mlsanityData to 14 Oct 2026

Towards Building ASR Systems for the Next Billion Users

arXiv 2111.03945 ↗First posted 6 Nov 2021cs.CL cs.SD eess.AScited by 47

Not on the map. Why?

Abstract (arXiv)

Recent methods in speech and language technology pretrain very LARGE models which are fine-tuned for specific tasks. However, the benefits of such LARGE models are often limited to a few resource rich languages of the world. In this work, we make multiple contributions towards building ASR systems for low resource languages from the Indian subcontinent. First, we curate 17,000 hours of raw speech data for 40 Indian languages from a wide variety of domains including education, news, technology, and finance. Second, using this raw speech data we pretrain several variants of wav2vec style models for 40 Indian languages. Third, we analyze the pretrained models to find key features: codebook vectors of similar sounding phonemes are shared across languages, representations across layers are discriminative of the language family, and attention heads often pay attention within small local windows. Fourth, we fine-tune this model for downstream ASR for 9 languages and obtain state-of-the-art results on 3 public datasets, including on very low-resource languages such as Sinhala and Nepali. Our work establishes that multilingual pretraining is an effective strategy for building ASR systems for the linguistically diverse speakers of the Indian subcontinent. Our code, data and models are available publicly at https://indicnlp.ai4bharat.org/indicwav2vec/ and we hope they will help advance research in ASR for Indic languages.

Key facts. Cited by 47 papers in the map (data through 14 Oct 2026). Builds on 23 of its 44 references that are in the map. First posted on arXiv 6 Nov 2021.

Builds on

23 of 44 references.

References · 23

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2019Unsupervised Cross-lingual Representation Learning at Scalecited by 3,935
  3. 2020wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representationscited by 3,905
  4. 2019fairseq: A Fast, Extensible Toolkit for Sequence Modelingcited by 1,488
  5. 2019SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognitioncited by 1,380
  6. 2019Cross-lingual Language Model Pretrainingcited by 1,090
  7. 2020Unsupervised Cross-lingual Representation Learning for Speech Recognitioncited by 501
  8. 2021VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation Learning, Semi-Supervised Learning and Interpretationcited by 463
  9. 2020Multilingual Translation with Extensible Multilingual Pretraining and Finetuningcited by 287
  10. 2019Massively Multilingual Neural Machine Translation in the Wild: Findings and Challengescited by 257
  11. 2019Pay Less Attention with Lightweight and Dynamic Convolutionscited by 211
  12. 2021Samanantar: The Largest Publicly Available Parallel Corpora Collection for 11 Indic Languagescited by 136
11 more
  1. 2019End-to-end ASR: from Supervised to Semi-Supervised Learning with Modern Architecturescited by 120
  2. 2020Iterative Pseudo-Labeling for Speech Recognitioncited by 83
  3. 2021UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Datacited by 77
  4. 2019Effectiveness of self-supervised pre-training for speech recognitioncited by 72
  5. 2021Multilingual and code-switching ASR challenges for low resource Indian languagescited by 53
  6. 2019Transformers with convolutional context for ASRcited by 44
  7. 2018wav2letter++: The Fastest Open-source Speech Recognition Systemcited by 42
  8. 2021CLSRIL-23: Cross Lingual Speech Representations for Indic Languagescited by 22
  9. 2021The heads hypothesis: A unifying statistical approach towards understanding multi-headed attention in BERTcited by 15
  10. 2017Letter-Based Speech Recognition with Gated ConvNetscited by 6
  11. 2020Utilizing Language Relatedness to improve Machine Translation: A Case Study on Languages of the Indian Subcontinentcited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 10; 2023: 10; 2024: 3; 2025: 7; 2026: 17.

47 papers.

Sort:

By area:

