InstructS-2S-200K

The 'InstructS2S-200K' dataset is used to evaluate the performance of spoken dialogue models in generating responses based on user instructions in the presence of interference from overlapping speakers.

Papers using InstructS-2S-200K (3)

IRAF: Interference-Resilient Adaptive Fusion for Noise-Robust End-to-End Full-Duplex Spoken Dialogue Systems2026Llama-omni: Seamless Speech Interaction With Large Language Models2024 · 2 citesLLaMA-Omni: Seamless Speech Interaction with Large Language Models2024 · 2 cites