Glossary

ASR confidence score

The probability the speech-to-text layer assigns to its own transcription. A useful signal when the agent should pause and confirm, ignored entirely by most vendors.

Every ASR system, when transcribing audio, produces not only a transcript but a confidence score per word or per utterance. A confident transcription ("appointment at three") scores high; an uncertain one ("ehmpoinmen at three?") scores low. The score is metadata most vendors throw away.

Using the score well is high-leverage. A turn with low ASR confidence on a critical entity (the date, the dollar amount, the spelling of a name) should trigger a confirmation rather than a silent action. 'I heard Thursday the 26th at three, did I get that right?' costs one extra turn and prevents a wrong-day booking, the kind of error that destroys customer trust.

A vendor that ignores ASR confidence is treating every transcription as if it were certain. That is fine on a quiet line with a clear-speaking caller. It is a liability on a noisy line, an accented call, or any conversation where the stakes of a misheard word are real.

How Vorel does this

Vorel confirms entities when ASR confidence drops below a per-vertical threshold. The clinic profile is more cautious about names and dates; the auto-shop profile is more cautious about VINs.

The next call doesn’t have to go to voicemail.

Book a thirty-minute demo. We point Vorel at one of your real numbers on the same call.