In this video, I look at the new Chatterbox TTS from Resemble.AI and how it's improving open-source text-to-speech with its impressive voice cloning and emotion control capabilities. We explore its features, including zero-shot voice cloning that requires only a few seconds of audio, and its unique ability to adjust the emotional intensity of speech.

Colab: https://dripl.ink/Vxs8D
Blog: https://www.resemble.ai/chatterbox/
Hugging Face Spaces: https://huggingface.co/spaces/ResembleAI/Chatterbox
Hugging Face: https://huggingface.co/ResembleAI/chatterbox
GitHub: Chatterbox-TTS-Extended https://github.com/petermg/Chatterbox-TTS-Extended

For more tutorials on using LLMs and building agents, check out my Patreon
Patreon: https://www.patreon.com/SamWitteveen
Twitter: https://x.com/Sam_Witteveen

🕵️ Interested in building LLM Agents? Fill out the form below
Building LLM Agents Form: https://drp.li/dIMes

👨‍💻Github:
https://github.com/samwit/llm-tutorials

⏱️Time Stamps:
00:00 Intro
00:24 Resemble.AI - Chatterbox
01:53 Samples
04:53 Hugging Face: Chatterbox
05:22 Demo
06:26 Adding Exaggeration
08:56 Voice Cloning
13:00 Chatterbox TTS Extended Github
14:07 Hugging Face: Chatterbox GGUF

Sam Witteveen AI is a publication offering insights, tutorials, and resources for artificial intelligence (AI) enthusiasts and practitioners. Readers can learn about machine learning algorithms, deep learning frameworks, and AI applications. With tutorials, case studies, and expert interviews, Sam Witteveen AI provides  guidance and expertise for building and deploying AI solutions.

Sam Witteveen

Chatterbox TTS is a new open-source text-to-speech model by Resemble AI that offers voice cloning capabilities with just 5 seconds of reference audio. The 500-million parameter model features emotion control, watermarking for fake audio detection, and can be easily installed via pip. Unlike proprietary solutions like ElevenLabs, it runs locally without per-token costs. The model includes exaggeration controls for emotional expression and classifier-free guidance weights for speed control, though it's limited to about 40 seconds of audio generation per request.

Building with Chatterbox TTS and Voice Cloning