refactor: define contrato de VAD e eventos de fala

README.md

@@ -390,6 +390,7 @@ src/voice_assistant/
   audio_gateway/app/            WebSocket endpoint and protocol
   application/                  VoiceSession and transport-independent events
   domain/                       Session state and domain rules
+  vad/                          VAD port and speech boundaries; runtime integration pending
   audio/                        Canonical audio models
   stt/                          STT contract, mock and NVIDIA adapter
   agent/                        Agent contract and mock adapter
@@ -415,3 +416,21 @@ tools/                          Integration and smoke scripts
 **Status:** V1 development baseline · Python 3.12+ · WebSocket protocol 1.0
 
 
+
+
+## Backend VAD contract (integration pending)
+
+`vad/adapter.py` defines one stateful detector per continuous input stream.
+Its input is the canonical PCM S16LE, mono, 16 kHz audio format. Arbitrary
+complete PCM frames are buffered into model-specific windows by the adapter.
+`vad/models.py` defines `SpeechStarted` and `SpeechEnded`; their sample offsets
+refer to input audio since initialization or reset, with inclusive starts
+and exclusive ends. Detection confirmation may arrive later than its audio
+boundary. Model padding never extends the input timeline.
+
+`process()` returns ordered boundaries, `flush()` handles normal end of input,
+`reset()` discards detection state and restarts the timeline, and `aclose()`
+releases resources without emitting new speech events. Calls are serialized.
+These contracts do not start turns, select providers, or carry transport IDs.
+The Silero implementation, application integration, and browser VAD replacement
+are subsequent tasks. This step leaves the active conversation flow unchanged.