Skip to content

📋 Wyoming Event Types

Each Wyoming message is defined by its type. This section documents all standard event types, grouped by functionality.

🔊 Audio Streaming Events

audio-start

Marks the beginning of an audio stream.

{ "type": "audio-start", "data": { "rate": 16000, "width": 2, "channels": 1 } }

Python:

from wyoming.audio import AudioStart

start = AudioStart(rate=16000, width=2, channels=1)
event = start.event()

audio-chunk

Contains a chunk of raw PCM audio.

{ "type": "audio-chunk", "payload_length": 3200 }

Payload: 3200 bytes of PCM audio.

Python:

from wyoming.audio import AudioChunk

chunk = AudioChunk(audio=b"...").event()

audio-stop

Marks the end of an audio stream.

{ "type": "audio-stop" }

Python:

from wyoming.audio import AudioStop

stop = AudioStop().event()

🧠 Speech Recognition (ASR)

transcribe

Start transcribing an audio stream.

{ "type": "transcribe", "data": { "name": "whisper", "language": "en" } }

transcript

Returns recognized text.

{ "type": "transcript", "data": { "text": "hello world", "confidence": 0.95 } }

đŸ—Ŗī¸ Text to Speech (TTS)

synthesize

Request speech audio from text.

{ "type": "synthesize", "data": { "text": "Hello!", "voice": { "language": "en" } } }

Response: stream of audio-start, audio-chunk, audio-stop

đŸŽ¯ Wake Word Detection

detect

Begin wakeword detection (streaming)

{ "type": "detect", "data": { "names": ["hey_lucy"] } }

detection

Wake word was detected.

{ "type": "detection", "data": { "name": "hey_lucy" } }

not-detected

Audio ended without a detection.

🎧 Voice Activity Detection (VAD)

voice-started

Detected speech start

voice-stopped

Detected speech end

🤖 Intent Recognition

recognize

Request to identify user intent from text.

intent

Returns structured result (intent name, entities)

not-recognized

Fallback when nothing matched.

🛠ī¸ Intent Handling

handled

Indicates successful command handling

not-handled

Indicates failure or no response

ℹī¸ Service Discovery

describe

Client asks for capabilities

info

Server responds with available ASR, TTS, Wake, etc.

⏱ī¸ Timer Events (see next file)

  • timer-started, timer-updated, timer-finished, timer-cancelled ➡ī¸ Full details in timers.md

❌ Errors

error

Used to indicate a failure:

{ "type": "error", "data": { "message": "Invalid format" } }

🛰ī¸ Satellite Events

For coordinating remote microphone devices.

  • run-pipeline
  • satellite-connected
  • streaming-started / streaming-stopped

(Details in pipeline.md)