LocalSpeech is a project that sets up and runs AI-powered speech models with just one command on MacOS. All voice models are setup in openai client sdk format. This guide provides step-by-step instructions to install all dependencies, set up the environment, and run both backend and playground services.

Ensure you have the following installed before proceeding:

If you are using macOS and don't have Homebrew installed, run:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Miniconda (recommended for lightweight installation):

curl -fsSL https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-x86_64.sh | bash

Restart your terminal and initialize Conda:

conda init zsh

For Linux users, replace MacOSX-x86_64 with Linux-x86_64 in the download link.

You can install Node.js and npm via Homebrew (macOS) or a package manager:

brew install node

For Linux:

sudo apt install nodejs npm -y

brew install ffmpeg  # macOS
sudo apt install ffmpeg -y  # Linux

To set up and run a speech model backend, execute the following command:

chmod+x master-script.sh
./master-script.sh

This script:

• Prompts you to select a model.
• Clones the corresponding repository.
• Creates and activates a Conda environment.
• Installs dependencies.
• Starts the model API server in an openai compliant format

To set up and run the playground service, execute:

chmod +x run-frontend.sh
./run-frontend.sh

This script:

• Navigates to the frontend directory.
• Installs dependencies using npm install.
• Starts the development server using npm run dev.

• Whisper Speech Recognition
• Kokoro TTS
• Spark TTS
• Zonos TTS
• Sesame CSM

• If the backend folder exists but the Conda environment does not, the script will delete the folder and re-clone it.
• Make sure to activate the Conda environment before running the backend manually:

  conda activate <env_name>

• If you encounter permission issues, try running the scripts with chmod +x script_name.sh and then executing them.
• Implement the APIs in your own apps by referring the ./frontend/app/page.tsx page to get idea about the api.

• Conda command not found: Restart your terminal and run conda init zsh again.
• ffmpeg missing: Ensure it's installed with brew install ffmpeg or sudo apt install ffmpeg -y.
• Port conflicts: If the backend or frontend fails to start, ensure no other service is running on the same port.

This README provides all necessary steps to get LocalSpeech up and running. If you encounter any issues, feel free to report them!