Annotated Speech Corpus for Low Resource Indian Languages: Awadhi, Bhojpuri, Braj and Magahi

TL;DR

This paper presents the development of an annotated speech corpus for four low-resource Indian languages, including data collection methods, linguistic annotations, and baseline speech recognition experiments, aiming to support NLP research and community income.

Contribution

It introduces a new speech corpus for Awadhi, Bhojpuri, Braj, and Magahi, with detailed annotations and baseline ASR results, addressing resource scarcity and pandemic-related challenges.

Findings

Approximately 18 hours of speech data collected and transcribed.

Baseline automatic speech recognition systems developed for each language.

Methodology adapted for pandemic conditions to support low-income communities.

Abstract

In this paper we discuss an in-progress work on the development of a speech corpus for four low-resource Indo-Aryan languages -- Awadhi, Bhojpuri, Braj and Magahi using the field methods of linguistic data collection. The total size of the corpus currently stands at approximately 18 hours (approx. 4-5 hours each language) and it is transcribed and annotated with grammatical information such as part-of-speech tags, morphological features and Universal dependency relationships. We discuss our methodology for data collection in these languages, most of which was done in the middle of the COVID-19 pandemic, with one of the aims being to generate some additional income for low-income groups speaking these languages. In the paper, we also discuss the results of the baseline experiments for automatic speech recognition system in these languages.