rzru/nightingale

Machine learning powered Karaoke app (with scores!)

What it solves

Nightingale allows users to turn any song in their existing music library into a karaoke track. It removes the need for pre-made karaoke files by automating the separation of vocals from instruments and the generation of synchronized lyrics.

How it works

The system uses a Rust-based core and a React frontend, utilizing a local Python analyzer powered by neural networks. This analyzer handles vocal separation (via UVR Karaoke or Demucs) and lyrics processing (through automatic transcription, alignment, or external sources like LRCLIB). To ensure performance, the analyzer runs as a persistent process to avoid model startup delays, and results are cached using BLAKE3 hashes.

Who it’s for

Music lovers and karaoke enthusiasts who want to create custom karaoke tracks from their own local audio/video files or connected media servers like Plex, Jellyfin, and Navidrome.

Highlights

  • Neural Vocal Separation: Uses ML models to strip lead vocals from songs.
  • Automated Lyrics: Supports automatic transcription and alignment for word-timed lyrics.
  • Flexible Library Integration: Connects to local folders and popular media servers.
  • Karaoke Features: Includes pitch scoring, key and tempo adjustment, and multiplayer support for 2-4 players.
  • Visual Customization: Offers dynamic backgrounds via GPU shaders, video, or images.
  • Cross-Platform: Available as a native desktop app (Windows, macOS, Linux) or a self-hosted web app.

Related

  • Project
    xcLee001/SonicVale

    An open-source AI voiceover platform that automates multi-character and multi-emotion speech synthesis for novels and scripts.

  • Project
    nomadkaraoke/python-audio-separator

    Audio Separator is a Python library/CLI that uses deep‑learning models (MDX‑Net, VR, Demucs, MDXC) to split music files into stems such as vocals, instrumental, drums, etc. It supports CPU, CUDA, Apple Silicon (MPS/CoreML), and experimental DirectML, offers Docker images, and provides a model catalog with filtering options.

  • Project
    sergree/matchering

    An open-source audio matching and mastering tool that makes a target audio track sound like a reference track by matching its RMS, frequency response, and stereo width.

  • Project
    mexicat/pdoom-video

    A generative, code-rendered music video system that uses audio analysis and three.js to create deterministic, word-synced visuals exportable in 4K60.

  • Project
    ServeurpersoCom/acestep.cpp

    acestep.cpp is a C++‑based local server that generates 48 kHz stereo music from text prompts. It uses GGML‑quantized models (LM, text encoder, diffusion DiT, VAE) and runs on CPU, CUDA, Vulkan or Metal. After downloading the four GGUF model files, you build the binary (CPU‑only, CUDA, Vulkan, or all‑in‑one) and start the server. A web UI at http://localhost:8085 lets you write a caption, select a model/LoRA adapter, and generate, play, or download tracks. The server also offers a tiny REST API (POST /lm, /synth, /understand, /vae) and CLI tools for scripting. It is a genuine open‑source AI music‑generation project, not a tutorial list.