Resemble AI Releases Dramabox Voice Cloning TTS Model

Resemble AI has publicly released Dramabox, a new text-to-speech (TTS) model designed for generating expressive and high-quality audio. The model's standout feature is its ability to perform one-shot voice cloning, replicating a speaker's voice from just a single short audio clip.

Under the hood, Dramabox employs a diffusion-transformer architecture. According to the company, this approach is built upon their LTX-2 flow-matching audio technology, which enables fine-grained control over the generated speech. This allows the model to produce not just clear narration but also audio with emotional nuance and expressiveness, a key challenge in speech synthesis.

Availability and License

Developers and researchers can access the model weights and inference code on the Hugging Face Hub. It's important to note that Dramabox is released under a custom Community License. This license permits non-commercial use and research, but requires a separate commercial license for any business applications.

The release of Dramabox provides the open-weights community with a powerful tool for creative and research-oriented audio projects. Its combination of a modern architecture and effective voice cloning makes it a significant new entry in the landscape of publicly available TTS models, offering a high-quality foundation for non-commercial applications.