How does language model size effects speech recognition accuracy for the Turkish language?
In this paper we aimed at investigating the effect of Language Model (LM) size on Speech Recognition (SR) accuracy. We also provided details of our approach for obtaining the LM for Turkish. Since LM is obtained by statistical processing of raw text, we expect that by increasing the size of availabl...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Pamukkale University
2016-05-01
|
Series: | Pamukkale University Journal of Engineering Sciences |
Subjects: | |
Online Access: | https://dergipark.org.tr/tr/pub/pajes/issue/20566/219179 |
_version_ | 1797914598708346880 |
---|---|
author | Behnam Asefisaray Erhan Mengüşoğlu Hayri Sever Murat Hacıömeroğlu |
author_facet | Behnam Asefisaray Erhan Mengüşoğlu Hayri Sever Murat Hacıömeroğlu |
author_sort | Behnam Asefisaray |
collection | DOAJ |
description | In this paper we aimed at investigating the effect of Language Model (LM) size on Speech Recognition (SR) accuracy. We also provided details of our approach for obtaining the LM for Turkish. Since LM is obtained by statistical processing of raw text, we expect that by increasing the size of available data for training the LM, SR accuracy will improve. Since this study is based on recognition of Turkish, which is a highly agglutinative language, it is important to find out the appropriate size for the training data. The minimum required data size is expected to be much higher than the data needed to train a language model for a language with low level of agglutination such as English. In the experiments we also tried to adjust the Language Model Weight (LMW) and Active Token Count (ATC) parameters of LM as these are expected to be different for a highly agglutinative language. We showed that by increasing the training data size to an appropriate level, the recognition accuracy improved on the other hand changes on LMW and ATC did not have a positive effect on Turkish speech recognition accuracy. |
first_indexed | 2024-04-10T12:28:57Z |
format | Article |
id | doaj.art-79bb4dadafee4db291ed50c423977927 |
institution | Directory Open Access Journal |
issn | 1300-7009 2147-5881 |
language | English |
last_indexed | 2024-04-10T12:28:57Z |
publishDate | 2016-05-01 |
publisher | Pamukkale University |
record_format | Article |
series | Pamukkale University Journal of Engineering Sciences |
spelling | doaj.art-79bb4dadafee4db291ed50c4239779272023-02-15T16:15:00ZengPamukkale UniversityPamukkale University Journal of Engineering Sciences1300-70092147-58812016-05-01222100105218How does language model size effects speech recognition accuracy for the Turkish language?Behnam AsefisarayErhan MengüşoğluHayri SeverMurat HacıömeroğluIn this paper we aimed at investigating the effect of Language Model (LM) size on Speech Recognition (SR) accuracy. We also provided details of our approach for obtaining the LM for Turkish. Since LM is obtained by statistical processing of raw text, we expect that by increasing the size of available data for training the LM, SR accuracy will improve. Since this study is based on recognition of Turkish, which is a highly agglutinative language, it is important to find out the appropriate size for the training data. The minimum required data size is expected to be much higher than the data needed to train a language model for a language with low level of agglutination such as English. In the experiments we also tried to adjust the Language Model Weight (LMW) and Active Token Count (ATC) parameters of LM as these are expected to be different for a highly agglutinative language. We showed that by increasing the training data size to an appropriate level, the recognition accuracy improved on the other hand changes on LMW and ATC did not have a positive effect on Turkish speech recognition accuracy.https://dergipark.org.tr/tr/pub/pajes/issue/20566/219179-dil modeli ses tanıma sistemleri dil modeli ağırlığı aktif token sayısı |
spellingShingle | Behnam Asefisaray Erhan Mengüşoğlu Hayri Sever Murat Hacıömeroğlu How does language model size effects speech recognition accuracy for the Turkish language? Pamukkale University Journal of Engineering Sciences - dil modeli ses tanıma sistemleri dil modeli ağırlığı aktif token sayısı |
title | How does language model size effects speech recognition accuracy for the Turkish language? |
title_full | How does language model size effects speech recognition accuracy for the Turkish language? |
title_fullStr | How does language model size effects speech recognition accuracy for the Turkish language? |
title_full_unstemmed | How does language model size effects speech recognition accuracy for the Turkish language? |
title_short | How does language model size effects speech recognition accuracy for the Turkish language? |
title_sort | how does language model size effects speech recognition accuracy for the turkish language |
topic | - dil modeli ses tanıma sistemleri dil modeli ağırlığı aktif token sayısı |
url | https://dergipark.org.tr/tr/pub/pajes/issue/20566/219179 |
work_keys_str_mv | AT behnamasefisaray howdoeslanguagemodelsizeeffectsspeechrecognitionaccuracyfortheturkishlanguage AT erhanmengusoglu howdoeslanguagemodelsizeeffectsspeechrecognitionaccuracyfortheturkishlanguage AT hayrisever howdoeslanguagemodelsizeeffectsspeechrecognitionaccuracyfortheturkishlanguage AT murathacıomeroglu howdoeslanguagemodelsizeeffectsspeechrecognitionaccuracyfortheturkishlanguage |