Add google Gemini API support(free) or Ollama self hosted server for custom translation LLM model

[smyhlin]

[smyhlin]

Working Code implementation can be found here

Instruction how to obtain key can be found here.

[umlx5h]

I will consider putting it in the roadmap, it will be Later.

I may consider adding a translation engine through a plugin mechanism, as it would be expensive to maintain all of them.

[Haks1]

Hi,

Noice, since the ASR is local, it would be sweet to have translation locally too.

[Haks1]

For the Ollama, we may just need to set a tiny http server and prompt the model (Mixtal seems the best for trad) to act error free, for client-server HTTP REST communication, exactly like one of the Google or DeepL API (we may need to precise the API version, maybe add the API manifest and ask him to mock anything that is not really translation),

[umlx5h]

okay, there seems to be demand for it, I would try add one or two translation services that could be implemented with just API calls soon.

[umlx5h]

I tried Gemini but had problems with very slow latency.
Also, I got 429 very quickly, probably due to the free API key.

  "error": {
    "code": 429,
    "message": "Resource has been exhausted (e.g. check quota).",
    "status": "RESOURCE_EXHAUSTED"
  }

If there is a fee, DeepL would definitely be a better choice.
I will add Ollama, which works locally and is versatile.

[umlx5h]

I have added support for Ollama, LM Studio, OpenAI and Claude. It will be released in the next version.
75041b7

I did not add support for Gemini because it is not OpenAI compatible.