9. IPSJ SIG-SLP Corpora and Environments for Noisy Speech Recognition
9-b. Noisy Speech Detection Evaluation Environment (CENSREC-1-C)
Data DOI
https://doi.org/10.32130/src.CENSREC-1-C
Producer, Project
Noisy Speech Recognition Evaluation Working Group,
Special Interest Group on Spoken Language Information Processing,
Information Processing Society of Japan (IPSJ)
Contents
Common platform for evaluating independently speech recognition accuracy and speech interval detection under noisy environment.
The data contained in CENSREC-1-C is constructed by concatenating the digit string which is the same as in AURORA-2J.
- Simulation data
- Set A: Subway, Babble, Car, Exhibition
- Set B: Restaurant, Street, Airport, Station
- Recording data in real environments
- the crowded university restaurant
- the vicinity of a highway mainline
Speaker
- Simulation data:
- 104 speakers (52 males, 52 females) 5824 utterances in total
- Recording data in real environments:
- 10 speakers（5 males, 5 females）160 utterances in total
Speech file format
RAW format (8kHz, 16bit, Mono, LittleEndian)
Distribution media
1 DVD
Licensing
For research and development purposes only
Price
No fee
Speech sample for test listening
Concatenating the digit strings
Noise environment: Restaurant
Utterance: 139O-27O6571-3-4625-5O217-6080064-83692-9443-O789-0877