Recognizing Every Voice: Towards Inclusive ASR for Rural Bhojpuri Women

2025/06/11 by Sakshi Joshi, Joshi, Sakshi, Eldho Ittan George +9 · 2 citations
Computer Science · #AI in Service Interactions #Audio and Speech Processing (eess.AS) #FOS: Electrical engineering #ICT in Developing Communities #Speech Recognition and Synthesis #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2506.09653

openalex publication_date 2025/06/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Digital inclusion remains a challenge for marginalized communities, especially rural women in low-resource language regions like Bhojpuri. Voice-based access to agricultural services, financial transactions, government schemes, and healthcare is vital for their empowerment, yet existing ASR systems for this group remain largely untested. To address this gap, we create SRUTI ,a benchmark consisting of rural Bhojpuri women speakers. Evaluation of current ASR models on SRUTI shows poor performance due to data scarcity, which is difficult to overcome due to social and cultural barriers that hinder large-scale data collection. To overcome this, we propose generating synthetic speech using just 25-30 seconds of audio per speaker from approximately 100 rural women. Augmenting existing datasets with this synthetic data achieves an improvement of 4.7 WER, providing a scalable, minimally intrusive solution to enhance ASR and promote digital inclusion in low-resource language.

Citations

• Enhancing Low-Resource ASR through Versatile TTS: Bridging the Data Gap
• F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching
• Rasa: Building Expressive Speech Synthesis Systems for Indian Languages in Low-resource Settings
• IndicVoices: Towards building an Inclusive Multilingual Speech Dataset for Indian Languages
• Stateful Conformer with Cache-based Inference for Streaming Automatic Speech Recognition
• SPRING-INX: A Multilingual Indian Language Speech Corpus by SPRING Lab, IIT Madras
• IndicTrans2: Towards High-Quality and Accessible Machine Translation Models for all 22 Scheduled Indian Languages
• Text Generation with Speech Synthesis for ASR Data Augmentation
• Effectiveness of Mining Audio and Text Pairs from Public Data for Improving ASR Systems for Low-Resource Languages
• IndicSUPERB: A Speech Processing Universal Performance Benchmark for Indian languages
• No Language Left Behind: Scaling Human-Centered Machine Translation
• Annotated Speech Corpus for Low Resource Indian Languages: Awadhi, Bhojpuri, Braj and Magahi
• ASR data augmentation in low-resource settings using cross-lingual multi-speaker TTS and cross-lingual voice conversion
• Conformer: Convolution-augmented Transformer for Speech Recognition
• Conformer: Convolution-augmented Transformer for Speech Recognition
• The IIT Bombay English-Hindi Parallel Corpus

Cited by

• Towards Fair ASR For Second Language Speakers Using Fairness Prompted Finetuning
• Synthetic Voice Data for Automatic Speech Recognition in African Languages

Related

• Model Adaptation for ASR in low-resource Indian Languages
• IndicVoices: Towards building an Inclusive Multilingual Speech Dataset for Indian Languages
• Towards Building ASR Systems for the Next Billion Users
• Bangla-Wave: Improving Bangla Automatic Speech Recognition Utilizing N-gram Language Models
• Bengali Common Voice Speech Dataset for Automatic Speech Recognition