Hi @starrytong, and thank you for SCNet — it's an excellent model and the results are very impressive.
I maintain an open-source, MIT-licensed audio stem-separation tool, and I'd like to redistribute the pretrained SCNet-large weights (the checkpoint linked from your README on Google Drive) as the default model, served from our own download mirror.
Your repository's LICENSE file is MIT, which clearly covers the source code. However, neither the README nor the weight files themselves state a license for the pretrained checkpoints specifically. To be careful about this, I'd be grateful for explicit written confirmation on two points:
- Under what license are the released SCNet / SCNet-large pretrained weights distributed? (We're hoping it's MIT, matching the code.)
- Are we permitted to redistribute those weights — including a format-converted ONNX export of them — as part of an MIT-licensed tool, with attribution to you?
A short reply in this issue confirming the license would be perfect — it gives us a citable record. And I'm very happy to include any attribution or NOTICE text you'd like us to carry.
Thank you very much for your time, and for releasing this work.
Hi @starrytong, and thank you for SCNet — it's an excellent model and the results are very impressive.
I maintain an open-source, MIT-licensed audio stem-separation tool, and I'd like to redistribute the pretrained SCNet-large weights (the checkpoint linked from your README on Google Drive) as the default model, served from our own download mirror.
Your repository's
LICENSEfile is MIT, which clearly covers the source code. However, neither the README nor the weight files themselves state a license for the pretrained checkpoints specifically. To be careful about this, I'd be grateful for explicit written confirmation on two points:A short reply in this issue confirming the license would be perfect — it gives us a citable record. And I'm very happy to include any attribution or NOTICE text you'd like us to carry.
Thank you very much for your time, and for releasing this work.