Audio file · Interface preview with sample dataText to speech · Interface preview with sample data
Run locally, set up in Freehand
Install NeMo for Nemotron or Parakeet transcription, with optional MagpieTTS speech in the same runtime. Add llama.cpp with S1-mini for English cleanup. Windows also supports managed whisper.cpp.
Use an existing server on this computer, on your network, or with a hosted provider. Choose each task’s connection independently, or share a server that supports several tasks.
Choose connections independently for Voice, audio files, cleanup, and speech generation. Keep them together in one workspace with task sidebars and a shared output panel.
Dictate into an app with a shortcut, or select an audio file and copy the result. Use completed transcription or live dictation with a qualified model.
Process completed transcripts with a separate chat model. Use custom instructions or S1-mini’s English controls; raw text remains the fallback if cleanup fails.
Your installed release may have fewer profiles than those listed here. Capabilities depend on the server version and model profile →
Native desktop integration
Global shortcuts, microphone capture, and a passive status overlay. Dictation is delivered only while the original app and window remain valid and focused; otherwise the result stays ready to copy.