🎙️ TwinMind Voice Recorder

Overview

TwinMind Voice Recorder is an Android application designed for efficient audio recording, chunked transcription, and AI-powered summary generation using the Google Gemini API. It leverages Room database for persistent storage and Jetpack Compose for a modern UI experience.

• Important: For details about classes, files, and function properties, refer to the documentation and class information.

⚙️ Minimum Requirements

• Android SDK 24+

🚀 What You'll Build

Three core features:

1. 🎤 Record Audio Robustly
   • Background recording with explicit interruptions handling
2. 📝 Generate Transcript
   • Convert audio to transcript
3. 📄 Generate Summary
   • Create structured summary from transcript

🛠️ Feature Details & Requirements

1. 🎤 Record Audio

• Foreground service for audio recording
• Split into 30-second audio chunks (with ~2-second overlap for continuity)
• Save chunks to local storage
• Persistent notification with Stop action
• Handles:
  • 📞 Incoming/outgoing phone calls (pause/resume, show status)
  • 🔇 Audio focus loss (pause/resume, show status)
  • 🎧 Microphone source changes (Bluetooth/wired headset, show notification)
  • 💾 Low storage (check before start, stop gracefully, show error)
  • 💀 Process death recovery (persist session state in Room, finalize chunk, resume transcription)
  • 🤫 Silent audio detection (warn after 10s silence)
• Live updates (Android 16+):
  • ⏱️ Recording timer
  • 🟢 Status indicator
  • ⏸️ Pause/Stop actions
  • 🎬 Visual indicator (icon)
• UI:
  • 🖥️ Single recording/post meeting screen
  • ⏺️ Record/Stop button
  • ⏲️ Timer
  • 🟢 Status indicator
  • 📋 Dashboard listing all meetings

2. 📝 Generate Transcript

• Upload chunks as soon as each 30s chunk is ready
• Uses Google Gemini 2.5 Flash (commented code for free version, limit <50kb)
• Save transcript to Room database (single source of truth)
• Ensure transcript order and chunk integrity

3. 📄 Generate Summary

• Send transcript to LLM API (Gemini)
• Generate structured summary
• Show specific error messages for failures

---

🏆 Accomplishments

• Used Gemini for both transcription and summary
• For transcription, use commented code for free Google AI Studio (limit <50kb)
• Added icons for UI clarity

✨ Features

• 🎤 Audio recording and session management
• 📝 Chunked audio transcription using Gemini API (gemini-pro:generateContent)
• 📄 AI-generated structured summaries from transcripts
• 💾 Persistent storage of transcripts and summaries via Room database
• 🖥️ Clean, minimal UI with navigation between transcript and summary screens
• 🛡️ Robust error handling and demo fallback for API/network issues

🧰 Technologies Used

• Kotlin
• Android Jetpack (Room, ViewModel, LiveData, Navigation)
• Jetpack Compose
• OkHttp (for API calls)
• Google Gemini API (v1beta)
• Gradle

📦 Installation Guide

Prerequisites

• 🛠️ Android Studio (latest recommended)
• 📱 Android SDK 33+
• 🔑 Google Gemini API key (obtain from Google AI Console)

Steps

1. Clone the repository:

   git clone https://github.com/akv-iu/New-TwinMind.git
   cd New-TwinMind

2. Open in Android Studio:
   • File > Open > Select the New-TwinMind folder.
3. Configure API Key:
   • Add your Gemini API key to local.properties:

     GEMINI_API_KEY=your_api_key_here
     

   • Or update the key in the code (see TranscriptionService.kt).
4. Build the project:
   • Click "Build" or run:

     ./gradlew assembleDebug

5. Run on device/emulator:
   • Select a device and click "Run" in Android Studio.

📖 Usage

• 🎤 Record audio sessions and view transcripts.
• 📄 Generate AI-powered summaries from transcripts.
• 🖥️ Navigate between transcript and summary screens.

© Copyright

Copyright © 2025 akv-iu. All rights reserved.

This software and its source code are licensed for personal and educational use only. Commercial use, redistribution, or modification without explicit permission is prohibited.

📝 License

See LICENSE file for details (if available).

---

🔜 Next Focus Areas

1. 📄 Generate Summary (Streaming & Resilience)

• Stream structured summary in the UI as the response arrives
• Continuously update the UI with new summary content
• Ensure summary generation continues and completes even if the app is killed (use WorkManager or foreground service for resilience)

2. 🎤 Record Audio (Live Updates & Lock Screen)

• Show live recording status on the lock screen (Android 16+)
• Display recording timer (updates every second)
• Indicate current status: "Recording", "Paused - Phone call"
• Provide Pause/Stop actions directly from the lock screen
• Show a visual recording icon

---

🗄️ Database Evolution & Usage Notes

• The Room database schema has evolved to support chunked audio, transcripts, and summaries.
• Important: There is currently no version control or error handling for schema/data changes. If you change the data model, you must delete all database elements before reopening the app, or it may crash.

---

For questions or contributions, please contact the repository owner via GitHub.