Lates News

On September 12th, according to Xiaomi Technology, the new generation Kaldi team of Xiaomi Group's AI Laboratory recently released the ZipVoice series speech synthesis (TTS) models based on the Flow Matching architecture - ZipVoice (zero-sample single speaker speech synthesis model) and ZipVoice-Dialog (zero-sample dialogue speech synthesis model). ZipVoice solves the pain points of the existing zero-sample speech synthesis models with large parameter sizes and slow synthesis speeds, while ZipVoice-Dialog addresses the bottlenecks in stability and reasoning speed of existing dialogue speech synthesis models.