Methodology for designing and creating Hindi speech corpus

Magdum, Damodar; Dubey, Manisha; Patil, Tejaswini; Shah, Ronak; Belhe, Swapnil; Kulkarni, Mahesh

doi:10.1109/spaces.2015.7058279

Cited by 11 publications

(3 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…According to [11] the major challenges when it comes to speech corpus creation is segmenting audio data into sentences. [12] presented the methodology the used for designing and creating Hindi speech corpus. The methodology involved crawling text, filtering, recording and annotation phases.…”

Section: Experimental Design Materials and Methodsmentioning

confidence: 99%

Development of Hausa dataset a baseline for speech recognition

2022

View full text Add to dashboard Cite

Section: Experimental Design Materials and Methodsmentioning

confidence: 99%

Development of Hausa dataset a baseline for speech recognition

2022

View full text Add to dashboard Cite

“…Recordings were collected from old people with the help of a microphone [36]. A comprehensive study was carried out on Hindi [37]. In this study, recordings were captured in a studio environment.…”

Section: Related Workmentioning

confidence: 99%

Building a Speech and Text Corpus of Turkish: Large Corpus Collection with Initial Speech Recognition Results

Polat

Oyucu

2020

Symmetry

View full text Add to dashboard Cite

To build automatic speech recognition (ASR) systems with a low word error rate (WER), a large speech and text corpus is needed. Corpus preparation is the first step required for developing an ASR system for a language with few argument speech documents available. Turkish is a language with limited resources for ASR. Therefore, development of a symmetric Turkish transcribed speech corpus according to the high resources languages corpora is crucial for improving and promoting Turkish speech recognition activities. In this study, we constructed a viable alternative to classical transcribed corpus preparation techniques for collecting Turkish speech data. In the presented approach, three different methods were used. In the first step, subtitles, which are mainly supplied for people with hearing difficulties, were used as transcriptions for the speech utterances obtained from movies. In the second step, data were collected via a mobile application. In the third step, a transfer learning approach to the Grand National Assembly of Turkey session records (videotext) was used. We also provide the initial speech recognition results of artificial neural network and Gaussian mixture-model-based acoustic models for Turkish. For training models, the newly collected corpus and other existing corpora published by the Linguistic Data Consortium were used. In light of the test results of the other existing corpora, the current study showed the relative contribution of corpus variability in a symmetric speech recognition task. The decrease in WER after including the new corpus was more evident with increased verified data size, compensating for the status of Turkish as a low resource language. For further studies, the importance of the corpus and language model in the success of the Turkish ASR system is shown.

show abstract

“…If all these words are processed without filtering, then there may be chance that irrelevant words are part of the corpus recording script, and those words are not adding value to the system. Also invalid words are very difficult to pronounce so filtering is very important step [13]. As text corpus size is very huge so manual checking is difficult and time consuming so algorithm are developed to create phonetically balanced optimized words [8] [9].…”

Section: Speech Corpus Design and Developmentmentioning

confidence: 99%

Designing Talking ATM System for People with Visual Impairments

Magdum¹,

Patil²,

Suman³

et al. 2018

IJET

Self Cite

View full text Add to dashboard Cite

In this paper, we discuss process of design and development of talking ATM for visually impaired people. Automated Teller Machine (ATM) has become vital part of our life to perform financial transactions without intervention of human banker. ATM facilitates cash withdrawal, balance check, mini statement and fund transfer. But, these banking services using ATM cannot be directly used by some set of people of society such as people with low vision, visually impaired, illiterate as lack of accessing ATM through screens. Even they can be defrauded at ATM centers. To digitally include these set of people, talking ATMs are evolved. Talking ATM provides accessibility to ATM services by providing audio component. Many ATMs employ headphone jack that facilitates user to do transaction with security. The audio information is generated either using pre-recorded speech corpus or through speech synthesis engine. The paper summarizes how ATM works, need, proposed solution of talking ATM for visually impaired users, design and development talking ATM using concatenated Text To Speech.

show abstract

Methodology for designing and creating Hindi speech corpus

Cited by 11 publications

References 3 publications

Development of Hausa dataset a baseline for speech recognition

Development of Hausa dataset a baseline for speech recognition

Building a Speech and Text Corpus of Turkish: Large Corpus Collection with Initial Speech Recognition Results

Designing Talking ATM System for People with Visual Impairments

Contact Info

Product

Resources

About