9. IPSJ SIG-SLP Corpora and Environments for Noisy Speech Recognition

9-e. In-car Isolated Word Data and Environment for Noisy Speech Recognition (CENSREC-3)

Data DOI


Producer, Project

Noisy Speech Recognition Evaluation Working Group,

Special Interest Group on Spoken Language Information Processing,

Information Processing Society of Japan (IPSJ)


Common platform for evaluating independently speech recognition accuracy and speech interval detection under noisy environment.

50 isolated word recognition in real driving car environments.


Training data:
293 speakers (202 males, 91 females) 14050 utterances with each microphone
Test data:
18 speakers (8 males, 10 females) 14216 utterances with each microphone

Speech file format

RAW format (16kHz, 16bit, Mono, LittleEndian)

Distribution media



For research and development purposes only


No fee


Those who use Phonetically Balanced Sentences recorded by the remote (hands-free) microphpone in the training data need to purchase another DVD from Nagoya Industrial Science Research Institute (21 600 yen for universities and 108 000 yen for companies).

Please let us know if you wish to use these data.

(update: July 2017)

Speech sample for test listening

Isolated words recorded in real car driving environments

Utterance: デジタルロッカー (digital locker)
close-talk microphone remote microphone
Low-speed driving
High-speed driving

Go to corpora list