[FEAT]: Support Deepgram APIs for TTS and STT in AnythingLLM

[pcgeek86]

What would you like to see?

Hello,

AnythingLLM is some pretty sweet software. I'm just getting started with it, along with Ollama and LM Studio, but so far everything works perfectly!

I noticed that AnythingLLM supports custom speech-to-text (STT) and text-to-speech (TTS) providers, since 2024. #909

It would be pretty cool if Deepgram were supported as a transcription and text-to-speech provider in AnythingLLM. There are WebSocket APIs for live STT and TTS, as well as HTTP endpoints for submitting pre-recorded audio.

Here are the WebSocket APIs that would be more suitable for real-time usage:

https://developers.deepgram.com/reference/speech-to-text/listen-streaming
https://developers.deepgram.com/reference/text-to-speech/speak-streaming