@startuml
!theme mono

title 음성-텍스트 변환 내부 시퀀스 (UFR-STT-020)

participant "API Gateway<<E>>" as Gateway
participant "SttController" as Controller
participant "SttService" as Service
participant "TranscriptionEngine" as Engine
participant "Azure Speech<<E>>" as Speech
participant "SttRepository" as Repository
database "PostgreSQL<<E>>" as DB
queue "Event Hub<<E>>" as EventHub

Gateway -> Controller: POST /api/v1/stt/transcribe\n{sessionId, audioFile}
activate Controller

Controller -> Service: transcribeAudio(sessionId, audioFile)
activate Service

Service -> Repository: findSessionById(sessionId)
activate Repository
Repository -> DB: STT 세션 조회\n(세션ID 기준)
DB --> Repository: session data
Repository --> Service: SttSession entity
deactivate Repository

alt 실시간 변환 모드
    Service -> Engine: streamingTranscribe(audioFile)
    activate Engine

    Engine -> Speech: createRecognizer()\nsetContinuousRecognition()
    note right
      Azure Speech 설정:
      - Mode: Continuous
      - Language: ko-KR
      - Enable diarization
      - Profanity filter
    end note

    Speech --> Engine: recognizer instance

    loop 오디오 청크 처리
        Engine -> Speech: recognizeOnceAsync(audioChunk)
        Speech --> Engine: recognition result
        note right
          결과 포함:
          - Text
          - Confidence
          - Duration
          - Speaker ID
        end note

        Engine -> Engine: validateConfidence(result)
        note right
          신뢰도 검증:
          - Threshold: 0.7
          - Low confidence 처리
        end note

        Engine --> Service: transcription segment

        Service -> Repository: saveSttSegment(segment)
        activate Repository
        Repository -> DB: STT 세그먼트 저장\n(세션ID, 텍스트, 신뢰도, 타임스탬프)
        DB --> Repository: saved
        Repository --> Service: segment saved
        deactivate Repository

        Service -> EventHub: publish(TranscriptionSegmentEvent)
        note right
          Event:
          - sessionId
          - segmentId
          - text
          - timestamp
        end note
    end

    Engine --> Service: streaming complete
    deactivate Engine

else 배치 변환 모드
    Service -> Engine: batchTranscribe(audioFile)
    activate Engine

    Engine -> Speech: batchTranscriptionAsync(audioUrl)
    note right
      배치 처리:
      - 전체 파일 업로드
      - 백그라운드 처리
      - Callback URL 제공
    end note

    Speech --> Engine: transcription job ID

    Engine --> Service: job submitted
    deactivate Engine

    Service -> Repository: updateSessionStatus(sessionId, "PROCESSING")
    activate Repository
    Repository -> DB: 세션 상태 업데이트\n(상태='처리중')
    DB --> Repository: updated
    Repository --> Service: updated
    deactivate Repository
end

Service -> Repository: aggregateTranscription(sessionId)
activate Repository
Repository -> DB: 세그먼트 목록 조회\n(세션ID 기준, 타임스탬프 순 정렬)
DB --> Repository: segments
Repository --> Service: ordered segments
deactivate Repository

Service -> Service: mergeSegments(segments)
note right
  세그먼트 병합:
  - 화자별 그룹화
  - 시간 순서 정렬
  - 문장 경계 보정
end note

Service -> Repository: saveTranscription(fullText)
activate Repository
Repository -> DB: 전체 텍스트 저장 및 상태 업데이트\n(전체텍스트, 상태='완료')
DB --> Repository: saved
Repository --> Service: updated session
deactivate Repository

Service -> EventHub: publish(TranscriptionCompletedEvent)
note right
  Event:
  - sessionId
  - meetingId
  - fullText
  - completedAt
end note

Service --> Controller: TranscriptionResponse\n{sessionId, text, segments}
deactivate Service

Controller --> Gateway: 200 OK\n{transcription, metadata}
deactivate Controller

@enduml