Udio

AI music creation platform

videono codeFree tier

Overview

Udio generates full songs with vocals, instruments, and professional production quality from text prompts. Supports many genres and styles. Known for more authentic vocal quality than competitors. Free tier includes 10 songs/day.

Best For

+Music creation
+Background tracks
+Jingles
+Creative projects

News & Updates

MiniMax H3 Gets Day-0 ComfyUI Support: Open Weights, Native Audio, 2K Video

ComfyUI has added day-0 support for MiniMax H3, a new open-weights model that brings native audio generation and 2K video capabilities to the platform. The integration, announced on ComfyUI's official blog, allows users to run H3 locally with full control over generation parameters. H3's open weights mean developers can fine-tune it for custom audio and video tasks without API costs. This marks a significant step in making advanced multimodal generation accessible to the open-source community.

new_tools8/3/2026

Production Company Sues Netflix for $105M Over Nicolas Cage Film Leaked by Vendor Breach

Netflix is being sued for $105 million after an unreleased Nicolas Cage movie, 'The Unbearable Weight of Massive Talent,' was stolen and leaked online. The lawsuit, filed by the production company, alleges Netflix failed to protect the film from cybercriminals who accessed it through a security breach at a post-production vendor. The leak occurred months before the film's scheduled release, causing significant financial damage and undermining the studio's marketing campaign. This case highlights the growing risk of intellectual property theft in the streaming era, where high-value content is increasingly targeted by hackers.

new_tools8/3/2026

ByteDance's Seedance 2.5 Generates 30-Second Video Clips with Built-in Audio

ByteDance has released Seedance 2.5, an AI video model that generates video and audio simultaneously in a single pass. Each clip runs up to 30 seconds, which is three times longer than Google's Gemini Omni Flash output. The model accepts dozens of reference images, videos, and audio files, enabling rich, multi-modal conditioning. For advertising teams, this could eliminate the tedious process of stitching together short clips one at a time, streamlining creative production workflows.

new_tools8/1/2026

MiniMax H3: Open-Weight Omni-Modal Model Released, Challenges GPT-4o

MiniMax, the Chinese AI startup behind the popular Talkie chatbot, has released H3, an open-weights omni-modal model that processes text, images, audio, and video. The model is designed to compete with closed-source rivals like OpenAI's GPT-4o, offering a rare fully open alternative for multimodal tasks. H3's release includes model weights, training code, and evaluation benchmarks, marking a significant step for open-source AI. This move could accelerate adoption of open models in production environments where data privacy and customization are critical.

breakthroughs8/1/2026

EU Mandates AI Content Labels by Aug 2026, Deepfakes Must Be Disclosed

The European Union's AI Act rules on labelling AI-generated content become legally binding on August 2, 2026. Companies building generative AI systems must mark images, audio, video, and text as artificial, and publishers of deepfakes or AI-written text on public interest matters must disclose this to viewers. This shifts from voluntary guidance to enforceable law, with penalties for non-compliance. The regulation aims to increase transparency and combat misinformation as AI-generated content becomes more prevalent.

new_tools7/31/2026

Researchers Discover Audio Injection Attack That Bypasses Safety Filters in AI Agents

Researchers have demonstrated a novel attack vector against multimodal LLM agents that uses concurrent audio prompts to inject malicious instructions, bypassing existing safety filters. The technique, detailed in a paper on arXiv, exploits the parallel processing of audio and text inputs to stealthily alter agent behavior. This attack can manipulate AI systems to perform unintended actions, such as exfiltrating data or executing unauthorized commands, without triggering typical detection mechanisms. The findings highlight a critical vulnerability in the integration of audio capabilities into AI agents, which are increasingly deployed in voice assistants and autonomous systems.

other7/31/2026

Gander: Android File Viewer with Zero Permissions Opens PDFs, Office Files

Gander, a new Android file viewer created by an independent developer, opens PDF, Word, Excel, PowerPoint, images, video, audio, Markdown, and code files without requesting any permissions. The app leverages Android's Storage Access Framework (SAF) to access files via user-selected documents, avoiding the need for broad storage permissions. This approach addresses privacy concerns about apps accessing sensitive data. Gander is available on GitHub as an open-source project, allowing developers to audit its code. The creator built it out of unease about opening files from others, prioritizing user privacy.

new_tools7/31/2026

Researchers Introduce MMAC Benchmark for Fine-Grained Audio AI Understanding

Researchers led by Weijie Wu, Junbo Li, and Lin Li have introduced MMAC, a massive multi-dimensional benchmark for audio captioning. Unlike existing evaluations that focus on generation quality or task performance, MMAC tests audio large language models (AudioLLMs) on open-ended, fine-grained free-form descriptions. The benchmark includes diverse audio types and multiple evaluation dimensions to diagnose information completeness and accuracy. This pushes the field beyond simple captioning toward richer, more detailed audio understanding.

breakthroughs7/30/2026

Google Releases Gemini Robotics ER 2 for Multi-Robot Coordination Without Central Control

Google has released Gemini Robotics ER 2, an embodied-reasoning AI model that acts as a high-level planner for robots, delegating motor control to a separate model. It is available to developers via the Gemini API and Google AI Studio, with private-preview access on the Gemini Enterprise Agent Platform. The model enables multiple robots to coordinate tasks like assembly or warehouse sorting without centralized control. This marks a shift from single-robot AI to multi-agent robotic teamwork, reducing the need for custom integration code.

new_tools7/30/2026

Developer Demonstrates Claude AI Editing and Transcribing Audio in YouTube Demo

A developer demonstrated using Anthropic's Claude AI as a single tool to edit, master, and transcribe audio files in a video posted on YouTube. The workflow shows Claude processing raw audio, applying noise reduction, adjusting levels, and generating accurate transcripts without needing separate software. The demo highlights Claude's ability to handle multi-step audio production tasks through natural language commands. This approach could replace traditional audio editing suites for basic to intermediate tasks, saving time and reducing tool complexity.