T5ynth 2.5: Stable Audio 3, Polyphonic Generative Sequencer, RePrompt, Snapshots, better UI

Learn about the project here: https://github.com/joeriben/T5ynth# (early version intro:    • Introducing T5ynth: A synthesizer that dec...   ). Demos of more experimental patches:   / t5ynth   T5ynth is a synthesizer that uses generative audio models like no other instrument, allowing to explore sonic spaces that simple prompt can not produce. It works purely locally on your machine without any online activities. T5ynth has evolved a lot in the last 2 months, huge UI improvements, Stable Audio 3 support, english prompt translation. T5ynth now features like ReSynth (feeding its output into its input) and RePrompt (it "listens" now to its own output and uses its interpretation of the sound to alter the A/B prompts in different ways), and it features a configurable polyphonic generative sequencer that is even able to manipulate its own parameters 8if you let it). Of course, T5ynth is and will always be OpenSource. Download from GitHub for Win, Mac and Linux: https://github.com/joeriben/T5ynth/re... (click "assets") Direct download for MacOs : https://github.com/joeriben/T5ynth/re... T5ynth will need either an RTX graphic card (gpu) with 16GB VRAM (could work with less, not tested) or a Mac with an M GPU, preferably Pro or Max versions (for inference speed, but does work on slower machines too).