Word Error Rate (WER)

The standard edit-distance metric for scoring automatic speech recognition by counting insertions, deletions, and substitutions against a reference transcript.

grounded in: Latest-trends theme 'On-device speech & audio AI': Apple SpeechAnalyzer, a native on-device speech-to-text API, benchmarked head-to-head with Whisper — WER is the canonical metric for exactly such ASR

Connected concepts

On-device speech recognition, Streaming ASR, Evals & benchmarks, Model Arena, Cost-aware agent eval