Features
On-device ONNX inference for low latency
Multilingual speech synthesis
Lightweight model runs on consumer GPU/CPU
Privacy-preserving local deployment
Interactive Hugging Face Space demo
Downloadable model weights for custom integration
Expressive output at faster-than-realtime speed
No cloud dependency or API calls
Configurable inference via ONNX runtime options
Supports batch processing for multiple texts
Real-time speech-to-text in 60+ languages
Text-to-speech with hallucination-free output
Real-time speech translation across 3,600 language pairs
Sub-200ms streaming latency
Multi-speaker diarization and language detection
Code-switching support mid-sentence
Audio never stored, processed in real-time
SOC 2 Type 2, ISO 27001, HIPAA, GDPR compliant
In-region processing for data residency
Voice cloning from few seconds of audio
v5 improvements in accuracy, endpointing, speaker separation
Unified API for STT, TTS, and translation
SDKs for Python, Node, Web, React, React Native
Optimized for high-noise environments
Token-based pricing with bundled features
