I just released this video where I explored the possibility to import samples into genopatch that have been created from text prompts. while doing so I realized there is a bit of a problem when the samples generated from texttosample are polyphonic. That needs workarounds and it's kinda sketchy, but fun. However it made me wonder wouldn't it be cool if Synplant had its own text prompt feature where natural language would be interpreted as plugin presets? i mean, that wouldn't be too different from what the dataset of genopatch already has in it, would it? the samples that were used to train the NN probably had names like "thumpy kick" or "break 120bpm" or "deep pad" and all that. so those names could serve as training data for that feature

Very cool! Thank you for sharing!