---
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](https://github.com/s-sakti/data_indsp_digit_cdsr)

### NusaCatalogue

For easy indexing and metadata: [https://indonlp.github.io/nusa-catalogue](https://indonlp.github.io/nusa-catalogue)