VoiceInsights

A privacy-first Android app that continuously records ambient audio 24/7 in the background, compresses it into lightweight .m4a chunks, and uploads them to Google Drive for offline AI transcription and insights.

Tech Stack

Component	Technology
Language	Kotlin
UI	Jetpack Compose + Material 3
Background	Android Foreground Service
Audio	MediaRecorder (AAC, 64kbps, Mono)
Cloud	Google Drive REST API
Architecture	MVVM

Features

• 24/7 Background Recording via Android Foreground Service with persistent notification
• Compressed Audio Chunks — 10-minute .m4a files (~4MB each) instead of raw PCM (~100MB each)
• Google Drive Sync — Automatic upload of audio chunks (in progress)
• Phone Call Import — Auto-detect and upload call recordings from native dialer or Cube ACR (planned)

Permissions

• RECORD_AUDIO — Microphone access
• FOREGROUND_SERVICE / FOREGROUND_SERVICE_MICROPHONE — Background recording
• POST_NOTIFICATIONS — Persistent notification (Android 13+)
• INTERNET — Google Drive uploads
• READ_PHONE_STATE — Call state detection

Project Structure

app/src/main/java/com/example/voiceinsights/
├── MainActivity.kt          # Entry point
├── MainScreen.kt            # Compose UI (permissions + start/stop)
├── RecordingService.kt      # Foreground Service
└── AudioCaptureManager.kt   # MediaRecorder chunking engine

Setup

1. Clone the repo
2. Open in Android Studio (Meerkat+)
3. Sync Gradle
4. Run on a physical Android device (emulator mic support is limited)

License

Private — Personal use only.

Audio Processing & Sarvam AI Integration (Python)

A set of Python scripts is provided in the scripts/ directory to process audio uploaded to Google Drive.

Features

1. Silence Removal: Uses the Silero VAD machine learning model to strip out silent background noise, preserving only actual speech.
2. Auto-Upload: Converts the processed audio back to .m4a and uploads it directly back into your Google Drive in the original folder.
3. Sarvam API: Transcribes the processed audio directly using Sarvam AI's Speech-to-Text API.

Setup Instructions

1. Install Python 3.8+ on your computer.
2. Install FFmpeg on your computer and make sure it is added to your system's PATH (required for converting audio to .m4a).
3. Open a terminal and navigate to the scripts/ folder inside the project.
4. Install the required Python packages:

   pip install -r requirements.txt

5. Get Google Drive API Credentials:
   • Go to the Google Cloud Console.
   • Enable the Google Drive API.
   • Create OAuth 2.0 Client ID credentials (Desktop app).
   • Download the JSON file, rename it to credentials.json, and place it inside the scripts/ folder.
6. (Optional) For transcription, set your Sarvam AI Pro Plan API key as an environment variable:
   • Windows (Command Prompt): set SARVAM_API_KEY=your_api_key_here
   • Windows (PowerShell): $env:SARVAM_API_KEY="your_api_key_here"
   • Mac/Linux: export SARVAM_API_KEY="your_api_key_here"

Running the Script

Run the script to automatically fetch all .m4a files from the last 45 days, clean them, re-upload them to Drive, and transcribe them:

python process_audio.py

(On first run, a browser window will open asking you to log into your Google Account to grant Drive access.)