AI-VoiceChat (Self Hosted everything)

GOAL: Develop a Self hosted Voice Chat application with a self hosted AI in real time. Everything needs to be self hosted, and in real time.You will be interacting the LLM using voice instead of typings.

v0.0.1 (Self host Ollama)

1. Self host Ollama on windows/WSL with any model. This model will be used for interaction. It may be switched to different model later.
2. Test it.

---

v0.0.2 (Accessing Ollama from WSL 2)

1. Start Ollama with env host to 0.0.0.0 so that request coming from WSL 2 will be accepted.
2. serve ollama with the updated env.
3. Find windows IP from WSL 2. We will need to use this ip to interact with ollama.
4. Test if we can communicate with ollama api from WSL 2 cli. (try to hit any ollam api from cli).

---

v0.0.3 (Starting a Nestjs project)

1. Start a nestjs project on WSL 2.
2. Create a module Ollama.
3. Create controller / services for it.
4. Connect with ollama running on windows 10. (NOTE: the endpoint needs to be of the windows machine, not localhost:11434.)
5. Create an endpoint POST /ollama/chat with body { prompt: 'Howdy!! }.
6. Pass the prompt to ollama.
7. Stream ollama response back to client instead of waiting for complete response.
8. Test it

---

v.0.0.4 (Audio pass-through between windows and WSL 2)

1. Setup PulseAudio on windows to allow audio pass-through from windows to WSL 2.
2. Install Sox / arecode in WSL 2 to receive audio from pulseAudio.
3. Test it.

---

v0.0.5 (Implementing audio recording)

1. Create a new module VoiceChat. All the voice chat related code (audio recording / streaming / processing / STT etc ) will be done here.
2. Create an endpoint GET /voice/chat.
3. Integrated npm package Mic here to capture audio, and stream it to a file.
4. Test it.

---

v0.0.6 (Integrating any STT)

1. Integrate real-time Speech to Text. (can be within same application or host a sperate server for it.)
   1. Using cloud service (Not a option since it has to be self hosted)
   2. Using Pre-existing solution that convert audio to text in real time. Host them locally on a server.

---