Local TTS (Coqui-TTS Stark Edition)
High-performance Text-to-Speech skill using a local Coqui-TTS server. Part of the Stark Fleet AI infrastructure.
Server Configuration
This skill connects to a local server at http://sphinx.espuny.net:5100 (OpenAI-compatible API).
Available Voices
- JARVIS: Calm, technical voice based on Paul Bettany's clone. (Default)
- HAL: Retro-tech voice inspired by HAL9000.
Usage
{baseDir}/scripts/say.sh "Hello sir, the analysis is complete." --voice JARVIS --lang en
Performance
- Latency: ~250ms for typical agent responses.
- Hardware: Optimized for NVIDIA RTX 5090 acceleration.
Scan to join WeChat group