Replies: 2 comments 9 replies
-
|
Thanks for the thoughtful list, @goermezer — really appreciated 🙏 Job persistence ("save the jobs permanently"): this is now sorted.
Spotify support: we've looked into this, and we won't be adding it, due to the legal issues around it. There's no clean way to pull audio from Spotify — its streams are DRM-protected, and the only tools that actually rip them circumvent that DRM (a Terms-of-Service violation that also runs into anti-circumvention law like DMCA §1201). The popular "Spotify downloaders" (e.g. spotDL) don't really download from Spotify either: they read the track metadata and fetch a YouTube match instead — YouTube under the hood, with the added risk of grabbing the wrong/alternate version. Given StemDeck's local, privacy-first, not-a-downloader stance, we'd rather not take the project there. Pasting a YouTube or SoundCloud link remains the supported way to bring in source audio. |
Beta Was this translation helpful? Give feedback.
-
|
Hey @goermezer, glad to hear Alpha 3 is working well! And good to know cellular is usable too, even if slower. Alpha 4 reduced the chunk size from 10s to 5s so the first audio should arrive a bit faster on slower connections now. One more update to close the loop on your original list: playback speed control shipped in 0.8.0 Alpha 4. You can now slow down or speed up a track with the speed slider in the player. It uses a pitch-preserving algorithm (WSOLA, the same technique SoundTouch uses), so voices and instruments stay at their original pitch regardless of tempo. Practicing guitar at 0.75x still sounds like the original recording, just slower. Works on both desktop and mobile. On Intel and AMD GPU support: this one is genuinely complex to add properly. GPU-accelerated ML inference in Python is almost entirely built around the CUDA/PyTorch stack, which is NVIDIA-specific. The alternatives exist: ROCm for AMD and OpenVINO/IPEX for Intel, but they each require a separate build pipeline, platform-specific driver dependencies, and dedicated hardware in CI to test against. That means each GPU vendor is effectively its own artifact, its own compatibility matrix, and ongoing maintenance burden. We are keeping focus for now, but it is absolutely on the list for the future if we can get access to the right hardware. It is not forgotten. So from everything you raised:
I think that genuinely covers everything you brought up. I really want to say thank you for the quality of your engagement here. Your feedback was specific and actionable, you tested every release, and when we needed more detail you recorded a screen capture. That kind of collaboration is rare and it made StemDeck concretely better. Big thank you for your ideas and for helping push the project forward. If you want to follow along with what we are working on between releases, we share weekly updates on Instagram at @stemdeckapp. Always happy to see familiar faces there :) |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Love this idea. Actually a very usable and helpful tool. I`m really happy that I found it. In case you are looking for further development ideas, here are mine from a musicians point of view:
Technical:
Beta Was this translation helpful? Give feedback.
All reactions