Block's Dessa AI Model Trained on Human Speech

Block's new AI model, Dessa, was trained entirely on human speech.

Block has introduced its latest AI model, Dessa, which was developed using only human speech data. This training approach aims to capture emotions that text-based models often miss. Coverage from outlets like Mashable, Gizmodo, and The Decoder highlights Dessa's ability to produce "uncanny, human-ish voices."

The model's unique training methodology is a key focus in the reporting, with multiple sources emphasizing the "entirely on human speech" aspect. This specific training is intended to allow Dessa to better understand and replicate the nuances of human emotion, a capability that distinguishes it from AI models primarily trained on text.

The development of Dessa by Block represents an advancement in AI's ability to process and generate speech with emotional depth. By focusing on human speech for its training, Block aims to create a more sophisticated and emotionally aware AI model, potentially leading to more natural and expressive AI-generated audio.

Related stories

Find the right AI tool for your business

Chat with Insta and get matched to the right tool in seconds.