The benchmark is using the latest stable of DS >>> https://github.com/Picovoice/speech-to-text-benchmark/blob/m...

the data for cloud-based is from 2022.