Dataset · TheAgenticDataCompany
Open Yap 1K: 1000 Hours of Dialogues
TheAgenticDataCompany/open-yap-1k
1000 hours of dual-channel English conversations at 48 kHz for full-duplex and speech models.
What's inside
Open Yap 1K is a corpus of 1000 hours of dual-channel English conversations recorded in natural environments at 48 kHz. It is available for free for commercial and research use. A sample is uploaded to Hugging Face: 8.92 hours, 16 conversations, CC-BY-4.0 license; the full corpus is provided separately.
What you can do with it
- Training full-duplex conversational speech models
- ASR and diarization of natural dialogues
- TTS and conversational speech generation
- Tasks
- Audio → audio, Automatic speech recognition, Text → speech
- Languages
- English
- Size
- up to 1K rows
- Downloads this month
- 8.3K
- Likes
- 265
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("TheAgenticDataCompany/open-yap-1k")