Reykjavik-based Treble has extended its Series A to build acoustic simulation infrastructure for voice models, wearables and robots.
MiniMax's H3 model generates 2K video with native stereo audio from mixed text, image and sound prompts.