I’d love to see certain time-consuming tasks off-loaded to the GPU.
So support for CUDA (nVidia) or ROCm (AMD, former OpenCL). Maybe also for Intel.
This is already common in video-processing software such as Vegas, or even in games like Go. GPU acceleration could potentially speed up AI-based functions like dialog transcription, downmixing, VST processing, and more.
Having a DAW with a capable graphics card is not unusual these days, as modern systems are often used for a wide range of tasks thanks to powerful hardware — office work, internet, recording, mixing, mastering, and gaming.
Another advantage is that many GPUs now support a zero-fan mode. Depending on the load, the fans often remain silent as long as the GPU temperature stays below roughly 60 °C.
Currently I am using a combination of SpectraLayers 12 Pro to demix vocals to perhaps make the quality of Dialog Transscription better, to extract the lyrics of a song.
I see that SpectraLayers takes a long time, but also doesn’t consume much CPU.
Maybe GPU could accelerate here .. whatever makes it slow.
Many thanks for considering / implementing this.
P.S.: now using demix complete song with vocals and several instruments I see,
that the GPU is being utilized a lot by SpectraLayers, nice!
I wished more like this in Nuendo (and Cubase) to relieve my CPU…
