@startuml
!theme mono

title 음성녹음 및 화자 식별 내부 시퀀스 (UFR-STT-010)

participant "API Gateway<<E>>" as Gateway
participant "SttController" as Controller
participant "SttService" as Service
participant "AudioStreamManager" as StreamManager
participant "SpeakerIdentifier" as Speaker
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/start-recording\n{meetingId, userId}
activate Controller

Controller -> Service: startRecording(meetingId, userId)
activate Service

Service -> Repository: findMeetingById(meetingId)
activate Repository
Repository -> DB: 회의 정보 조회\n(회의ID 기준)
DB --> Repository: meeting data
Repository --> Service: Meeting entity
deactivate Repository

Service -> StreamManager: initializeStream(meetingId)
activate StreamManager
StreamManager -> Speech: createRecognizer()\n(Azure Speech API)
note right
  Azure Speech 설정:
  - Language: ko-KR
  - Format: PCM 16kHz
  - Continuous recognition
end note
Speech --> StreamManager: recognizer instance
StreamManager --> Service: stream session
deactivate StreamManager

Service -> Speaker: identifySpeaker(audioFrame)
activate Speaker
Speaker -> Speech: analyzeSpeakerProfile()\n(Speaker Recognition API)
note right
  화자 식별:
  - Voice signature 생성
  - 기존 프로필과 매칭
  - 신규 화자 자동 등록
end note
Speech --> Speaker: speakerId
Speaker --> Service: speaker info
deactivate Speaker

Service -> Repository: saveSttSession(session)
activate Repository
Repository -> DB: STT 세션 저장\n(회의ID, 상태, 시작일시)
DB --> Repository: session saved
Repository --> Service: SttSession entity
deactivate Repository

Service -> EventHub: publish(SttStartedEvent)
note right
  Event:
  - meetingId
  - sessionId
  - startedAt
end note

Service --> Controller: RecordingStartResponse\n{sessionId, status}
deactivate Service

Controller --> Gateway: 200 OK\n{sessionId, streamUrl}
deactivate Controller

== 음성 스트리밍 처리 ==

Gateway -> Controller: WebSocket /ws/stt/{sessionId}\n[audio stream]
activate Controller

Controller -> Service: processAudioStream(sessionId, audioData)
activate Service

Service -> StreamManager: streamAudio(audioData)
activate StreamManager

StreamManager -> Speech: recognizeAsync(audioData)
Speech --> StreamManager: partial result
note right
  실시간 인식:
  - Partial text
  - Confidence score
  - Timestamp
end note

StreamManager --> Service: recognized text
deactivate StreamManager

Service -> Speaker: updateSpeakerMapping(text, timestamp)
activate Speaker
Speaker --> Service: speaker segment
deactivate Speaker

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

Service --> Controller: streaming response
deactivate Service

Controller --> Gateway: WebSocket message\n{text, speaker, timestamp}
deactivate Controller

@enduml