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
Für die meisten Sprachen ist die automatische Spracherkennung (ASR) weniger durch Modellarchitektur begrenzt als durch fehlende präzise transkriptierte spontane Sprache. Wir präsentieren Monsoon, ein Rahmenwerk für die Erhebung solcher Daten in der Größe. Es erzeugt unskriptierte Gespräche zwischen gepaarten Beiträgen, Bildschirmaufnahmen mit automatisierten Klassifikatoren, korrigiert Transkripte über fünf Ebenen menschlicher Überprüfung und wählt Segmente für lexische und akustische Abdeckung. Es hat Sprache in 50 Sprachen gesammelt; wir analysieren sieben 3,500-Stunden-Corpora. Bei gleichem Volumen reduzieren seine Qualitätsstufen die semantische Wortfehlerrate (WER) um 10-39% im Vergleich zu einem konventionellen
Machine translation (Meta M2M100); the English original follows.
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)
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.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange