DentaNote: dental clinical notes with local Gemma, no audio leaves the operatory

Description and use

DentaNote cuts dental note creation time to under a minute by locally transcribing audio and extracting structured data, keeping all patient data on the device and allowing clinician review.

Problem

Clinicians lose time building structured dental notes after dictating audio, and existing solutions risk patient data leaving the machine.

Outcome

DentaNote reduces note creation time from ~10 min to under 1 min per appointment, keeps all data local, and allows model swapping and validation to mitigate hallucinations.

Workflow

• Record patient visit audio
• Run dentanote run visit.wav to transcribe with Whisper and extract structured data with Gemma via Ollama
• Validate JSON with Pydantic
• Render note with deterministic template
• Review uncertain_items flagged for clinician verification
• Export or edit note in Streamlit UI

Evidence

audio in, structured clinical note out, running 100% locally, with not a single byte leaving my machine.

Source

The full pipeline — transcription with Whisper, extraction with Gemma via Ollama, schema validation, and rendering — took 11.2 seconds to turn a 30-second audio clip into a complete note (transcription: 7.7 s; extraction: 2.1 s; rendering is a deterministic template and doesn't register in the timing).

Source

Both produced valid JSON on 100% of first attempts. Both failed on the same case (urgencia_dor predicted tooth 37 instead of 47 — the model "picked" a neighbor on the arch).

Source

The small Whisper model on CPU transcribes ~3.8× faster than real time. So a 3-minute dictation becomes text in about 45 s, plus ~2 s of extraction — under a minute for a note I would have dictated entirely.

Source