Utter

A voice capture app for Apple Watch and iPhone.
Record quick thoughts on your wrist. Transcribe and organize them on your phone.

What it does

Utter is built around a simple idea — the fastest way to capture a thought is to say it out loud. Hold a button on your Apple Watch, speak, and release. That recording gets sent to your iPhone where it's transcribed and automatically sorted into one of four categories based on what you said.

No typing. No unlocking your phone. Just speak and it's saved.

Demo

Watch app	iPhone app
Hold to record a thought, watch the waveform as you speak, then review and send it straight to your iPhone — all from your wrist.	Import the latest watch recording, transcribe it with one tap, review the auto-detected category, and save it to your inbox.

App screens

Watch — intro

Watch — recording & review

Watch — syncing & confirmed

iPhone — full workflow

How it works

1. Record on Apple Watch — Hold the screen to record, release to stop. Review the audio clip before sending.
2. Send to iPhone — Tap the arrow to transfer the recording via WatchConnectivity.
3. Transcribe — On iPhone, tap "Import from Apple Watch" to convert the audio to text using on-device speech recognition.
4. Review and save — Check the transcript, confirm or change the category, then save it to your inbox.
5. Inbox — Browse notes grouped by category. Tap a category to open it, check items off, or delete them.

Categories

	Category	Used for
🟢	Todo	Tasks and actions
🟠	Reminder	Time-based follow-ups
🟣	Idea	Concepts and brainstorms
🔵	Note	Everything else worth keeping

Auto-detection is keyword-based. Phrases like "remind me", "tomorrow", "call" → Reminder. "Buy", "pick up", "do" → Todo. "Idea", "what if", "build" → Idea. Everything else → Note.

Design process

Early sketches exploring different UI directions for the watch app before landing on the final design.

Tech stack

• SwiftUI — iOS + watchOS UI
• WatchConnectivity — Watch → iPhone audio file transfer
• AVFoundation — Audio recording and playback
• Speech framework — On-device transcription
• UserDefaults — Local memo persistence

Requirements

• iOS 17+
• watchOS 10+
• Xcode 15+
• Apple Watch paired to iPhone (for full transfer flow)

Project structure

Utter/
├── Utter/                          # iPhone app
│   ├── ContentView.swift           # Main UI — recording, inbox, review sheet
│   ├── VoiceMemo.swift             # Data model
│   ├── SpeechManager.swift         # Recording + transcription
│   ├── UtterApp.swift              # App entry + PhoneConnectivityManager
│   └── Color+Hex.swift             # Hex color helper
│
└── UtterWatch Watch App/           # Watch app
    ├── ContentView.swift           # Watch UI — idle, recording, review, confirmed
    ├── WatchAudioRecorder.swift    # Audio recording + playback
    └── UtterWatchApp.swift         # App entry + WatchConnectivityManager

Setup

1. Clone the repo

   git clone https://github.com/ritikajoshi/utter.git

2. Open Utter.xcodeproj in Xcode
3. Select your development team in Signing & Capabilities for both the Utter and UtterWatch Watch App targets
4. Build and run on a real device — microphone access and WatchConnectivity require physical hardware for the full flow

Developer notes

• Simulator testing — WatchConnectivity transferFile does not work between simulators. To test transcription in the simulator, manually copy a .m4a file into the app's Documents/utter-recordings/ folder.
• File naming — Watch recordings must be named utter-<UUID>.m4a to be picked up by latestWatchRecordingURL().
• Category detection — Lives in detectedCategory(for:) in ContentView.swift. Reminder keywords are checked first to avoid false Todo matches on phrases like "remember to".
• Persistence — Memos are stored in UserDefaults under the key utter_saved_memos as JSON-encoded [VoiceMemo].

Built by Ritika Joshi · 2026