ovos-tts-plugin-edge-tts 0.3.2a1

TTS plugin for OVOS based on Edge-TTS

Configuration

Configuration parameters to add in the user config: ~/.config/mycroft/mycroft.conf

"tts": {
    "module": "ovos-tts-plugin-edge-tts",
    "ovos-tts-plugin-edge-tts": {
        "voice": "en-US-AriaNeural", 
        // =100% speed; use "+50%" for 150%, or "+100%" for 200% speech rate etc for adjusting speed
        "rate": "+0%" 
    }
}

See for voices: list of voices available in Edge TTS.

Also a lot of the latest generation multilingual voices (from March 2024) work with this plug-in. For instance: "voice": "en-US-EmmaMultilingualNeural".

Installation

pip install ovos-tts-plugin-edge-tts

Docker (ovos-tts-server)

A container image runs the plugin as an ovos-tts-server (ElevenLabs-compatible API), built and pushed to GHCR by CI on every push to dev/master:

docker run -p 9666:9666 ghcr.io/openvoiceos/ovos-tts-plugin-edge-tts:latest
curl "http://localhost:9666/synthesize/hello%20world?lang=en-US" --output hello.wav

The served voice is baked in via the EDGE_VOICE build arg (default en-US-AriaNeural); rebuild to change it, e.g. docker build --build-arg EDGE_VOICE=ar-SA-HamedNeural -t edge-tts .. See the bundled docker-compose.yml. Edge streams from Microsoft's cloud, so the container needs network access.