Yang Ai

DAIEN-TTS achieves unprecedented control over speech synthesis by disentangling speaker characteristics from environmental factors, resulting in highly natural and contextually relevant audio outputs.

Achieving high-quality speech reconstruction at just 0.5 kbps could revolutionize low-bandwidth communication systems.

VoCodec achieves a remarkable 27% bitrate reduction while enhancing speech quality by intelligently allocating resources based on voicing characteristics.

Minor acoustic noise can nearly double the rate of unsafe outputs in clinical documentation, despite only a slight increase in Word Error Rate.

Seamless transitions between speech and singing modes are now driven purely by text context, achieving state-of-the-art results in code-switching synthesis.

Achieve near state-of-the-art speech coding quality at extremely low bitrates (0.65 kbps) with a surprisingly compact model.