Scriber - Windows Voice Dictation

Scriber is a high-performance, AI-driven voice dictation application for Windows. It replicates the functionality of tools like Aqua Voice and Wispr Flow, allowing you to dictate text into any application system-wide.

Quick Start (Windows)

1. Download the repository.
2. Double-click start.bat.
   • It will automatically set up the environment, install dependencies, and prompt you for API keys.
3. Dictate: Press Ctrl+Alt+S to start/stop listening.

Features

• System-Wide Dictation: Works in any application (Word, IDEs, Browser, etc.).
• Global Hotkey: Activate voice capture with Ctrl+Alt+S (configurable).
• Multi-Engine Support:
  • Soniox: Ultra-low latency streaming with custom vocabulary.
  • AssemblyAI: High accuracy with punctuation.
  • Deepgram: Fast and cost-effective streaming.
  • OpenAI (Whisper): High accuracy via Whisper API.
  • Azure Speech: Microsoft's enterprise STT.
  • Gladia: Audio intelligence API.
  • Groq: Fast inference for Whisper models.
  • Speechmatics: Specialized ASR.
  • Google Cloud STT: Enterprise-grade recognition.
  • ElevenLabs: Scribe model integration.
• Smart Turn Detection: Optional pause detection when supported by the environment.
• One-Click Setup: Automated start.bat script for easy installation.

Manual Installation

If you prefer to run it manually or are on Linux/Mac:

1. Clone the repository.
2. Install dependencies:

   pip install -r requirements.txt

3. Create a .env file with your API keys.
4. Run:

   python src/main.py

Configuration

The start.bat script will create a .env file for you. You can also edit it manually:

# STT Service API Keys
SONIOX_API_KEY=your_key
ASSEMBLYAI_API_KEY=your_key
DEEPGRAM_API_KEY=your_key
OPENAI_API_KEY=your_key
AZURE_SPEECH_KEY=your_key
AZURE_SPEECH_REGION=westus
GLADIA_API_KEY=your_key
GROQ_API_KEY=your_key
SPEECHMATICS_API_KEY=your_key
ELEVENLABS_API_KEY=your_key
GOOGLE_APPLICATION_CREDENTIALS=path/to/json

# App Settings
SCRIBER_HOTKEY=ctrl+alt+s
SCRIBER_DEFAULT_STT=soniox  # Options: soniox, assemblyai, deepgram, openai, azure, gladia, groq, speechmatics, google, elevenlabs
SCRIBER_MODE=toggle         # toggle or push_to_talk
SCRIBER_CUSTOM_VOCAB=Scriber, Pipecat, Soniox

Requirements

• Windows 10/11 (Recommended)
• Python 3.10+
• Microphone