Zoom Scribe V2.1 Pro Live tops streaming transcription benchmark
The result shows streaming speech-to-text accuracy has effectively converged at the top, so buyers should now weigh latency, language coverage, and per-minute pricing instead of leaderboard position. Teams evaluating live captioning for meetings or webinars can use AA-WER Streaming as an independent reference when comparing Zoom against Microsoft's offerings. Zoom's integration advantage means the model is most directly useful to existing Zoom customers.
Zoom's Scribe V2.1 Pro Live has taken the number one position for Final Transcript accuracy on Artificial Analysis's AA-WER Streaming benchmark, achieving a 2.4% word error rate. Microsoft's MAI-Transcribe-2-Streaming follows closely in second place at 2.5%, making the gap between the two leading streaming speech-to-text models just 0.1 percentage points.
Artificial Analysis conducted its own independent benchmarking of Zoom's newly launched streaming speech-to-text model rather than relying on vendor-reported figures. The AA-WER Streaming evaluation measures how accurately a model transcribes live audio streams, where transcripts are produced in real time rather than after the full audio is available — a harder task that typically produces higher error rates than batch transcription.
Streaming transcription accuracy matters for live captioning, real-time meeting notes, and accessibility features embedded in video conferencing platforms. With Zoom edging out Microsoft's MAI transcription model, competition in streaming speech-to-text is now effectively down to tenths of a percentage point, pushing vendors to differentiate on latency, pricing, and integration rather than raw accuracy alone.
Zoom Scribe V2.1 Pro Live takes the #1 spot for Final Transcript accuracy on AA-WER Streaming, with a 2.4% word error rate, just ahead of MAI-Transcribe-2-Streaming at 2.5%
We benchmarked @Zoom 's new streaming Speech to Text model independently. Congratulations to the Zoom team on the launch! 🧵
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Zoom, speech transcription, AA-WER Streaming, MAI-Transcribe.
Source attribution: curated from Artificial Analysis (@ArtificialAnlys).