[WIP] feat: implement OpenAI TTS client and integrate with TTS controller #1858
Conversation
Signed-off-by: Shahriyar Jalayeri <shahriyar@posteo.de>
|
PRs are always welcome, and this looks great! There are a few permission details you may want to check for Tauri builds. With the latest code, tauri http fetch should be able to connect to any endpoint (see default.json#L95). For the built-in window.fetch, you’ll still need to adjust the CSP here: tauri.conf.json#L18. Ref: https://github.com/readest/readest/blob/main/apps/readest-app/src/services/sync/KOSyncClient.ts#L65 |
|
I’m also planning to implement a BaseHttpTTSClient so that other TTS clients can reuse common code. I’ll likely start working on this next week. |
Awesome! and thanks for the pointers, I will test it this afternoon. Also for setting the endpoint and API key, do you have any preference in terms of which menu? |
|
@shjala, could you let me know if you resolved the PR issue? |
The Settings - Custom dialog is a good place for setting the API key. |
I wasn't sure this project accepts PR form outsiders. This is currently WIP, but works fine when I build for web-dev
pnpm dev-web. I have tested in with my locally running Kokoro-FastAPI model which implements OpenAI compatible API. The logic is copy-pasted from EdgeTTS. But lacks the following :Also for some reason it wont even connect to the TTS engine when I build the macOS application. I'll figure it out.