tags:
- speech-recognition
language:
- ind
indspeech_digit_cdsr
INDspeech_DIGIT_CDSR is the first Indonesian speech dataset for connected digit speech recognition (CDSR). The data was developed by TELKOMRisTI (R&D Division, PT Telekomunikasi Indonesia) in collaboration with Advanced Telecommunication Research Institute International (ATR) Japan and Bandung Institute of Technology (ITB) under the Asia-Pacific Telecommunity (APT) project in 2004 [Sakti et al., 2004]. Although it was originally developed for a telecommunication system for hearing and speaking impaired people, it can be used for other applications, i.e., automatic call centers that recognize telephone numbers.
Dataset Usage
Run pip install nusacrowd before loading the dataset through HuggingFace's load_dataset.
Citation
@inproceedings{sakti-icslp-2004,
title = "Indonesian Speech Recognition for Hearing and Speaking Impaired People",
author = "Sakti, Sakriani and Hutagaol, Paulus and Arman, Arry Akhmad and Nakamura, Satoshi",
booktitle = "Proc. International Conference on Spoken Language Processing (INTERSPEECH - ICSLP)",
year = "2004",
pages = "1037--1040"
address = "Jeju Island, Korea"
}
License
CC-BY-NC-SA-4.0
Homepage
https://github.com/s-sakti/data_indsp_digit_cdsr
NusaCatalogue
For easy indexing and metadata: https://indonlp.github.io/nusa-catalogue