Queuing runs

[DUR6NA]

I like to to use Kokoro to make audio versions of very large texts. the time to complete each can take quite a long time, hours in some cases. It would be super helpful if you could set a save file path and be able to queue multiple separate generations that autosave upon completion so multiple texts could easily be generated even well sleeping, at work, or otherwise occupied.

[RBEmerson970]

It would be helpful to know what hardware you're using, the OS, etc.

As a benchmark, I just let FastKoko process 262K characters (47,165 words) and it took 5:20 from start to finish. The resulting MP3 takes up 201 Mb, and has a playback time of 3:39.

The machine is a Legion 7 laptop, i9, RTX-4090, 64 Gb RAM, 16 Gb VRAM, Win 11, Docker 4.55.0. The instance of FastKoko is based on a Git from early December (saying V0.2.4 is meaningless - I've had what VERSION says is V0.2.4 since June, though I've done a few fresh builds over the last six months).