ELRA - ELRA-U-S0287 : COSINE Corpus

You are here » Universal Catalogue » Spoken Resources » Desktop/microphone

Language Resources

Search Catalogue

Send us information

Would you like to collaborate ?
Contact Us

Languages

Catalog Reference : ELRA-U-S0287

COSINE Corpus

The COnversational Speech In Noisy Environments (COSINE) corpus consists of multi-party conversations recorded in a real world environment with background noise.

The corpus includes 33 recording sessions of 45 to 90 minutes each. The conversations (between 2 and 7 participants) were recorded through a 7-channel recording system worn by each speaker: a 4-channel array in front of the speaker's chest, a throat microphone, a shoulder-mounted microphone and a close-talking microphone in front of the mouth.

From the 150 hours of recordings, 42.5 hours are transcribed (at the word level), including 9.5 hours of speech and 33 hours of non-speech.

Such a database can be used to train noise-robust speech recognition systems.

Production

Creation date : 2009

Contents

Click on the arrow to display content.

speech corpus
Language(s) : English (USA)
Duration : 150 hours
Quantisation : 16 bit
Source Channel : Microphone
Transcription Entries : Orthographic
Transcription Segmentation Level : Partial