Investigation on cross- and multilingual MLP features under matched and mismatched acoustical conditions

Citations Over TimeTop 10% of 2013 papers

Abstract

In this paper, Multi Layer Perceptron (MLP) based multilingual bottleneck features are investigated for acoustic modeling in three languages - German, French, and US English. We use a modified training algorithm to handle the multilingual training scenario without having to explicitly map the phonemes to a common phoneme set. Furthermore, the cross-lingual portability of bottleneck features between the three languages are also investigated. Single pass recognition experiments on large vocabulary SMS dictation task indicate that (1) multilingual bottleneck features yield significantly lower word error rates compared to standard MFCC features (2) multilingual bottleneck features are superior to monolingual bottleneck features trained for the target language with limited training data, and (3) multilingual bottleneck features are beneficial in training acoustic models in a low resource language where only mismatched training data is available-by exploiting the more matched training data from other languages.

Related Papers

• Dictation: An Effective Means of FL Teaching and Teaching Evaluation(2001)
• Research On Strengthening Dictation Practices in English Teaching(2004)
• A Study of the Psychological Mechanism and Strategies in TEM4 Dictation(2011)
• On the Application of Dictation in High School English Classroom Teaching(2007)
• → Error Analysis of TEM-4 Dictation and Teaching Suggestions to TEM-4 Dictation(2016)