CENSREC-3: An evaluation framework for Japanese speech recognition in real car-driving environments

被引：14

作者：

Fujimoto, Masakiyo

Takeda, Kazuya

Nakamura, Satoshi

机构：

[1] ATR, Spoken Language Commun Res Labs, Keihanna Sci City, Kyoto Fu 6190288, Japan

[2] Nagoya Univ, Nagoya, Aichi 4648603, Japan

来源：

IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS | 2006年 / E89D卷 / 11期

关键词：

noisy speech recognition; common evaluation framework; incar speech database; CENSREC-3;

D O I：

10.1093/ietisy/e89-d.11.2783

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

This paper introduces a common database, an evaluation framework. and its baseline recognition results for in-car speech recognition. CENSREC-3, as an outcome of the IPSJ-SIG SLP Noisy Speech Recognition Evaluation Working Group. CENSREC-3, which is a sequel to AURORA-2J. has been designed as the evaluation framework of isolated word recognition in real car-driving environments. Speech data were collected using two microphones, a close-talking microphone and a hands-free microphone. under 16 carefully controlled driving conditions, i.e.,combinations of three car speeds and six car conditions. CENSREC-3 provides six evaluation environments designed using speech data collected in these conditions.

引用

页码：2783 / 2793

页数：11