Medical Audio Pipeline — Test Interface
Test system only — not for clinical use. Outputs are not validated medical advice.
Audio input
Upload an audio file:
— or record directly —
Start Recording
Device
Compute device:
CUDA (GPU)
CPU
Pipeline components
ASR (speech-to-text transcription)
Model:
MedASR (medical, gated)
NVIDIA Parakeet RNNT 0.6B (general, open)
pyannote (speaker diarization)
MedGemma (clinical note generation — requires MedASR)
Run Pipeline
Execution Time
Results
Transcript (ASR)
Speaker Diarization (pyannote)
Speaker-Labeled Transcript (ASR + pyannote)
Clinical Note (MedGemma)