Chatterbox (documentation: https://github.com/resemble-ai/chatterbox and demonstration: https://huggingface.co/spaces/ResembleAI/Chatterbox-Multilingual-TTS) is a “text to speech” model open-source available in 23 languages for its Multilingual version. Two other versions are offered only in English, TTS and Turbo with the possible addition of noises (laughter, sniffling, clucking etc.) in the generation.

In the demo area, 300-character text with the option of adding a reference voice or recording one.

Export wav.

Note: Noises may appear at the end of the generation.

Free and open-source.

Tests since Art poétique, Verlaine, https://fr.wikipedia.org/wiki/Art_po%C3%A9tique_(Verlaine)

With recorded voice: