Voice Preferences

Customize your speech-to-text model and text-to-speech voice. These settings apply to all your recording and read-aloud features.

Speech-to-Text Model (Transcription)

Choose the AI model for transcribing your audio recordings. More accurate models may be slower.

Text-to-Speech Voice (Read Aloud)

Choose the voice for the "Read Aloud" feature. Preview available after selection.

Browser Transcription (Local Whisper)

Transcribes audio entirely in your browser — no audio sent to any server. Powered by OpenAI Whisper running in WebAssembly. Model is downloaded once and cached locally.

Enable browser transcription: Off

Model:

When enabled, overrides server transcription. Falls back to server if browser transcription fails.

Network/Firewall Issue: If model download fails, check that cdn.jsdelivr.net and huggingface.co are not blocked. Server transcription will be used as fallback.

Real-Time Streaming Transcription

Privacy Warning: Uses your browser's built-in speech recognition, which may send audio to cloud servers (Chrome/Edge send to Google). Only enable if you accept this trade-off for real-time transcription.

See words appear as you speak (streaming). Overrides browser and server transcription when enabled. Not HIPAA-compliant in most browsers.

Enable real-time streaming: Off

Current Browser:

Detecting...

Trade-off: Immediate transcription vs. privacy. For maximum privacy, use Browser Whisper (offline batch mode) or Server transcription with HIPAA-eligible provider.

Two-Factor Authentication

Status: Loading...

Nextcloud Integration

Export generated documents to your Nextcloud.

Not connected

Nextcloud URL

Username

App Password Go to Nextcloud → Settings → Security → Create new app password

My Templates

Save reusable templates for physical exam, ROS, encounter format, etc. The AI will use these when generating notes. You can reference them by saying "use my normal physical exam" in dictation.

Loading templates...

Documents

Upload and manage documents via S3 storage (PDF, images, Word docs, text files). Max 10 MB per file.

Loading...

AI Learning (Corrections)

The AI automatically learns from your edits. When you modify AI-generated text and save, corrections are stored here and applied to future notes. Latest 20 per section.

Loading corrections...

Audio Backups

Recordings are automatically backed up locally and kept for 24 hours. Retry transcription if it failed.

Loading...

Saved Encounters

Encounters are automatically deleted after 7 days per site policy.

Loading...

Compliance & Usage

AWS Bedrock is available with a Business Associate Agreement (BAA) for HIPAA-eligible workloads.

✅ All connections use HTTPS/TLS encryption
✅ Authentication with optional 2FA
✅ No patient data stored on server beyond session
✅ AWS Bedrock supports BAA for HIPAA compliance
✅ Azure OpenAI supports BAA for HIPAA compliance

Important: Check with your institution's guidelines and policies before use. This tool is not intended for production clinical use without proper organizational authorization and provider BAAs in place. Use with caution.