  1. 2023Scaling Speech Technology to 1,000+ Languagescited by 411
  2. 2022IndicSUPERB: A Speech Processing Universal Performance Benchmark for Indian languagescited by 40
  3. 2023AfriSpeech-200: Pan-African Accented Speech Dataset for Clinical and General Domain ASRcited by 35
  4. 2022Towards Building Text-To-Speech Systems for the Next Billion Userscited by 27
  5. 2022Vision+X: A Survey on Multimodal Learning in the Light of Datacited by 22
  6. 2022A Survey of Multilingual Models for Automatic Speech Recognitioncited by 21
  7. 2023Vistaar: Diverse Benchmarks and Training Sets for Indian Language ASRcited by 21
  8. 2022Effectiveness of Mining Audio and Text Pairs from Public Data for Improving ASR Systems for Low-Resource Languagescited by 16
  9. 2022BLASER: A Text-Free Speech-to-Speech Translation Evaluation Metriccited by 16
  10. 2024Predicting positive transfer for improved low-resource speech recognition using acoustic pseudo-tokenscited by 11
  11. 2023AfriNames: Most ASR models "butcher" African Namescited by 7
  12. 2022Accidental Learners: Spoken Language Identification in Multilingual Self-Supervised Modelscited by 6
35 more
  1. 2024Empowering Low-Resource Language ASR via Large-Scale Pseudo Labelingcited by 5
  2. 2026Towards Orthographically-Informed Evaluation of Speech Recognition Systems for Indian Languagescited by 5
  3. 2023Label Aware Speech Representation Learning For Language Identificationcited by 4
  4. 2024CoSTA: Code-Switched Speech Translation using Aligned Speech-Text Interleavingcited by 4
  5. 2023Textless Speech-to-Speech Translation With Limited Parallel Datacited by 3
  6. 2025BanglaTalk: Towards Real-Time Speech Assistance for Bengali Regional Dialectscited by 3
  7. 2022Analyzing the factors affecting usefulness of Self-Supervised Pre-trained Representations for Speech Recognitioncited by 2
  8. 2022An Automatic Speech Recognition System for Bengali Language based on Wav2Vec2 and Transfer Learningcited by 2
  9. 2025Vedavani: A Benchmark Corpus for ASR on Vedic Sanskrit Poetrycited by 2
  10. 2025Beyond WER: Probing Whisper's Sub-token Decoder Across Diverse Language Resource Levelscited by 2
  11. 2026Benchmarking Automatic Speech Recognition for Indian Languages in Agricultural Contextscited by 2
  12. 2026PSP: An Interpretable Per-Dimension Accent Benchmark for Indic Text-to-Speechcited by 1
  13. 2026Seeds Before Objectives: Rethinking Evaluation for Low-Resource Garhwali ASRcited by 1
  14. 2022WavFT: Acoustic model finetuning with labelled and unlabelled datacited by 0
  15. 2023MASR: Multi-label Aware Speech Representationcited by 0
  16. 2023"We care": Improving Code Mixed Speech Emotion Recognition in Customer-Care Conversationscited by 0
  17. 2023Multimodal Modeling For Spoken Language Identificationcited by 0
  18. 2023Temporally Aligning Long Audio Interviews with Questions: A Case Study in Multimodal Data Integrationcited by 0
  19. 2025LAPS-Diff: A Diffusion-Based Framework for Singing Voice Synthesis With Language Aware Prosody-Style Guided Learningcited by 0
  20. 2025The Limits of Data Scaling: Sub-token Utilization and Acoustic Saturation in Multilingual ASRcited by 0
  21. 2025TeluguST-46: A Benchmark Corpus and Comprehensive Evaluation for Telugu-English Speech Translationcited by 0
  22. 2025Navigating the Reality Gap: On-Device Continual Adaptation of ASR for Clinical Telephonycited by 0
  23. 2026Nw=ach=a Mun=a: A Devanagari Speech Corpus and Proximal Transfer Benchmark for Nepal Bhasha ASRcited by 0
  24. 2026ShobdoSetu: A Data-Centric Framework for Bengali Long-Form Speech Recognition and Speaker Diarizationcited by 0
  25. 2026Enhancing ASR Performance in the Medical Domain for Dravidian Languagescited by 0
  26. 2026Jointly Improving Dialect Identification and ASR in Indian Languages using Multimodal Feature Fusioncited by 0
  27. 2026Rethinking Indic AI from a Lens of Cultural Heritage Preservationcited by 0
  28. 2026Robust Assamese Speech Recognition through Controlled Fine-Tuning of Whisper Modelscited by 0
  29. 2026Staged Depth-Pruning Distillation of a Flow-Matching Text-to-Speech Teacher: A Compact Hindi Speech Synthesizercited by 0
  30. 2026Comparative Analysis of Multilingual Pre-trained Models for Nepali Automatic Speech Recognitioncited by 0
  31. 2026Model-Agnostic and Language-Agnostic Voice Pipeline Improvement for the Agriculture Domaincited by 0
  32. 2026Vimarsha: Faithful ASR Evaluation for Indian Languages with Demographic Diversity, In-the-Wild Audio and Spelling Variationscited by 0
  33. 2026BanglaKontho: Closing the Long-Form Gap in Bangla Text-to-Speechcited by 0
  34. 2026Automatic Speech Recognition for Low-Resource Sinhala: A Critical Review of Methods, Challenges, and Future Directionscited by 0
  35. 2026Anuvani-Diarization: Zero Shot Speaker Diarization and Annotation Conventions Across 22 Languagescited by 0