Speed-IA Speech Datasets and Models for Indo-Aryan languages

About Dataset

This dataset is an annotated speech corpus developed for four low-resource Indo-Aryan languages: Awadhi, Bhojpuri, Braj, and Magahi. The corpus contains approximately 18 hours of speech collected from 40 native speakers using the Karya mobile crowdsourcing platform. Data collection combined translation-based elicitation and narrative speech tasks designed using linguistic fieldwork methodologies. The dataset was organized into lifecycle and translation category, with recordings grouped in the same folder. Each folder contains audio file and corresponding Json metadata containing transcription and recording information. The Json file were parsed to extract audio filename, transcription text and folder identifier, which were consolidated into a structured table linking the audio with the transcription. The citation of the dataset is: Kumar, Ritesh; Singh, Siddharth; Ratan, Shyam; Raj, Mohit; Sinha, Sonal; Lahiri, Bornini; Seshadri, Vivek; Bali, Kalika; and Ojha, Atul Kr. (2022), 'Annotated Speech Corpus for Low Resource Indian Languages: Awadhi, Bhojpuri, Braj and Magahi', Proceedings of Speech for Social Good Workshop, Interspeech 2022. This dataset was identified and facilitated for onboarding as part of the Dataset Onboarding Support Team (DOST) initiative led by CivicDataLab (CDL), partnering with the Gates Foundation in collaboration with BHASHINI. CivicDataLab provided support for dataset discovery, metadata preparation and onboarding facilitation. All dataset ownership and intellectual property rights remain with the original author(s).