From the very beginning of Soundry AI we asked the question: what if we used AI to empower artists rather than replace them? We remain committed to this vision. Rather than blindly ripping data from the web, we have been very careful to respect the needs of creators. Here we aim to shed more of a light on how we did this and what our plans are for the future.
Those of you who know how AI works are aware that an insanely large amount of data is required to train these neural networks. Thankfully, the open source community has been incredibly generous and has provided us a lot of data to work with. In particular, we have found Freesound and the NSynth Dataset to be very helpful. Freesound includes sounds under the CC0, CC BY, and CC BY-NC licences, and NSynth is under the CC BY 4.0 licence. We have filtered out any Freesound sounds which use CC BY-NC (not acceptable for commercial use), and you can find our full list including attribution for every single creator here.
We will always have the need to keep some data sources private because we don’t want to give our competitors a blueprint for how to replicate our work, but you can trust us to respect the needs of artists and copyright law.
We want to support artists both by providing them a powerful tool and also by licencing their work. If you are interested in partnering with us, then please fill out our application. We look forward to working with you to make our tool even better!