Relaxation Station

Free guided meditations and soothing soundscapes. Everything is built into the app, so it works fully offline, even in airplane mode.

• 15 guided meditations in five categories: Sleep, Stress Relief, Breathing, Body Scan and Focus. Each category has a 3, 5 and 10 minute session.
• 12 soundscapes: Gentle Rain, Wind Chimes, Ocean Waves, Music Box Lullaby, Heartbeat, Crackling Fire, Guitar Lullaby, Box Fan, Brown Noise, Relaxing Bells, Car Ride and Gentle Shush. Each plays on a seamless loop until you stop it, like a sound machine.
• Keeps playing with the screen locked, with lock screen and notification controls.
• No network, no accounts, no data collection.

Built with Expo SDK 58, React Native 0.88 (New Architecture), Expo Router, the React Compiler, TypeScript and expo-audio.

Getting started

Requirements: Node 24 LTS. Simulator builds also need Xcode and CocoaPods.

npm install
npx expo start --go       # Expo Go on your phone (scan the QR code)
npx expo run:ios          # development build on the iOS Simulator

Background playback and lock screen controls need a development or production build. They are native settings that Expo Go can't change.

Scripts

Command	Purpose
npm test	Unit tests (jest-expo)
npm run typecheck	TypeScript
npm run lint	ESLint (eslint-config-expo)
npm run generate:audio	Regenerate the meditations and soundscapes (see below)
npm run generate:brand	Re-render the app icon, splash, Play icon and website logo
node scripts/generate-play-assets.mjs	Re-render the Google Play feature graphic and screenshots

How the audio is made

All audio is generated on a Mac and committed to assets/audio/. The app never downloads anything.

• Catalog: content/catalog.json lists every category, meditation and soundscape. The app and the generator both read it.
• Scripts: content/meditations.txt holds the meditation scripts. Each ## <id> <bed> header names the background bed (warm, guitar, ocean, piano or flute). Lines are spoken, ~ adds a longer pause and [N] adds an exact pause of N seconds.
• Generator: audio-tools/generate_audio.py:
  • speaks each line with the open-source Qwen3-TTS text-to-speech model (Apache 2.0). The narrator is a gentle female voice. Qwen3-TTS VoiceDesign created her once from a written description, and her clip is kept in audio-tools/narrator-reference.wav. The Base model clones that clip for every sentence. The generator speeds each sentence up by 8% without changing its pitch, then softens the voice slightly
  • spreads the pauses so every session is exactly 3, 5 or 10 minutes
  • mixes the voice over the session's background bed, which dips slightly while the narrator speaks, with a singing bowl at the start and end
  • synthesizes the soundscapes from filtered noise and tones, and crossfades each one's end into its start so it loops seamlessly
  • encodes everything to AAC with macOS's afconvert
• Soundscapes: the lullaby is Brahms' Lullaby (1868, public domain) on a synthesized music box. The lullaby and bells are built as exact circles, so their notes and reverb wrap around the loop point.
• Audio map: the generator also writes src/content/audio-files.ts, the static require() map that bundles the files.

One-time generator setup. It needs Python 3.12 and an Apple Silicon Mac. The virtual environment lives in the git-ignored .tts/ folder. The Qwen3-TTS models, about 9 GB, download to the Hugging Face cache on the first run:

python3.12 -m venv .tts/venv
.tts/venv/bin/pip install qwen-tts soundfile scipy audiotsm
npm run generate:audio                 # everything
npm run generate:audio -- meditations sleep-3   # one meditation
npm run generate:audio -- sounds rain                   # one soundscape
npm run generate:audio -- voice         # redesign the narrator from VOICE_DESCRIPTION

Project layout

content/                  Catalog and meditation scripts (source of truth for the audio)
audio-tools/              Audio generator (Python, run on a Mac only)
assets/audio/             Generated meditations and soundscapes (AAC)
src/
  app/                    Expo Router routes: (tabs)/ Meditate, Sounds, About; player.tsx modal
  components/             Play/pause icon, sound icons, button; brand/ logo, hero, splash
  content/                Typed catalog and the generated audio map
  context/                Playback: one shared player, background audio, lock screen controls
  lib/                    Small pure helpers with tests
patches/                  Fixes to dependencies, applied by `npm install`

patches/react-native-screens+4.30.0.patch makes the tab bar appear when the app launches on an unfolded iPhone Duo. It is the fix from react-native-screens#4990. npm install applies it through postinstall. Delete the patch after upgrading to a react-native-screens release that includes the fix.

Releasing

• Website: docs/ holds the public home page, privacy policy and support page, served by GitHub Pages.
• App Store: store/README.md has the build command, listing copy, App Privacy and age rating answers, App Review notes, and the iPhone and iPhone Duo screenshots.
• Google Play: store/android/README.md has the Play Console steps, including the foreground service declaration that background audio needs.

npx eas-cli@latest build --platform ios --profile production --auto-submit
npx eas-cli@latest build --platform android --profile production

License

Copyright (c) 2026 Preston Davies. All rights reserved. See LICENSE.