Sub-20ms precision lyrics alignment and pristine audio separation. Built for the modern KTV experience.
Karaoke+ is a hyper-optimized, hybrid machine learning pipeline designed to solve the two hardest problems in automated karaoke generation: mathematical timestamp drift and audio separation artifacts.
By utilizing a bifurcated architecture, Karaoke+ delivers studio-quality MDX audio for the human ear while routing raw audio through a DSP-traced CTC alignment matrix for the machine. The result? Zero AI hallucination, perfect legato vocal tracking, and sub-20ms lyric synchronization.
Engineered with a focus on functional minimalism and industrial-grade automation by Ian Tirta Harjono.
- Website: iantirta.com
- Instagram: @iantirtaa
Karaoke+ is an independent engineering project.