Nima Mesgarani

Explore how the auditory cortex integrates sound over time.

Mamba shows promise against transformers in speech tasks, especially for long inputs.

Style-Talker improves conversations between humans and machines through emotional depth.

SpeechLLMs show promise but struggle with speaker identification in conversations.

StyleTTS-ZS offers efficient, high-quality speech synthesis without extensive speaker training.

Exploring the understanding of language by large language models.