Shinnosuke Takamichi

Core-set selection improves text-to-speech models by focusing on diverse data.

This study examines if learned speech symbols mimic word frequency patterns.

Coco-Nut offers diverse Japanese voice samples for advanced text-to-speech applications.

A study on improving TTS systems with diverse voice samples.

RALL-E enhances text-to-speech synthesis for clearer, more natural speech.

Introducing spatial voice conversion to enhance audio realism and immersion.

This study examines how voice preferences vary among different listeners.

Researchers explore textless approaches for better understanding of spoken language.

J-CHAT provides a large, open-source dataset for enhancing spoken dialogue systems.

Researchers develop SaSLaW to enhance machine speech adaptation in various environments.

BigCodec improves sound quality in low-bitrate audio transmission.

A new method enhances synthesized ensemble singing by modeling singer interactions.