sulabhkatiyar/ne-asr-dataset-grt

Garo (grt) — ASR dataset

A small Garo (grt) speech-to-text dataset for automatic speech recognition (ASR) of a low-resource North-East India language. Each example pairs a short audio clip with its Romanized (Latin-script) transcript.

Source

Derived from the ARTPARK-IISc Vaani project (https://vaani.iisc.ac.in/)

Splits

Split	Samples
train	33,480
validation	4,253
test	4,101

Data fields

Each example has:

• audio — the audio recording (16 kHz mono).
• text — the Romanized (Latin-script) transcript.
• language — the ISO 639-3 language code (grt).

Used by

This dataset trains and evaluates the paired ASR model sulabhkatiyar/ne-asr-grt — a character-level CTC adapter over a frozen facebook/mms-1b-all encoder. An augmented training variant is available at sulabhkatiyar/ne-asr-dataset-grt-aug.

Part of the NE Speech AI V1 collection.

Limitations

• Garo is a low-resource language; this dataset is small and may not cover the full range of speakers, dialects, domains, and recording conditions.
• Transcriptions are in Latin / Roman script (romanized); romanization conventions can introduce label noise.

License

cc-by-4.0, following the ARTPARK-IISc Vaani source dataset (https://vaani.iisc.ac.in/), from which this data is derived. Please retain that attribution and comply with the license terms when using this dataset.