ovos-audio

ovos-audio is the audio daemon for OVOS. It handles all speech output (TTS) and legacy media playback, running as a background service that connects to the OVOS MessageBus.

---

Responsibilities

• TTS synthesis: receives speak bus events, runs them through DialogTransformersService, synthesizes audio via the configured TTS plugin, and queues it for playback
• TTS playback: PlaybackThread dequeues synthesized audio, runs it through TTSTransformersService, plays it, and fires lifecycle bus events (start/end of speech, listen trigger)
• Fallback TTS: if the primary TTS plugin fails, falls back to a separately configured plugin
• Legacy audio service: optionally loads AudioService backends (VLC, MPV, etc.) for media playback via mycroft.audio.service.* bus events
• Plugin introspection: responds to opm.tts.query and opm.g2p.query with installed plugin metadata for external UIs

---

Architecture

MessageBus
    │
    ▼
PlaybackService (Thread)
    ├── DialogTransformersService   ← rewrites dialog text before TTS
    ├── TTSFactory → TTS plugin     ← synthesizes audio
    ├── PlaybackThread (Thread)
    │       └── TTSTransformersService  ← post-processes wav files
    └── AudioService (optional, legacy)
            └── AudioBackend plugins (VLC, MPV, ...)

---

Navigation

Document	Contents
playback-service.md	PlaybackService — main daemon, TTS lifecycle, fallback, bus events
tts.md	TTSFactory, PlaybackThread — synthesis queue and playback lifecycle
audio-service.md	AudioService — legacy audio backends, playback control, audio ducking
transformers.md	DialogTransformersService, TTSTransformersService — plugin pipeline

---

Quick Start

The service is normally started by ovos-core or run standalone:

ovos-audio

Or in Python:

from ovos_audio.service import PlaybackService

service = PlaybackService()
service.start()
service.join()

---

Package Layout

ovos_audio/
├── service.py       # PlaybackService — main daemon class
├── tts.py           # TTSFactory wrapper
├── playback.py      # PlaybackThread — dequeue and play TTS audio
├── audio.py         # AudioService — legacy audio backend manager
├── transformers.py  # DialogTransformersService, TTSTransformersService
├── utils.py         # require_default_session decorator, report_timing
└── __main__.py      # Entry point: ovos-audio CLI

---

Configuration

All configuration is read from mycroft.conf via ovos-config.

Key	Default	Description
tts.module	—	TTS plugin entry point name
tts.fallback_module	—	Fallback TTS plugin if primary fails
tts.preload_fallback	true	Load fallback TTS eagerly at startup
tts.pulse_duck	false	Use PulseAudio modules for audio ducking instead of bus events
tts.ocp_cork	false	Cork (pause) OCP media during speech
tts.ocp_duck	false	Duck (lower volume) OCP media during speech
enable_old_audioservice	true	Enable legacy AudioService backends
disable_ocp	false	Disable built-in OCP backend inside AudioService
g2p.module	—	Grapheme-to-Phoneme plugin for mouth/viseme animations
dialog_transformers	{}	Dialog transformer plugin configs
tts_transformers	{}	TTS (wav) transformer plugin configs
Audio.default-backend	—	Preferred legacy audio backend name
Audio.backends	{}	Legacy audio backend configs