TTS Options for OpenVoiceOS
OpenVoiceOS, or OVOS, provides a flexible plugin system for text-to-speech (TTS) synthesis. This system allows you to choose the TTS engine that best suits your needs, whether you prioritize voice quality, speed, privacy, or language support.
Running Plugins Directly
OVOS maintains several TTS plugins that can run directly on your voice assistant hardware. Here are the currently maintained plugins, categorized by their primary use case:
| Hardware/Environment | Model | GitHub Link | Notes |
|---|---|---|---|
| GPU-Optimized | Coqui TTS | ovos-tts-plugin-coqui | Local TTS using Coqui models, best performance with GPU but some models work on CPU |
| CPU-Capable | Edge TTS | ovos-tts-plugin-edge-tts | Microsoft Edge TTS voices, requires internet connection but provides fast, high-quality output and streaming capability |
| CPU-Capable | Piper | ovos-tts-plugin-piper | Local TTS with multiple voice options, good performance on modern hardware (optimized for Raspberry Pi) |
| CPU-Capable | Mimic | ovos-tts-plugin-mimic | Classic Mycroft offline TTS engine, lightweight but robotic sounding |
| CPU-Capable | ESpeak-NG | ovos-tts-plugin-espeakNG | Lightweight offline TTS, supports many languages but robotic sounding |
| CPU-Capable | Pico | ovos-tts-plugin-pico | Very lightweight offline TTS engine |
| API-Based/Cloud | Azure | ovos-tts-plugin-azure | Microsoft Azure Cognitive Services TTS, paid service with high quality but no guarantee of privacy |
| API-Based/Cloud | Polly | ovos-tts-plugin-polly | Amazon Polly TTS service, paid service with high quality but no guarantee of privacy |
| API-Based/Cloud | MaryTTS | ovos-tts-plugin-marytts | Connect to MaryTTS server, can be self-hosted, considered outdated and not well maintained |
| Language-Specific | Nòs | ovos-tts-plugin-nos | Galician language TTS |
| Language-Specific | Cotovia | ovos-tts-plugin-cotovia | Alternative Galician language TTS |
| Language-Specific | Matxa | ovos-tts-plugin-matxa-multispeaker-cat | Catalan language multi-speaker TTS |
| Novelty | BeepSpeak | ovos-tts-plugin-beepspeak | R2D2-style robotic sounds, works well but requires subtitles to be understood |
| Novelty | SAM | ovos-tts-plugin-SAM | Software Automatic Mouth, retro-style speech synthesis, too digitized for regular use |
Note: While Mimic2 and Mimic3-server plugins exist in archived form, they are no longer supported. Users are recommended to use Piper or other maintained alternatives for better performance and ongoing support.
OVOS-TTS-Server
If you want to run your TTS engine on a different machine from your voice assistant, you can use the OVOS TTS Server. This server provides a consistent API interface for any supported OVOS TTS plugin, allowing you to:
- Run TTS on more powerful hardware separate from your voice assistant
- Share a single TTS server among multiple voice assistants
- Cache TTS results for improved performance
- Use the REST API for custom applications beyond voice assistants
The server can be particularly useful when running resource-intensive TTS models that might not perform well on lower-powered devices like a Raspberry Pi. I run the Coqui TTS plugin on a separate server to take advantage of its GPU acceleration and provide mid-high-quality TTS to all my voice assistants. I also have Piper running as a remote backup server.
Conclusion
Questions? Comments? Feedback? Let me know on the Open Conversational AI Forums or OVOS support chat on Matrix. I’m available to help and so is the rest of the community.