GoKawiilWispr Flow launches Canto speech model for real-world dictation
Wispr Advanced Interfaces Lab has released Canto, a new speech recognition model built specifically for real-time dictation in noisy, everyday conditions rather than clean studio audio. In tests using 10 hours of real Wispr Flow dictations from over 2,300 speakers, Canto posted the lowest word error rate compared with models from Google, OpenAI, AssemblyAI, and Deepgram.
GoKawiil's interpretation of the reporting above, not reported fact.
Most speech models are benchmarked on pristine audio, but actual dictation happens through laptop mics and earbuds amid background noise, meetings, and commutes—conditions where accuracy typically degrades. By optimizing specifically for these messy real-world scenarios and outperforming established competitors, Wispr signals a shift toward speech models tailored to actual usage patterns rather than lab conditions, which could meaningfully improve voice-based productivity tools.
- Canto is Wispr's new speech model designed for real-time, real-world dictation conditions.
- It outperformed models from Google, OpenAI, AssemblyAI, and Deepgram on word error rate in a real-user test set.
- The evaluation used opted-in data from 2,300+ speakers with strict train/test separation to avoid overfitting.
Jabra Evolve2 65 Headset — If you're dictating on the go through busy offices, commutes, or coffee shops like the scenarios Canto is built for, a headset with a noise-isolating boom mic can dramatically improve transcription accuracy. The Jabra Evolve2 65 is designed for exactly this kind of real-world use, with background noise suppression and a clear mic pickup that pairs well with speech-to-text tools like Wispr Flow. It's a practical companion for anyone relying on voice dictation throughout their workday.
See Jabra Evolve2 65 Headset on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.Source: wisprflow.ai, 2026-09-17
Published there as: “Canto: A speech model built for the real world”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.