Post by Sierra

Only about a quarter of the world speaks English — but most ASR benchmarks are built around it. Today we’re introducing μ-Bench, an open multilingual transcription benchmark built from real customer service phone conversations across multiple locales. It evaluates how systems perform on the errors that matter in production — not just how closely transcripts match a reference, but whether they preserve the speaker’s intent. Learn more: https://lnkd.in/gZwZCWPJ