Newest papers

Monsoon: Scalable Spontaneous Speech Collection for Low-Resource ASR in 50 Languages

Shobhit Banga, Manas Dhir, Aaditya Pareek, Amritansh Walecha, Bhaskar Singh, Hanuman Sidh, Manmeet Kaur, Sagar C. Jain, Supriya Paul, Vanshika Chhabra, Utkarsh Singh · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26776 · cited by 0 works (OpenAlex) · open-access PDF

Hindi

अधिकांश भाषाओं के लिए, स्वचालित भाषण पहचान (एएसआर) मॉडल आर्किटेक्चर द्वारा कम सीमित है, जो सटीक रूप से ट्रांसक्रिप्शन किए गए स्पोनानिक भाषण की कमी के कारण है। हम मॉन्सून को प्रस्तुत करते हैं, एक पैमाने पर ऐसे डेटा को एकत्र करने के लिए एक फ्रेम। यह जुड़े योगदानकर्ताओं के बीच अनस्क्रिप्ट किए गए वार्तालापों को उत्तेजित करता है, स्वचालित वर्गीकरण के साथ स्क्रीन रिकॉर्डिंग करता है, मानव समीक्षा के पांच स्तरों पर ट्रांसक्रिप्शन को सही करता है, और lexical और acoustic कवरेज के लिए खंडों का चयन करता है। यह 50 भाषाओं में एकत्र किए गए भाषण है; हम सात 3,500 घंटे corpora

Machine translation (Meta M2M100); the English original follows.

Abstract

For most languages, automatic speech recognition (ASR) is limited less by model architecture than by a lack of accurately transcribed spontaneous speech. We present Monsoon, a framework for collecting such data at scale. It elicits unscripted conversations between paired contributors, screens recordings with automated classifiers, corrects transcripts over five levels of human review, and selects segments for lexical and acoustic coverage. It has collected speech in 50 languages; we analyse seven 3,500-hour corpora. At equal volume, its quality stages reduce semantic word error rate (WER) by 10-39% relative to a conventional protocol. Compared with an architecture-matched baseline trained on public corpora, Whisper fine-tuned on Monsoon performs comparably on the baseline's own test sets and reduces semantic WER by 20-44 points on conversational benchmarks. The gains transfer to a second architecture and grow with data volume. We release the framework, its quality-control models, and the fine-tuned models.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Speech recognition, Natural language processing, Word error rate, Artificial intelligence, Word (group theory)

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.26776, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange