Skip to content

Sesame CSM

Generate conversational speech from text and audio

Visit official site

What Sesame CSM is

The model is described as a conversational speech generation model developed by Sesame, producing high-quality speech from text and audio input. The material states it is built on the Llama architecture, uses the Mimi audio encoder and is open source for research and education.

What you can do with it

  • Generate speech from text for synthesis tasks
  • Refine speech output with context information
  • Work with multiple voice styles and tones
  • Extend and customise the open source model
  • Deploy with the provided pre-trained models and code

Application scenarios

  • Build voice assistant applications with natural interaction
  • Generate teacher narration for online education platforms
  • Explore improvements to speech synthesis in research

Getting started

  • Clone the repository to a local machine
  • Create a virtual environment and install dependencies
  • Download the pre-trained models
  • Use the model to generate speech

Pros & cons

✓ What we like

  • Multiple voice styles and tones are listed
  • Open source model allows further development and customisation

! What to watch out for

  • The described inputs are text and audio
  • The model is stated for research and education purposes

FAQ

What does it generate?

The material states it generates conversational speech from text and audio input.

Is the model open?

The material states the model is open source with pre-trained models and code.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review