Skip to content

atomlayer/llama_cute_voice_assistant

Repository files navigation

llama_cute_voice_assistant

Motivation: Many voice assistants with artificial intelligence have an unpleasant voice. While it may be a matter of personal preference. I aim to create a more human assistant that allows you to plug in cute anime character voices and vtuber voices.


Solution diagram:



How to install

This guide may not be the most detailed. It will need to be improved.

Step 1 - Install llama_cute_voice_assistant

Installation procedure for Windows

  1. Install git https://git-scm.com/downloads
  2. Install cudatoolkit (You should only choose versions of cuda that pytorch supports. See supported versions here https://pytorch.org/get-started/locally/)
  3. Install miniconda https://docs.conda.io/en/latest/miniconda.html
  4. Open miniconda console
  5. Create new conda environment
conda create --name llama_cute_voice_assistant python=3.11
  1. Activate conda environment
conda activate llama_cute_voice_assistant
  1. Clone project
git clone https://github.com/atomlayer/llama_cute_voice_assistant.git
  1. Go to project directory
cd llama_cute_voice_assistant
  1. Install pytorch
  1. Install the libraries
pip install SpeechRecognition==3.10.0 
pip install pyttsx3==2.90 
pip install soundfile==0.12.1
pip install simpleaudio==1.0.4
pip install pygame==2.5.1
conda install PyAudio
pip install openai-whisper --no-cache-dir
pip install omegaconf==2.3.0
pip install git+https://github.com/openai/whisper.git
conda install -c conda-forge ffmpeg


Step 2 - Install oobabooga Text generation web UI

  1. https://github.com/oobabooga/text-generation-webui#one-click-installers

  2. Open the oobabooga Text Generation web UI using the -api parameter.

  3. On the model tab: download and run your favorite AI model.

  4. On the Chat settings > Character tab: set your character name and description.



Step 3 - Install Realtime Voice Changer

Detailed instruction - https://www.youtube.com/watch?v=_JXbvSTGPoo

  1. Download https://huggingface.co/wok000/vcclient000/blob/main/MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.11.zip

  2. Unpack the archive

  3. Run start_http.bat

  4. Join the AI Hub Discord: https://discord.gg/aihub

  5. Go to the search-models channel

  6. Find and download the model you like

  7. Click the edit button in the Realtime Voice Changer Client

  8. Upload the model to a free cell

  9. Adjust the best TUNE parameter for your voice.



Step 4 – Install the virtual microphone

  1. Download and install VB-CABLE Virtual Audio Device https://vb-audio.com/Cable/

  2. Open Realtime Voice Changer

  3. Set up an audio input: Cable Output (VB-Audio Virtual Cable)

  1. Press start button


Step 5 - Change the settings in the settings.py file

  1. Replace oobabooga_api_name and wake words.
    oobabooga_api_name - the name of one of your characters in oobabooga text generation web UI (Paremeters > Character tab)
oobabooga_api_name = "Lisa"
wake_words = ["lisa"]


How to use

  1. Open conda console in the project folder
  2. Run the command:
python voice_chat.py
  1. Say the wake word and the command for your assistant.

Links

https://github.com/davabase/whisper_real_time