Artificial Analysis Launches Diarization Index; NVIDIA Nemotron 3.5 Transcribe Leads at 3.3%

Artificial Analysis announced the Diarization Index, a benchmark scoring speech-to-text APIs on Word Diarization Error Rate across 97 conversations of two to eight people. NVIDIA's Nemotron 3.5 Transcribe leads at 3.3%, ahead of Voxtral Small Transcribe 3 at 7.5%.
Event version 1.
Key insights
New benchmark measures speaker attribution
The Diarization Index scores APIs on Word Diarization Error Rate, the share of words assigned to the wrong speaker or to no speaker. It covers 97 conversations of two to eight people, with equal weight on each group size.
Nemotron 3.5 Transcribe leads
Nemotron 3.5 Transcribe from NVIDIA leads the index at 3.3% WDER. Voxtral Small Transcribe 3 ranks second at 7.5%.