LAHAJA: A Robust Multi-accent Benchmark for Evaluating Hindi ASR Systems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Javed, Tahir, Nawale, Janki, Joshi, Sakshi, George, Eldho, Bhogale, Kaushal, Mehendale, Deovrat, Khapra, Mitesh M.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910571845124096
author Javed, Tahir
Nawale, Janki
Joshi, Sakshi
George, Eldho
Bhogale, Kaushal
Mehendale, Deovrat
Khapra, Mitesh M.
author_facet Javed, Tahir
Nawale, Janki
Joshi, Sakshi
George, Eldho
Bhogale, Kaushal
Mehendale, Deovrat
Khapra, Mitesh M.
contents Hindi, one of the most spoken language of India, exhibits a diverse array of accents due to its usage among individuals from diverse linguistic origins. To enable a robust evaluation of Hindi ASR systems on multiple accents, we create a benchmark, LAHAJA, which contains read and extempore speech on a diverse set of topics and use cases, with a total of 12.5 hours of Hindi audio, sourced from 132 speakers spanning 83 districts of India. We evaluate existing open-source and commercial models on LAHAJA and find their performance to be poor. We then train models using different datasets and find that our model trained on multilingual data with good speaker diversity outperforms existing models by a significant margin. We also present a fine-grained analysis which shows that the performance declines for speakers from North-East and South India, especially with content heavy in named entities and specialized terminology.
format Preprint
id arxiv_https___arxiv_org_abs_2408_11440
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LAHAJA: A Robust Multi-accent Benchmark for Evaluating Hindi ASR Systems
Javed, Tahir
Nawale, Janki
Joshi, Sakshi
George, Eldho
Bhogale, Kaushal
Mehendale, Deovrat
Khapra, Mitesh M.
Computation and Language
Hindi, one of the most spoken language of India, exhibits a diverse array of accents due to its usage among individuals from diverse linguistic origins. To enable a robust evaluation of Hindi ASR systems on multiple accents, we create a benchmark, LAHAJA, which contains read and extempore speech on a diverse set of topics and use cases, with a total of 12.5 hours of Hindi audio, sourced from 132 speakers spanning 83 districts of India. We evaluate existing open-source and commercial models on LAHAJA and find their performance to be poor. We then train models using different datasets and find that our model trained on multilingual data with good speaker diversity outperforms existing models by a significant margin. We also present a fine-grained analysis which shows that the performance declines for speakers from North-East and South India, especially with content heavy in named entities and specialized terminology.
title LAHAJA: A Robust Multi-accent Benchmark for Evaluating Hindi ASR Systems
topic Computation and Language
url https://arxiv.org/abs/2408.11440