AuK: Tencent Hunyuan's Unified Audio Generation and Editing Model (TTS Demo)

Melvin Vivas · X video post · 2026-09-10 · 1:53 · 572 views · Open on X

Topics: Industry Trends & Job Market · Level: beginner

Summary

Melvin Vivas shares a demo of AuK, a new audio model from Tencent Hunyuan that combines text-to-speech and audio editing in one model. The caption says it handles zero-shot voice generation, speech editing, acoustic control, audio cleanup and source separation, and that it can remove an accent. The video is a run of before/after audio samples with no spoken explanation.

Key points

• AuK is described as a unified audio generation and editing model from Tencent Hunyuan.
• Capabilities listed in the caption: zero-shot voice generation, speech editing, acoustic control, audio cleanup and source separation.
• The quoted post says AuK can edit out a speaker's accent.
• The demo plays paired samples (original, then edited) with the same sentence re-rendered, which shows content-preserving speech editing.
• Some samples show edited phrases (e.g. 'I'm not taking anything away from that' changed to 'I'm not taking any shots'), which shows text-based word replacement inside existing speech.
• The project page auk-project.github.io has the official demos and details.

Resources mentioned

• AuK — Unified Audio Generation and Editing · website · auk-project.github.io · free
  Project page for Tencent Hunyuan's AuK model, with demos of TTS, speech editing, accent editing, cleanup and source separation.
• Tencent Hunyuan · website · x.com · free
  Tencent's AI model team, which makes the Hunyuan family of models, including Hy3.
  Also in: Hy3 (Tencent Hunyuan) Free on Nous Portal for a Limited Week (Melvin Vivas on X · notes)
• AuK · tool · github.com · free
  Open-source foundation model for unified speech generation and editing (zero-shot TTS, instruction control, content editing, whisper conversion).
  Also in: AuK: Open-Source Model for Generating and Editing Speech (Melvin Vivas on X · notes)