Two-stage training for Chinese dialect recognition

Zongze Ren; Guofu Yang; Shugong Xu

doi:10.21437/Interspeech.2019-1522

Two-stage training for Chinese dialect recognition

Zongze Ren, Guofu Yang, Shugong Xu

Shanghai University

Research output: Contribution to journal › Conference article › peer-review

16 Citations (Scopus)

Abstract

In this paper, we present a two-stage language identification (LID) system based on a shallow ResNet14 followed by a simple 2-layer recurrent neural network (RNN) architecture, which was used for Xunfei (iFlyTek) Chinese Dialect Recognition Challenge¹ and won the first place among 110 teams. The system trains an acoustic model (AM) firstly with connectionist temporal classification (CTC) to recognize the given phonetic sequence annotation and then train another RNN to classify dialect category by utilizing the intermediate features as inputs from the AM. Compared with a three-stage system we further explore, our results show that the two-stage system can achieve high accuracy for Chinese dialects recognition under both short utterance and long utterance conditions with less training time.

Original language	English
Pages (from-to)	4050-4054
Number of pages	5
Journal	Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH
Volume	2019-September
DOIs	https://doi.org/10.21437/Interspeech.2019-1522
Publication status	Published - 2019
Externally published	Yes
Event	20th Annual Conference of the International Speech Communication Association: Crossroads of Speech and Language, INTERSPEECH 2019 - Graz, Austria Duration: 15 Sept 2019 → 19 Sept 2019

Keywords

Acoustic model
Convolutional recurrent neural network
Dialect recognition

Access to Document

10.21437/Interspeech.2019-1522

Cite this

@article{d6993872dcc6414599860d429b0c1ed3,

title = "Two-stage training for Chinese dialect recognition",

abstract = "In this paper, we present a two-stage language identification (LID) system based on a shallow ResNet14 followed by a simple 2-layer recurrent neural network (RNN) architecture, which was used for Xunfei (iFlyTek) Chinese Dialect Recognition Challenge1 and won the first place among 110 teams. The system trains an acoustic model (AM) firstly with connectionist temporal classification (CTC) to recognize the given phonetic sequence annotation and then train another RNN to classify dialect category by utilizing the intermediate features as inputs from the AM. Compared with a three-stage system we further explore, our results show that the two-stage system can achieve high accuracy for Chinese dialects recognition under both short utterance and long utterance conditions with less training time.",

keywords = "Acoustic model, Convolutional recurrent neural network, Dialect recognition",

author = "Zongze Ren and Guofu Yang and Shugong Xu",

note = "Publisher Copyright: Copyright {\textcopyright} 2019 ISCA; 20th Annual Conference of the International Speech Communication Association: Crossroads of Speech and Language, INTERSPEECH 2019 ; Conference date: 15-09-2019 Through 19-09-2019",

year = "2019",

doi = "10.21437/Interspeech.2019-1522",

language = "English",

volume = "2019-September",

pages = "4050--4054",

journal = "Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH",

issn = "2308-457X",

}

TY - JOUR

T1 - Two-stage training for Chinese dialect recognition

AU - Ren, Zongze

AU - Yang, Guofu

AU - Xu, Shugong

PY - 2019

Y1 - 2019

N2 - In this paper, we present a two-stage language identification (LID) system based on a shallow ResNet14 followed by a simple 2-layer recurrent neural network (RNN) architecture, which was used for Xunfei (iFlyTek) Chinese Dialect Recognition Challenge1 and won the first place among 110 teams. The system trains an acoustic model (AM) firstly with connectionist temporal classification (CTC) to recognize the given phonetic sequence annotation and then train another RNN to classify dialect category by utilizing the intermediate features as inputs from the AM. Compared with a three-stage system we further explore, our results show that the two-stage system can achieve high accuracy for Chinese dialects recognition under both short utterance and long utterance conditions with less training time.

AB - In this paper, we present a two-stage language identification (LID) system based on a shallow ResNet14 followed by a simple 2-layer recurrent neural network (RNN) architecture, which was used for Xunfei (iFlyTek) Chinese Dialect Recognition Challenge1 and won the first place among 110 teams. The system trains an acoustic model (AM) firstly with connectionist temporal classification (CTC) to recognize the given phonetic sequence annotation and then train another RNN to classify dialect category by utilizing the intermediate features as inputs from the AM. Compared with a three-stage system we further explore, our results show that the two-stage system can achieve high accuracy for Chinese dialects recognition under both short utterance and long utterance conditions with less training time.

KW - Acoustic model

KW - Convolutional recurrent neural network

KW - Dialect recognition

UR - http://www.scopus.com/inward/record.url?scp=85074713961&partnerID=8YFLogxK

U2 - 10.21437/Interspeech.2019-1522

DO - 10.21437/Interspeech.2019-1522

M3 - Conference article

AN - SCOPUS:85074713961

SN - 2308-457X

VL - 2019-September

SP - 4050

EP - 4054

JO - Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH

JF - Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH

T2 - 20th Annual Conference of the International Speech Communication Association: Crossroads of Speech and Language, INTERSPEECH 2019

Y2 - 15 September 2019 through 19 September 2019

ER -

Two-stage training for Chinese dialect recognition

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this