Nano Banana for Audio
Tencent has introduced AuK for speech generation and editing, featuring powerful voice customization capabilities.
Nano Banana for Audio
Tencent has launched AuK for speech generation and editing. This is a 1.5B model that transforms text into speech and edits ready recordings based on text instructions through a single interface. It clones voices based on references, changes words without re-recording the entire phrase, removes accents and noise, adjusts timbre, emotion, speed, and pitch, and separates the speaker or vocals from the mix.
It was trained on 1.95 million hours of "effective training audio." For understanding instructions, Qwen2.5-Omni-3B is used separately, so 1.5B is not the full size of the model. Along with the base version, AuK-Flash was released. It performs 4 steps instead of 32 and works 4.5 times faster. The code, weights, and ComfyUI nodes are already available for hands-on experimentation.
Links
Why it matters
AnalysisThis technology has the potential to significantly simplify the process of speech generation and editing, which can be beneficial in various fields, from media to education.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Tencent releases Hy4: an open-weight model challenging DeepSeek
Tencent has released Hy4, a multimodal model that scored 75.2 on APEX-Agents in internal tests, ahead of DeepSeek V4.1 Flash-1017. The company plans to publish the model weights.
Nano Banana: Graphic Design Prompts
A curated collection of production-ready prompts for generating design visuals — from packaging mockups to trending 2025 posters.
Prompts for hyper realistic skin in Nano Banana
This selection is a set of working prompts for shooting ultra-detail skin and body in an editorial, fashion context.
NotebookLM and Audio Overview (Podcast from Summaries)
A free research tool from Google based on Gemini models. It strictly adheres to uploaded sources (no hallucinations) and can transform dull reports into engaging conversational audio podcasts.