chatterbox vs csm
Open-source speech generation projects compared
chatterbox and csm both provide Python-based models for generating speech from text. chatterbox offers several model variants, including multilingual and lower-compute options, while csm focuses on conversational speech shaped by optional audio and transcript context.
chatterbox: Generate Speech with Resemble AI TTS Models
Chatterbox is Resemble AI’s open-source text-to-speech model family, with multilingual voice cloning and smaller English models for lower-latency or CPU use. It is a Python project suited to developers building speech generation into applications and workflows.
csm: Generate Conversational Speech from Text and Audio
CSM is Sesame’s speech-generation model, producing audio from text and optional conversation context. It suits developers building voice experiences who can run large models on a CUDA-compatible GPU and provide the required Hugging Face checkpoints.
| chatterbox | csm | |
|---|---|---|
| Language | Python | Python |
| License | MIT | Apache-2.0 |
| Stars | 27.1k | 14.7k |
| Forks | 3.6k | 1.5k |
| Last analyzed | Oct 21, 2026 | Oct 3, 2026 |
Key differences
- chatterbox supports voice cloning from reference audio, while csm generates speech using selected speaker identities and conversational context.
- chatterbox has models for multilingual generation and lower-compute English use; csm runs its documented 1B model and notes that non-English performance is likely limited.
- chatterbox examples show CUDA, CPU, and MPS options, while csm documents a requirement for a CUDA-compatible GPU.
- chatterbox is licensed under MIT, while csm is licensed under Apache-2.0.
- chatterbox includes expressive controls and tags such as [laugh] for some models; csm emphasizes conditioning generation on prior utterances and multiple speaker identities.
- chatterbox reports 26.7k stars and 3.6k forks, compared with csm's 14.7k stars and 1.5k forks.
Choose chatterbox if you…
- need multilingual speech generation or cross-language voice cloning.
- want to compare English, multilingual, and smaller CPU-oriented model variants.
- need generation controls or expressive tags for speech output.
Choose csm if you…
- want to condition spoken responses on transcript and audio context.
- need speech generation with multiple speaker identities in conversation.
- can run the documented model on a CUDA-compatible GPU and pair it with a separate text-generating model.
Compare the numbers side by side, or add more projects
This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.