Open Yap 1K: 1,000 hours of natural two-speaker conversations, free for commercial use
realmrfakename · x · 2026-09-10
The Agentic Data Company released Open Yap 1K, billed as the world's largest free conversational speech dataset: 1,000 hours (1,602 conversations, 239 speakers) of natural English two-speaker audio at 48 kHz dual-channel, free for commercial and research use. Unlike Fisher or Switchboard, speakers are friends/family talking freely, preserving interrupts, backchannels, and short turn gaps that full-duplex models must learn. Transcriptions are word-level from Deepgram Nova-3 (not human-verified), and it's part of a larger licensed corpus built with frontier labs.
More from Research
- Hallucination rate as a kill gate: replay pipeline reveals pitfalls of LLM model swaps — dl_weekly · 2026-09-10
- Joseph Suarez launches Reinforcement Learning series — jsuarez · 2026-09-10
- SimWorld Teases Code4Scene Benchmark: Prompt-to-Explorable-3D Container Port Demo — Lianhuiq · 2026-09-10
- Salesforce releases EvoHarnessBench to test agents against evolving tool harnesses — Salesforce · 2026-09-10
- AI-built app visualizes Rentosertib anti-aging drug mechanism in ~40 minutes — DeryaTR_ · 2026-09-10
- AlphaFold helps Cornell researchers discover hidden cellular 'off switch' linked to cancer — Dr_Singularity · 2026-09-10