Hugging Face and Voice Arena have partnered to add the Open ASR Leaderboard's first Global South language, introducing evaluation sets for Hindi and Indian English. The new Monsoon datasets aim to address a critical gap in automatic speech recognition benchmarking: research has shown that ASR error rates vary significantly across demographic groups, with commercial systems performing roughly twice as poorly for Black speakers compared to white speakers. By making speaker demographics visible in the benchmark, the initiative challenges the prevailing approach of optimizing for single aggregate metrics that can mask disparities affecting specific populations. The Monsoon datasets comprise 4,888 speakers and deliberately vary across nine dimensions—geography, age, gender, vocabulary, devices, acoustic environments, speech type, speech rate, and transcript variants—to expose failure modes that aggregate performance metrics typically hide. Hindi, spoken by over half a billion people, becomes the first Indic language on the multilingual evaluation tab, which previously covered only European languages. Both public and private dataset splits are released to enable community evaluation while preventing benchmark-specific model optimization. The benchmark reflects growing recognition that ASR systems require demographic-stratified testing to identify and remediate bias. By prioritizing speaker diversity over raw audio duration and systematically recording speaker attributes, the benchmark provides researchers and developers with tools to measure performance disparities and drive more equitable AI development in speech recognition.