Tencent open-sources 1.5B AuK audio model that replaces a whole stack of audio tools
solyarisoftware · x · 2026-09-10
Tencent open-sourced the AuK family, a 1.5B-parameter audio model controlled by natural-language instructions that replaces a stack of separate tools:
- Zero-shot voice cloning and TTS
- Replace, insert or remove spoken words
- Change emotion, pitch, speed, volume; remove accents
- Normal speech ↔ whisper conversion
- Rewrite lyrics while preserving melody and voice
- Denoising, dereverberation, speaker separation, vocal extraction
MIT licensed with weights and code released; supports CLI, Gradio and ComfyUI. A distilled AuK-Flash needs only 4 inference steps, with a reported 4.5× wall-clock speedup, and Tencent claims it beats Qwen3-TTS.
More from Models
- Open Models Match GPT-6 on MedMCQA but Trail 17 Points on Hard Clinical Cases — keunwoochoi · 2026-10-27
- Health-Focused Benchmark Reranks LLMs: Muse Spark Open Jumps from #9 to #1 — keunwoochoi · 2026-10-27
- DeepSeek cut KV cache per token from 389KB to 890B in two years as open-closed gap debate rages — teortaxesTex · 2026-10-27
- GPT-6 Astra controls dexterous hands zero-shot at just 5.2% success in 500 trials — stingning · 2026-10-27
- Unverified: fable 5.5 speculated at 4-5T parameters to outscale 1-3T Opus — haider1 · 2026-10-27
- Kimi K3 Is 'Open' Until You Succeed: License Caps Free Use at $20M Revenue — alex_verem · 2026-10-27