Block Releases Local Voice Model With Human-Like Expressions

内容摘要

Financial technology firm Block has introduced a new voice model developed within its internal research lab. The company describes the system as capable of running locally on consumer hardware while producing speech that mimics human expressions, including laughs, pauses, and cries. According to Block, the model was trained entirely on human speech data using a single rack of NVIDIA graphics processors, distinguishing it from many existing text-to-speech solutions that rely on intermediate text processing.

本文根据公开信息编译与整理，提供主题摘要及词汇辅助阅读。