Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Jul 7, 2026

Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis

Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis

Published on: August 9, 2024

Modular recurrent neural networks for Mandarin syllable recognition.

S H Chen¹, Y F Liao

¹Department of Communication Engineering, National Chiao Tung University, Hsinchu, Taiwan, R.O.C.

IEEE Transactions on Neural Networks

|February 8, 2008

Summary

This study introduces a modular recurrent neural network (MRNN) for Mandarin speech recognition, effectively distinguishing 1280 confusable syllables. The novel approach improves accuracy and efficiency compared to traditional methods.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

[The use of central volaris forearm flap in oral and maxillofacial reconstruction]

Shanghai kou qiang yi xue = Shanghai journal of stomatology·2004

Same author

Observation of a re-entrant kinetic glass transition in a micellar system with temperature-dependent attractive interaction.

The European physical journal. E, Soft matter·2004

Same author

Antibody-mediated endothelial cell damage via nitric oxide.

Current pharmaceutical design·2004

Same author

Primer Design Assistant (PDA): A web-based primer design tool.

Nucleic acids research·2003

Same author

[The development of cerebral circulation analyzer].

Zhongguo yi liao qi xie za zhi = Chinese journal of medical instrumentation·2003

Same author

Gene sequencing for identification of Paragonimus eggs from a human case.

Zhongguo ji sheng chong xue yu ji sheng chong bing za zhi = Chinese journal of parasitology & parasitic diseases·2003

Area of Science:

Artificial Intelligence
Speech Processing
Machine Learning

Background:

Recognizing large vocabularies of highly confusable Mandarin syllables presents significant challenges.
Existing Artificial Neural Network (ANN) approaches face issues with time-alignment and scaling for large-vocabulary speech recognition.

Purpose of the Study:

To propose a novel modular recurrent neural network (MRNN)-based speech-recognition method.
To effectively recognize the entire vocabulary of 1280 highly confusable Mandarin syllables.
To address time-alignment and scaling problems in ANN-based large-vocabulary speech recognition.

Main Methods:

The proposed method decomposes the complex speech recognition task into simpler subtasks (subsyllable and tone discrimination).
Two weighting Recurrent Neural Networks (RNNs) generate dynamic weighting functions to integrate sub-solutions.

More Related Videos

Examining Online Syntactic Processing of Spoken Complex Sentences in Chinese Using Dual-Modal Interference Tasks

Examining Online Syntactic Processing of Spoken Complex Sentences in Chinese Using Dual-Modal Interference Tasks

Published on: September 5, 2019

Related Experiment Videos

Last Updated: Jul 7, 2026

Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis

Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis

Published on: August 9, 2024

Examining Online Syntactic Processing of Spoken Complex Sentences in Chinese Using Dual-Modal Interference Tasks

Examining Online Syntactic Processing of Spoken Complex Sentences in Chinese Using Dual-Modal Interference Tasks

Published on: September 5, 2019

The MRNN architecture incorporates a priori linguistic knowledge of Mandarin syllable initial-final structures.

Main Results:

The MRNN method demonstrates effectiveness and efficiency in discriminating highly confusable Mandarin syllables.
Experimental results show the proposed MRNN method outperforms an advanced Hidden Markov Model (HMM) method.
Extensions like reverse-time MRNN (Rev-MRNN) and bidirectional MRNN (Bi-MRNN) also show superior performance.

Conclusions:

The developed MRNN-based speech recognition method successfully handles large vocabularies of confusable Mandarin syllables.
The approach offers improved recognition rates and reduced system complexity compared to HMMs.
Incorporating linguistic knowledge into the MRNN architecture is key to its success.