How Accurate?

The Per Word Confidence score for each transcribed word enables VoiceBase customers to analyze their content more granularly. This can power valuable predictive algorithms, which supply businesses with actionable insights.

How Accurate?

The Per Word Confidence score for each transcribed word enables VoiceBase customers to analyze their content more granularly. This can power valuable predictive algorithms, which supply businesses with actionable insights.


See Results

Per word confidence is a measure of the acoustic similarity between the sound in the audio recording and the word that was transcribed. Confidence scores can be returned as JSON data and denoted ‘c’ in the words object. The score can be any number between 0 and 1. A score closer to 1 indicates a strong similarity, and a word closer to 0 indicates poor acoustic similarity (a word with low confidence may still be transcribed correctly due to strong context). A score of 1 indicates that the word was submitted through a human process, either user submitted or ordered through the human transcription service.