On the evening of April 8, Baidu official WeChat announced that ERNIE Bot officially launched its voice customization function. It is reported that Baidu's new voice synthesis technology can replicate people's voices in just 2 seconds. It is based on offline personalized work, relying on the large model of text and speech synthesis through extensive speech training, allowing AI to truly understand the correspondence between text and sound. Combined with the large model Prompt technology, it does not require fine-tuning and can quickly generate natural and smooth personalized synthesized sound in a zero shot manner.