UltraSinger
Automates the creation of UltraStar karaoke files, MIDI, and note data from audio using AI-driven transcription.
Convert audio and YouTube links into fully synced UltraStar karaoke files with automated pitch and lyrics.
UltraSinger is an automation tool for karaoke enthusiasts and producers who want to generate UltraStar-compatible files from raw audio. It uses AI models to handle the heavy lifting of transcription, pitch detection, and note alignment.
What it is
The tool processes audio or video files to generate UltraStar.txt files, MIDI, and sheet music. It utilizes Whisper for lyrics transcription and various separation models to isolate vocals, then automatically handles pitch detection and note tapping.
Who it's for
It suits users who want to digitize music for karaoke without manual note entry. It is not for users who lack technical familiarity with Python environments or those without a capable GPU for efficient model inference.
Key features
Automated Vocal Transcription
Uses Whisper models to convert singing into synchronized text lyrics.
Pitch Detection and Quantization
Extracts vocal pitch and aligns notes to the musical key.
Audio Track Separation
Employs demucs models to isolate vocals from background music.
YouTube Integration
Directly processes audio from YouTube links via yt-dlp integration.
Up and running in 3 steps
Our honest verdict
What's great
- Automates the tedious manual labor of tapping and syncing karaoke notes.
- Supports re-pitching existing UltraStar files to improve accuracy.
- Includes specific features for musical key quantization and hyphenation.
- Provides an interactive mode for users who prefer guided configuration.
Where it falls short
- Requires a local Python environment and manual dependency management.
- High VRAM requirements for optimal performance on larger Whisper models.
- Performance is heavily dependent on the quality of the source audio.
- Setup involves installing Python, ffmpeg, and managing CUDA drivers.
Pricing & license
The tool is free under its license. There are no paid hosted options; the only cost is the electricity and hardware resources required to run the AI models locally.
Open remote AI-training roles
879 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Can I use this on any song?
The developers recommend using it only on Creative Commons licensed songs.
What if my GPU is too weak?
You can force CPU processing or reduce the Whisper batch size to prevent crashes, though this will significantly increase processing time.
