search

erew123 / alltalk_tts

AllTalk is based on the Coqui TTS engine, similar to the Coqui_tts extension for Text generation webUI, however supports a variety of advanced features, such as a settings page, low VRAM support, DeepSpeed, narrator, model finetuning, custom models, wav file maintenance. It can also be used with 3rd Party software via JSON calls.
GNU Affero General Public License v3.0
686 stars 71 forks source link

Tune docker #249

Closed dpkirchner closed 4 weeks ago

dpkirchner commented 1 month ago

These commits update the docker configs to make it easier to work on the code locally. The main change that users will care about is the reordering of the dockerfile rules, reducing how much time it takes to test code updates that don't involve changing dependencies. The rest is just a little quality-of-life gravy.

I know there's a v2 coming, however I saw there's no dockerfile (yet). I figure you might be able to adapt this for v2 when alltalkbeta is merged to main.

erew123 commented 1 month ago

Hi @dpkirchner

Im spending most my days apologising to people at the moment for slow replies. Im very bogged down with lots of things between code and emails etc. So not to break that, apologies for my late reply!

Im guessing the code youve sent over youve tested and it works fine? (sorry to ask a dumb question). Im just not in a great position to test it out, but if you say it works, then Im happy to just import it in!

And yes, v2, Ive not got around to Docker yet, but it should be a bit easier to work with. Im not sure it would need the separate launch.sh script that v1 needs and the requirements I think will be simpler.

Im not quite a docker expert to be honest, so your help is very appreciated :)

dpkirchner commented 4 weeks ago

The changes are working well for me. The main difference is just moving where the source is copied so it happens after all of the dependencies are installed (which can take a while). I haven't tested them with v2 beta yet, however.

erew123 commented 4 weeks ago

Thanks for the work/help on this. I think v2 should be cleaner with Docker (no need to run a separate launch.sh script is my guess) but Ive not gotten there yet to test.

Ill pull the PR in.

Thanks again