Speech workspace
Run speech recognition or generate speech with your installed models.
Audio input
WAV, MP3, M4A or FLAC · up to 25 MiB · max length 10 min
Advanced settings
Transcription
Your transcription will appear here.
0 characters
61 characters
Advanced settings
Generated speech
Models
Manage installed models and the local model catalog.
Loading models…
Smart Router
Inspect and update language-aware model routing.
Configured call flow
Choose a task and language to inspect its ordered model candidates.
Routing rule
Adjust model selection or priority for this existing rule.
All changes saved
API integration
Use these examples to connect an agent or application.
Base URL
STT · Transcribe audio
Multipart upload. Select an installed model in the workspace to update this example.
TTS · Generate speech
JSON request returning MP3 by default; WAV is available.
About
Local speech recognition and synthesis, with language-aware model routing.
SmartVoice is a local speech-to-text (STT) and text-to-speech (TTS) service. It offers both through one OpenAI-style audio API, with models installed and run on your machine. Smart routing chooses an installed model based on the request language. Inference runs locally on CPU with no per-request cloud fee; model files need to be downloaded during setup.