Using Resources from a Closely-related Language to Develop ASR for a Very Under-resourced Language: A Case Study for Iban

Abstract : This paper presents our strategies for developing an automatic speech recognition system for Iban, an under-resourced language. We faced several challenges such as no pronunciation dictionary and lack of training material for building acoustic models. To overcome these problems, we proposed approaches which exploit resources from a closely-related language (Malay). We developed a semi-supervised method for building the pronunciation dictionary and applied cross-lingual strategies for improving acoustic models trained with very limited training data. Both approaches displayed very encouraging results, which show that data from a closely-related language, if available, can be exploited to build ASR for a new language. In the final part of the paper, we present a zero-shot ASR using Malay resources that can be used as an alternative method for transcribing Iban speech.
Document type :
Poster communications
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-02015501
Contributor : Benjamin Lecouteux <>
Submitted on : Tuesday, February 12, 2019 - 11:11:37 AM
Last modification on : Monday, July 8, 2019 - 3:09:38 PM
Long-term archiving on : Monday, May 13, 2019 - 2:49:39 PM

File

poster_IS2015.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02015501, version 1

Citation

Sarah Samson, Laurent Besacier, Benjamin Lecouteux, Mohamed Dyab. Using Resources from a Closely-related Language to Develop ASR for a Very Under-resourced Language: A Case Study for Iban. Interspeech 2015, Sep 2015, Dresden, Germany. 2015. ⟨hal-02015501⟩

Share

Metrics

Record views

13

Files downloads

14