aifollow.news
Back Rohan Paul
Rohan Paul· @rohanpaul_ai · X· · Original publication time AI score47

Joy dual-channel conversation dataset introduced with 100,000 hours of unscripted dialogueMachine translation

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

AI introduction

Joy contains unscripted conversations between two people, each recorded on a separate track. The dataset totals 100,000 hours from 50,000 speakers across 10 languages. Its introducer says it is designed to help voice models learn when people listen, jump in or give brief responses.

Original post · AI translation

@rohanpaul_ai
Translation · refer to the original

无脚本通话很难大规模收集。 如今的语音 AI 主要从由大型语言模型编写、再由文本转语音系统朗读的虚构对话中学习,因此它从未听过人们相互抢话;

真想知道他们是怎么让对话持续 10 到 20 分钟的。

他们发布了 Joy,一个涵盖 10 种语言、总计 100,000 小时的无脚本双人对话数据集

https://t.co/KFRib2aMI6

QuoteShobhit Banga@shobhitbanga
Translation
第 1/15 条发布内容:介绍 Joy,一个双通道对话数据集。 两个人之间真实、无脚本的对话,每个人的声音都录在各自独立的音轨上。它旨在教会语音模型人们实际上如何交谈:什么时候倾听,什么时候插话,什么时候说“嗯”。 100,000 小时。 50,000 位说话者。 10 种语言。
Original

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 ↗

Source: Rohan Paul · x.com

Industry
Found an error? Send a correction