Skip to content
/ lycoris Public

Real-time speech recognition & AI-powered note-taking app for macOS with offline/online modes, multilingual transcription, and Japanese translation support.

License

Notifications You must be signed in to change notification settings

solaoi/lycoris

Repository files navigation

lycoris

Real-time Speech Recognition & AI Note-taking App

license GitHub release (latest by date) GitHub Sponsors PIXIV FANBOX MacOs

Lycoris is a feature-rich, user-friendly app for macOS, designed to enhance your note-taking experience with real-time speech recognition, transcription, and translation capabilities. With offline and online modes, Lycoris balances flexibility with privacy.

⚠️ Note: Currently, Lycoris is optimized for Japanese users. While it aims to support transcription and translation in 19 languages in the future, the current version may have limited functionality in other languages.

📌 Introduction

日本語 English 中文 한국어 Français Deutsch Русский Español Português Türkçe

Tiếng Việt Italiano Nederlands Català Українська Svenska हिंदी(हिन्दी) Čeština Polski

Table of Contents

✨ Key Features

key-features

  • 🎙️ Real-time Speech Recognition:

    • Offline Mode:
      • Vosk for Real-time Transcription: Lycoris uses Vosk to quickly generate initial real-time transcriptions, which are then refined using Whisper for enhanced accuracy and detail.
      • Whisper for Detailed Transcription: The primary transcription model for offline mode, Whisper delivers high-quality, multi-language transcription. For Japanese-only use cases, Kotoba-Whisper is available as an alternative for optimized Japanese transcription.
    • Online Mode:
      • OpenAI Whisper API: For users seeking cloud-based transcription, Lycoris integrates with the OpenAI Whisper API, allowing for high-quality transcriptions with cloud resources.
      • AmiVoice API: Additionally, users can choose amiVoice for specialized Japanese transcription via API key.
  • 🌐 Advanced Translation:

    • Whisper + FuguMT: Supports translation from 18 languages into Japanese. Whisper transcribes any of the 18 languages into English, and FuguMT then translates the English transcription into Japanese, ensuring high accuracy.
    • Kotoba-Whisper-Bilingual: Specialized for direct bi-directional translation between Japanese and English.
  • 📄 Flexible Note-Taking with Markdown Support: Notes support Markdown formatting, allowing for organized and visually enhanced note-taking.

  • 💬 Contextual Speech Suggestions:

    • Sentiment-Based Suggestions: Generate contextual responses in Neutral, Positive, and Negative tones with just a click.
    • Text-to-Speech Capabilities: Use Style-BERT-VITS2 JP-Extra to convert suggested responses into natural-sounding speech.
  • 🤖 Real-Time ChatGPT Integration:

    • Interact with transcriptions or summaries in real-time using ChatGPT for tasks like summarization and contextual insights. This feature helps streamline discussions or capture essential points without losing context.
  • 📶 Offline & Online Modes: Fully functional offline with local models to ensure privacy; online mode available with API integration for cloud-based services like OpenAI Whisper API.

💻 System Requirements

  • Operating System: macOS 13 Ventura or later.
  • Processor: Compatible with Apple Silicon (M1, M2, M3) and Intel processors.
  • Memory: Minimum 8GB RAM recommended for optimal performance.

⚙️ Installation

  1. Download: Obtain the latest version of Lycoris from the GitHub Releases page.
    • For Intel processors: Download the .dmg file ending in x64.dmg.
    • For Apple Silicon processors: Download the .dmg file ending in aarch64.dmg.
    • Note: Only the version number in the file name changes with each release.
  2. Install: After downloading the .dmg file, double-click it to open. You’ll see the Lycoris application and the Applications folder. Simply drag the Lycoris app into the Applications folder.
  3. Permissions: On first launch, grant necessary permissions for microphone and screen recording as prompted.

🚀 Usage Guide

  1. Download Language Packs:

    • Open the app and go to settings (⚙️).
    • Download language packs for speech recognition as needed.
  2. Configure API Key (for online features):

    • In settings, enter your API keys for services like amiVoice or OpenAI Whisper API to enable extended transcription and AI features.
  3. Create a Note:

    • Click "Add Note" next to the search bar.
    • Set the speaker’s language, transcription method, and audio source.
    • Click "Start Recording" to begin capturing notes in real time!
    • With OpenAI’s online feature, you can also receive sentiment-based response suggestions (Neutral, Positive, Negative) for discussions, and use real-time ChatGPT integration to summarize or discuss the transcribed content directly.

🙌 Support & Contributions

Lycoris is available for free on BOOTH to reach a wide audience. Download the appropriate version for your macOS processor:

If you’d like to support further development:

  • For Japanese users: Contributions are welcome via Fanbox or by making a voluntary purchase on BOOTH.
  • For international users: You can support Lycoris by becoming a sponsor on GitHub Sponsors.

Whether you’re a developer interested in contributing or a user wanting to support this project, every contribution helps Lycoris grow and improve for all!

👥 Contributors

solaoi
Solaoi
mohemohe
Mohemohe

📈 Star History

Star History Chart

Your support keeps Lycoris accessible and continually improving. Thank you for helping us make smarter, AI-powered note-taking a reality! 📒✨

About

Real-time speech recognition & AI-powered note-taking app for macOS with offline/online modes, multilingual transcription, and Japanese translation support.

Topics

Resources

License

Security policy

Stars

Watchers

Forks

Sponsor this project

 

Contributors 3

  •  
  •  
  •