Inno-sama

A voice to voice artificial conversation system.

She can sing!

Neural Voice Model checkpoint with example link

Inference using RVC, for example link

---

Authors

• n.kurkulskiu@innopolis.university https://t.me/dadagust
• a.rakhmetov@innopolis.university https://t.me/R2R_6

---

Key Components and Workflow

1. Server side

• Purpose: Runs the LLM and TTS voice model

  • Gets the text data from the client and then, returns the .wav file with the voice to it.

2. Client side

• Purpose: Runs the vosk model to generate text, from the audio of a client
  • Connects to the server and send the text data to the server, then voices .wav, that the server returned

3. Ollama

• Purpose: Creates and serves a model from the .guff file
• Steps:
  • ollama serve to create an ollama server for the model.
  • ollama create innosama -f ModelFile to create an instance of an LLM model
  • The rest of the work is done by the server code