Back Rohan Paul

Rohan Paul· @rohanpaul_ai · X· 2026-10-30 02:07 · Original publication time AI score47

Joy introduces a dual-channel conversation dataset with 100,000 hours of speechMachine translation

Automatically verified and published · Generated and evidence-checked automatically; not reviewed by a human.

AI introduction

Joy is a dual-channel dataset of unscripted conversations between two people, each recorded on a separate track. It contains 100,000 hours of speech from 50,000 speakers across 10 languages. Its creators say it is designed to help voice models learn when to listen, interject or give a brief response.

Original post · Original

Rohan Paul@rohanpaul_ai

@shobhitbanga this is really rare stuff.

do the timing labels come per channel, or as a separate annotation file?

ReplyShobhit Banga@shobhitbanga

Drop 1/15: Introducing Joy, a dual-channel conversational dataset. Real, unscripted conversations between two people, each on their own track. Built to teach voice models how people actually talk: when to listen, when to jump in, when to say “hmm”. 100,000 hours. 50,000 speakers. 10 languages.

View this post on X ↗

02:07 · 2026-10-30 · UTC+8

Found an error? Send a correction