Freespeech

A native macOS video journaling app. Record short videos, get automatic transcriptions, and use AI to improve your language skills — all stored locally on your Mac.

---

What it does

Record → Transcribe → Analyze.

1. Hit the camera button, speak, stop.
2. Run Analyze to get a high-accuracy transcript (ElevenLabs), pronunciation scoring (Azure), grammar corrections, and AI language coaching (Claude) — all in one step.

All data stays on your Mac under ~/Documents/Freespeech/.

---

Requirements

• macOS 14.0 or later
• Xcode 15 or later
• API keys to unlock AI features (see Setup)

---

Setup

1. Clone and open

git clone https://github.com/gabruini/freespeech.git
cd freespeech
open freespeech.xcodeproj

2. Build and run

Select the freespeech scheme and press ⌘R. Or from the terminal:

xcodebuild -project freespeech.xcodeproj -scheme freespeech -configuration Debug build

3. Grant permissions

On first launch the app requests:

• Camera — to record video
• Microphone — to capture audio
• Speech Recognition — for live transcription

4. Add API keys

Open Settings (gear icon, bottom-right) or go through the onboarding flow. Keys are stored in the macOS Keychain.

Anthropic (Claude)

Used for AI language coaching and video analysis.

• Get a key at console.anthropic.com

Microsoft Azure Speech

Used for pronunciation assessment, TTS synthesis, and word-timing extraction.

• Create a Speech resource in the Azure Portal
• You need the Key and the Region (e.g. eastus)

ElevenLabs

Used for high-accuracy audio transcription (scribe_v2 model).

• Get a key at elevenlabs.io

Note: Recording work without any API keys. The Analyze pipeline and Language Coach require all three keys.

---

License

MIT — see LICENSE.