CHiME-9

This Challenge has now ended

The CHiME-9 challenge ran between 1st July 2025 and 6th February 2026. The challenge had two sub-tasks, MCoRec and ECHI, the full details of which can be found below. Results of the challenge were presented at the HSCMA&CHiME 2026 workshop.

CHiME-9 Task 1, MCoRec: Multi-Modal Context-aware Recognition

MCoRec targets the problem of Multi-Modal Context-aware Recognition in a single-room environment. The goal is to process a single 360° video and audio recording of a room where multiple, separate conversations are happening simultaneously, and to both transcribe each speaker’s speech and identify which speakers belong to the same conversation.

Get the dataset on the MCoRec Hugging Face page.

CHiME-9 Task 2, ECHI: Enhancing Conversations to address Hearing Impairment

ECHI invites participants to tackle the challenge of Enhancing Conversations to address Hearing Impairment. We consider signal processing that enables a device, which will block out unwanted noise but while still allowing the wearer to hear their conversational partners.

Get the dataset on the ECHI Hugging Face page.