# SesameAILabs/csm

A Conversational Speech Generation Model

Repository: https://github.com/SesameAILabs/csm
Canonical: https://ross.abutalabs.com/products/csm
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2025-05-27T12:21:52+00:00

## Health v2 (maintenance only)
Score: 30/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 23, release rhythm 35, longevity 39
- inputs: {"age_days": 553, "days_push": 463, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 14720, forks 1478 (observed 2026-08-28T04:11:07.542405+00:00)

## What it is
CSM (Conversational Speech Model) is Sesame's speech generation model that produces conversational audio from text and audio context, using a Llama backbone with a Mimi audio decoder. The repository provides Python code for loading the 1B checkpoint and generating speech, including context-aware multi-speaker conversations.

## Use cases
- generate natural-sounding speech from text
- build a conversational voice assistant
- create multi-speaker dialogue audio with context prompts
- clone speaker identity from audio context segments
- integrate TTS into Python applications via Hugging Face Transformers

## When to choose
- you need context-aware, conversational-quality speech generation
- you have a CUDA GPU and want a Python API for TTS
- you want an open-weights 1B speech model with Llama backbone

## When to avoid
- you need production TTS without GPU hardware
- you need non-English speech or fine-grained voice control beyond context prompting
- you need a lightweight CPU-only text-to-speech solution

## Facets
- artifact type: library
- maturity: active
- function: tts, speech-recognition, machine-learning, deep-learning, llm-inference
- domain: speech-processing, artificial-intelligence, large-language-models
- platform: python, windows
- tags: conversational-speech, voice-generation, llama-backbone, rvq-audio-codes, mimi-codec, huggingface, audio, gpu, linux, macos

## Member repositories
- SesameAILabs/csm (main) score 30

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:07.542405+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:07:14.849184+00:00, confidence not recorded.
  - readme: https://github.com/SesameAILabs/csm (fetched 2026-08-28T04:11:07.542405+00:00, sha 599f8c4e3dc0)
- Data as of 2026-08-30T08:39:29.467469+00:00.