internal_run)로 전환하고, WebSocket STT 누적 로직을 최종 파트 리스트로 변경했으며, 의도 분류 노드에 STT 정정(corrected_text)을 추가했습니다. 테스트 WAV 경로와 불필요한 스크립트도 정리되었습니다.

Changes

Cohort / File(s)	Summary
QA 실행 경로 개선 app/api/qa.py, app/core/dependencies.py	app/api/qa.py의 내부 QA 실행이 직접 text_pipeline.graph.invoke(initial_state) 대신 traced_internal_run(initial_state)(asyncio.to_thread 통해 호출)으로 변경됨. app/core/dependencies.py에 @traceable(name="internal_qa_pipeline")로 장식된 traced_internal_run(initial_state: dict)->dict 추가.
STT 트랜스크립트 누적 app/api/websocket.py	ws_stream에서 최종 트랜스크립트를 단일 문자열 덮어쓰기 대신 final_parts: list[str]로 누적하고, 쿼리는 (sep.join(final_parts) or last_interim).strip()로 구성되도록 변경(언어별 구분자 적용 포함).
의도 분류 및 STT 정정 app/graph/nodes/classify/intent_gate.py	LLM 프롬프트/출력 계약을 확장하여 먼저 corrected_text(STT 오류 정정)를 반환하고 그 후 의도 랭킹을 수행하도록 로직 변경. normalized_text에 정정된 텍스트를 넣고 추적 페이로드에 original_text/corrected_text를 기록. max_tokens 증가(60→150).
테스트 및 정리 app/core/ws_stt_test.py, app/graph/graph_builder.py, audio/m4a.py	테스트 오디오 파일 경로를 audio/wav/6_sh.wav로 변경, 임포트/주석 사이의 빈 줄 제거, 더 이상 필요 없는 audio/m4a.py 스크립트 삭제.

Sequence Diagram(s)

sequenceDiagram
    participant Client
    participant QA_API as "app/api/qa.py"
    participant Thread as "asyncio.to_thread"
    participant Traced as "traced_internal_run\n(`@traceable`)"
    participant Graph as "text_pipeline.graph.invoke"

    Client->>QA_API: 내부 QA 요청
    QA_API->>Thread: to_thread(traced_internal_run, initial_state)
    Thread->>Traced: 실행
    Traced->>Graph: graph.invoke(initial_state)
    Graph-->>Traced: 결과(state)
    Traced-->>Thread: 반환(state)
    Thread-->>QA_API: 결과(state)
    QA_API-->>Client: 응답 반환

Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~30 minutes

Possibly related PRs

• Feature/event #55: WebSocket STT→쿼리 조립 및 실시간 LLM 제어 흐름 변경과 관련된 코드(유사 수정 지점).
• Feature/tts stream #25: app/api/websocket.py의 STT 최종/중간 트랜스크립트 집계 로직을 수정한 PR로 코드 레벨 연관성 있음.

시

🐰 달그락 달그락 귀를 세우고,
잘못된 말도 살며시 고쳐 담아,
조각난 음성들을 모아 묶으면,
의도가 반짝 얼굴을 내미네.
추적의 실로 길을 잇는 나, 춤추며 축하해요. 🥕✨

🚥 Pre-merge checks | ✅ 1 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 33.33% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.
Title check	❓ Inconclusive	PR 제목이 변경사항의 주요 내용을 모호하게 설명하고 있습니다. 'Feature/stt refine'은 STT 관련 개선이라는 의도는 있으나, 구체적인 변경 내용(STT 오류 수정, 다중 최종 토사본 수집, 내부 QA 파이프라인 등)을 명확하게 전달하지 못합니다.	PR 제목을 더 구체적으로 변경하세요. 예: 'Refine STT transcript handling with multi-final support and intent correction' 또는 '다중 최종 토사본 수집 및 의도 분류 개선'.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feature/stt-refine

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@app/api/websocket.py`:
- Line 346: The current join always uses a space: change the construction of
query so it uses an empty separator for CJK languages and a space otherwise —
e.g., determine sep = "" if last_lang_code in ("ja","zh") else " ", then set
query = (sep.join(final_parts) or last_interim).strip(); update the code around
variables final_parts, last_interim and last_lang_code (where query is assigned)
to use that language-aware separator.

In `@app/graph/nodes/classify/intent_gate.py`:
- Around line 74-80: The parsed JSON from llm.chat (variable parsed in
intent_gate.py) must be validated as a dict before using parsed.get(); modify
the logic in the block around raw = llm.chat(...) and parsed = json.loads(raw)
to: 1) catch json.JSONDecodeError/ValueError and route to the existing fallback
flow (e.g., set ranking=_DEFAULT_RANKING, place_category=None,
method="llm_fallback_invalid_json", error=str(e)); 2) if json.loads succeeds but
parsed is not a dict (e.g., list, string, null), treat that as an invalid schema
and perform the same fallback handling; and 3) only access
parsed.get("corrected_text") when isinstance(parsed, dict) to avoid
AttributeError and ensure fallback is used for malformed responses.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: ff52fdb0-e161-498f-b0d4-4b78dc10420c

📥 Commits

Reviewing files that changed from the base of the PR and between bff6597 and 3fd4661.

📒 Files selected for processing (7)

• app/api/qa.py
• app/api/websocket.py
• app/core/dependencies.py
• app/core/ws_stt_test.py
• app/graph/graph_builder.py
• app/graph/nodes/classify/intent_gate.py
• audio/m4a.py

💤 Files with no reviewable changes (2)

• app/graph/graph_builder.py
• audio/m4a.py