Real-Time STT WER Comparison: Three Major Models Evaluated
ArtificialAnlys · x · 2026-07-06
In partial transcription of the first sentence, Universal-3.5 Pro Realtime's Max Accuracy mode achieves 4.1% WER 0.44 seconds after speech ends. Its accuracy is second only to ElevenLabs Scribe v2 Realtime (3.6%, 0.13s) and slightly better than Cartesia Ink-2 outer endpoint (4.3%, 0.07s), though with slower text output. The Min Latency mode returns the first sentence in 0.39s with 6.1% WER, trading accuracy for lower latency.
Related event: AssemblyAI Launches Universal-3.5 Speech Models(4 posts)→
More from Multimodal
- AI filmmaker PJaccetturo discusses building his NEXUS universe with Dreamina — DavidmComfort · 2026-10-27
- Runway launches App Studio: build no-code creative apps that generate image, video and audio — runwayml · 2026-10-27
- MiniMax H3 runs Maestro locally for free; Kling 4.0 vs Seedance 2.5 video editing compared — cocktailpeanut · 2026-10-27
- Claude scripts, generates and edits an emotional AI commercial end-to-end in Higgsfield — aakashgupta · 2026-10-27
- AI-generated movie-grade clip goes viral, blurring the line with real cinema — cneuralnetwork · 2026-10-27
- 10 practical ways to run Astra computer use on Premiere Pro for video editing — beechinour · 2026-10-27