Data/ML/AI
Speech Recognition
Also written as ASR, Automatic Speech Recognition, Speech-to-Text, Whisper
Turning recorded speech into text. Accuracy swings enormously with accent, background noise and specialist vocabulary, so a headline accuracy figure means little on its own.
Think of it like
Transcribing a meeting. Easy in a quiet room, much harder in a café with three people talking over each other.
Junior or senior?
The signal is whether they measured on their own audio.
Senior sounds like
Talks about word error rate on real recordings and about domain vocabulary.
Ask them
“How well did it work on your actual audio, compared with the benchmark number?”