INVITED PEOPLE ONLY
Sign in to Voice notes
Use the account your host created for you.
Need access or a password reset? Contact your host. Accounts cannot be created on this page.
Audio is sent over an encrypted connection to your host's computer. Speech recognition and summaries run there. Download your notes before leaving.
LAB 13 · LOCAL VOICE
Voice notes
Record a thought or a meeting. Turn speech into searchable knowledge.
Checking local models…
Local setup · no API key
Run these commands in starter_repo, then start the API. The speech model is downloaded once; recording and summaries run locally.
.venv/bin/python -m pip install -e '.[voice]' .venv/bin/python scripts/setup_voice.py ollama pull qwen2.5:3b sh scripts/start_local.sh
Use localhost or HTTPS in a browser with microphone access. If the embedded browser cannot open the microphone, open this same page in Chrome/Safari or import audio below.
Up to 10 minutes per note. Transcript updates in ~8-second chunks plus processing time. Audio stays in this page until you download it; only saved notes enter the document library.
Or import an audio recording
WAV and browser-supported MP3/M4A/OGG. Converted to mono audio in your browser.
Start recording or import audio. You can also paste a transcript below.
Review names, numbers and words near chunk boundaries. Speech recognition can make mistakes. Maximum 24,000 characters.
Draft notes from a local model. Unstated owners and dates should remain unspecified.