Microsoft Launches Paza Speech Recognition for Low-Resource Languages

Microsoft Research has launched Paza, an initiative to advance automatic speech recognition (ASR) for low-resource languages, starting with several in Africa.

Paza combines a transparent benchmarking platform with speech models for real-world use. At the heart of the initiative is PazaBench, an evaluation platform dedicated to low-resource languages that helps developers identify performance gaps and accelerate improvements in speech technology for underserved communities.

Alongside the Paza benchmark, Microsoft Research is fine-tuning its ASR models for Swahili and several Kenyan languages.

"Paza is about co-creating speech technologies in partnership with the communities who use them. Guided by this principle, Paza puts human use first, which enables model improvement," Microsoft engineers wrote in a blog post about the innovation.

"We plan to expand PazaBench beyond African languages and evaluate state-of-the-art ASR models across more low-resource languages globally. The Paza ASR models are an early step; truly supporting small and under-represented languages requires dedicated datasets, strong local partnerships, and rigorous evaluation. Meaningful progress depends on sustained collaboration with the communities who speak these languages, and expanding responsibly means prioritizing depth and quality over broad but shallow coverage."