Speech Resynthesis

• Speech Resynthesis from Discrete Disentangled Self-Supervised Representations
• self-supervised discrete representations for the task of speech resynthesis
• separately extract low-bitrate representations for speech content, prosodic information, and speaker identity
• This allows to synthesize speech in a controllable manner
• evaluate the F0 reconstruction, speaker identification performance (for both resynthesis and voice conversion), recordings’ intelligibility, and overall quality using subjective human evaluation
• ultra-lightweight speech codec