Apple's karaoke model ships natively on macOS with on-the-fly voice dampening
banteg · x · 2026-11-01
Noah D highlights that Apple's karaoke feature runs an AI model natively on macOS, performing real-time voice dampening on songs. The model even accepts a -100 to 100 value, so you can boost vocals and duck the music. banteg is surprised: he assumed Apple would have pressured labels into handing over stems/masters for the whole catalogue rather than shipping a model.
More from Multimodal
- Z Image Turbo on an 8GB VRAM Laptop: 40-60s per Image, a Beginner Is Impressed — Tonka-Jahari-Pizza · 2026-11-01
- "Found Footage" From a Hiker's Camera in 1997 — Actually AI-Generated — ati29 · 2026-11-01
- Beyond the Text Bottleneck: Why Native Speech-to-Speech Changes Voice AI — afigueiredo · 2026-11-01
- Slopus 0.4.0: open-source local AI studio adds music gen, MOSS, Krea 2, Fibo 2.0 — BittiAI · 2026-11-01
- Halloween Cast List Includes 'Content Blocked' in All-AI Luma Short Film — mrjonfinger · 2026-11-01
- BOTZILLA: a Mechagodzilla-style short film made with Codex and Google Omni — BotLordBotflix · 2026-11-01