VoiceKeeper

Speak it. Keep it. Organized.

Try the web demo →  ·  Mobile app available on Android in the release and coming soon to the App Store on iOS.

---

VoiceKeeper is a voice-first note-taking app. Tap to record, and your voice is instantly transcribed and auto-formatted into the structure you want — bullet lists, paragraphs, action items, meeting notes, or a fully custom template you define by example.

Built with React Native (Expo), Supabase, and OpenAI. Runs on iOS, Android, and Web.

---

Features

• One-tap recording with live waveform feedback
• Auto-transcription powered by OpenAI Whisper
• Auto-formatting into your chosen structure (5 built-in formats + custom templates)
• Custom instructions — set persistent rules for tone, language, length, etc.
• Card-based grid UI with full-text search (Google Keep style)
• Cross-platform — iOS, Android, and Web from a single codebase

Supported formats: Bullet List · Paragraph · Action Items · Meeting Notes · Custom Template

---

How it works

You speak into your phone
        │
        ▼
Supabase Edge Function
  ├── OpenAI Whisper  →  transcription
  └── OpenAI GPT      →  formatted note + title
        │
        ▼
Saved to Supabase Postgres (with full-text search)

All OpenAI calls go through a Supabase Edge Function — your API key is never exposed to the client.

---

Tech stack

Layer	Technology
Frontend	React Native (Expo) — iOS, Android, Web
Speech-to-text	OpenAI Whisper API
Auto-formatting	OpenAI GPT
Backend	Supabase (Auth + Postgres + Storage + Edge Functions)
Deployment	Vercel (web)

---

Self-hosting / Development setup

VoiceKeeper is open source under the AGPL-3.0 license. You can run your own instance — you just need to bring your own API keys (BYOK).

Prerequisites

• Node.js 18+
• A Supabase account (free tier works)
• An OpenAI API key
• (Optional) Vercel account for web deployment

1. Clone and install

git clone https://github.com/End2EndAI/voicekeeper.git
cd voicekeeper
npm install

2. Set up Supabase

1. Create a new project at supabase.com/dashboard
2. Go to Authentication > Providers and enable Email (disable "Confirm email" for faster testing)
3. Note your Project URL and anon key from Settings > API

3. Configure environment variables

cp .env.example .env

Edit .env with your Supabase credentials:

EXPO_PUBLIC_SUPABASE_URL=https://your-project-ref.supabase.co
EXPO_PUBLIC_SUPABASE_ANON_KEY=your-anon-key-here

Also create supabase/.env.local for local Edge Function development:

OPENAI_API_KEY=sk-your-openai-key

4. Set up the database and Edge Functions

# Install Supabase CLI if needed
brew install supabase/tap/supabase   # macOS
# or: npm install -g supabase

# Set your project ID in the Supabase config
# Replace YOUR_PROJECT_REF with your project reference from Settings > General
sed -i '' 's/project_id = ""/project_id = "YOUR_PROJECT_REF"/' supabase/config.toml

# Login and link your project
supabase login
supabase link --project-ref YOUR_PROJECT_REF

# Push all migrations (tables, RLS policies, triggers, storage bucket)
supabase db push

# Set your OpenAI API key as a Supabase secret
supabase secrets set OPENAI_API_KEY=sk-your-openai-key

# Deploy the Edge Functions
supabase functions deploy process-recording
supabase functions deploy delete-account

Note: Do not commit your project_id to the repo — supabase/config.toml uses an empty project_id by default so each self-hoster sets their own.

5. Run locally

# Web
npx expo start --web

# iOS (requires Xcode)
npx expo start --ios

# Android (requires Android Studio)
npx expo start --android

Note: Expo automatically loads the .env file. The EXPO_PUBLIC_ prefix is required for client-side access.

---

Build mobile apps (EAS)

Requires an Expo account and the EAS CLI:

npm install -g eas-cli
eas login

Android APK (preview — direct install)

eas build --platform android --profile preview

Produces a downloadable .apk you can install directly on any Android device. No Play Store needed.

Production build (Play Store / App Store)

# Android (AAB for Play Store)
eas build --platform android --profile production

# iOS (for App Store)
eas build --platform ios --profile production

---

Deploy to Vercel (web)

1. Push your repo to GitHub
2. Import the repository at vercel.com/new
3. Add environment variables in the Vercel dashboard:
   • EXPO_PUBLIC_SUPABASE_URL
   • EXPO_PUBLIC_SUPABASE_ANON_KEY
4. Deploy — Vercel auto-detects the vercel.json config

The included vercel.json configures the build command (npx expo export --platform web), output directory (dist), and SPA routing.

---

Project structure

voicekeeper/
├── app/                          # Expo Router screens
│   ├── _layout.tsx               # Root layout (auth guard, providers)
│   ├── index.tsx                 # Home (note grid + search)
│   ├── login.tsx                 # Login / signup
│   ├── record.tsx                # Recording (modal)
│   ├── preview.tsx               # Note preview after processing
│   ├── settings.tsx              # User preferences
│   └── note/[id].tsx             # Note detail / edit
├── components/                   # Reusable UI components
├── contexts/                     # React Context (Auth, Notes, Preferences)
├── services/                     # Business logic (Supabase, processing, GDPR)
├── types/                        # TypeScript type definitions
├── constants/                    # Colors, format options
├── utils/                        # Helpers (alerts, title generation)
├── supabase/
│   ├── migrations/               # SQL migrations (6 files)
│   └── functions/
│       ├── process-recording/    # Edge Function (Whisper + GPT)
│       └── delete-account/       # Edge Function (GDPR account deletion)
├── .env.example                  # Environment template
├── PRIVACY_POLICY.md             # Privacy policy (GDPR)
├── SECURITY.md                   # Security vulnerability reporting
├── app.json                      # Expo config
├── vercel.json                   # Vercel deployment config
└── package.json

---

Database schema

notes — voice notes with title, formatted text, raw transcription, format type, and optional audio URL. Row-Level Security ensures users can only access their own notes.

user_preferences — per-user settings: default format, custom template example, and custom instructions. Auto-created on signup via a database trigger.

audit_log — GDPR accountability log tracking account creation, deletion, and data export requests.

recordings bucket — private Supabase Storage bucket for audio files, with per-user RLS.

The database also includes server-side functions for GDPR compliance: export_user_data() (data portability) and account deletion with full storage cleanup.

---

Edge Function: process-recording

The Edge Function acts as a secure proxy between the app and OpenAI:

1. Validates the user's Supabase JWT
2. Sends audio to OpenAI Whisper for transcription
3. Sends transcription to GPT for formatting (with the user's chosen format + custom instructions)
4. Returns { transcription, formatted_text, title } to the app

This design keeps API keys server-side and centralizes all AI logic for easy iteration.

---

About

VoiceKeeper is developed by End2EndAI. The web app is live and mobile apps are coming soon to the App Store and Play Store. The free tier allows 5 notes per day (limited due to OpenAI API costs). There is no paid subscription yet — one is coming soon on Android and iOS apps.

Developers are welcome to self-host using their own API keys under the AGPL-3.0 license. See CONTRIBUTING.md for how to get involved.

---

License

This project is licensed under the GNU Affero General Public License v3.0.

You are free to use, modify, and distribute this software under the terms of the AGPL-3.0. If you deploy a modified version as a network service, you must make your source code available to users of that service.

See LICENSE for the full text.