Speech Pro sources and licenses
McNova Speech Pro records and decodes audio in the browser. Its transcription engine uses the bundled Transformers.js (Apache-2.0) and Xenova/whisper-tiny (Apache-2.0; ONNX conversion of OpenAI Whisper Tiny). Model files are fetched from Hugging Face on first use and cached by the browser. Audio is processed locally and is not uploaded by this app.
The small model has limited accuracy, especially for specialist vocabulary and some languages. It is not a medical transcription service. Users must review names, numbers, and meaning.
Other voice projects evaluated
Vosk API and Vosk Server support local/server speech recognition but need a dedicated runtime. Silero VAD detects speech segments but does not transcribe. Dograh, Pipecat, AgentScope and BentoVoiceAgent are voice-agent frameworks needing backend infrastructure and service review. Realtime Voice Conversion and Voice-Converter-CycleGAN transform voices; they are not speech-to-text engines. None of these projects' source code is bundled into the current Speech Pro build.