Skip to content

README.md

Latest commit

 

History

History
86 lines (57 loc) · 3.5 KB

README.md

File metadata and controls

86 lines (57 loc) · 3.5 KB

🎙️ CoreScribe

Blazingly fast, local-first voice transcription for students and researchers. Built with 🦀 Rust, powered by OpenAI's Whisper via WGPU.

Interface

🚀 Quick Start (Windows x64)

Download CoreScribe

Important

Latest Release Integrity (v0.1.7)

To ensure you are using the authentic build, you can verify the executable:

  • SHA-256: b89e2d2cc56562c0092e4ca5bb8e42ac2483aa1021947138cb3e25b56c7f3f1e

(Verify via PowerShell: Get-FileHash .\CoreScribe.exe)

🎓 Why CoreScribe?

Transcribing lectures, interviews, or study group recordings can be expensive and time-consuming. Most services charge by the minute or require monthly subscriptions. CoreScribe changes that by putting the power of AI directly on your computer.

  • 💰 Zero Cost: No API keys, no subscriptions. It uses your own hardware.
  • 🔒 Privacy First: Your recordings never leave your device. Perfect for sensitive research interviews.
  • 🚀 Student-Proof Performance: Works on high-end gaming rigs (NVIDIA/AMD) and standard study laptops (Intel iGPU) alike.
  • ⚡ Built for Speed: No Python overhead. Native Windows performance.

✨ Features

  • Universal GPU Acceleration: Optimized for NVIDIA, AMD, and Intel graphics via WGPU.
  • Offline Mode: Once the model is downloaded, no internet connection is required.
  • Simple Workflow: Drag and drop your audio files and get your text in seconds.
  • Automated Setup: CoreScribe handles the heavy lifting, including model downloading and extraction.

🛠️ Tech Stack

CoreScribe is built on a modern, high-performance Rust stack:

  • UI: eframe / egui for a responsive, hardware-accelerated interface.
  • Inference: Whisper-based transcription with WGPU 0.19 backend.
  • Audio: hound for precise WAV processing.
  • Async: tokio to keep the UI fluid during heavy computations.
  • Systems: windows-rs for deep OS integration.

🚀 Getting Started

  1. Download: Grab the latest .exe from the Releases page.
  2. Run: Open CoreScribe.exe.
  3. Setup: On the first run, the app will automatically download the necessary AI models (approx. 150MB - 500MB depending on your choice).
  4. Transcribe: Select your .wav file and let your GPU do the work!

Note: Currently, CoreScribe supports .wav files. Support for .mp3 and .m4a is coming soon.

🤝 Contributing

This is an Alpha release, and we love feedback! Whether you are a fellow student developer or a Rust enthusiast:

  • Report bugs via Issues.
  • Suggest features (like PDF export or summary generation).
  • Submit Pull Requests to improve the audio pipeline.

📄 License & Attribution

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0).

Branding & Trademark Policy

CoreScribe is open-source, but its identity is protected. If you redistribute this software:

  • Attribution: You must retain the original "About" section and a link to this repository.
  • Naming: You may not use the name "CoreScribe" for derivative works without clear distinction.
  • Integrity: Credits to the original author must remain visible in the UI.

Made with ❤️ for students who value their time and privacy.