What is automatic speech recognition (ASR)?
Automatic speech recognition converts spoken audio into written text. It is the step that lets software work with what a caller says.
Why it matters
Everything a voice system does with a caller’s words depends on the text it receives. If a name, a number or a single word such as “not” is misheard, later steps act on the wrong request. Accurate recognition across accents, background noise and narrow phone-line audio is what allows callers to speak naturally instead of pressing keys or repeating themselves.
How it works
The audio is split into very short slices and passed to an acoustic model that estimates which speech sounds are present. A language model then chooses the most likely sequence of words for those sounds. In a live call this happens as a stream, so partial text appears while the caller is still talking. Many systems accept a list of expected terms, such as product names, to improve accuracy.
In practice
A caller says, “I’d like to move my delivery to the fourteenth.” The recogniser returns that sentence as text within a moment, along with a confidence score. It does not decide what the caller wants. That is the job of the next step, which reads the text and identifies a request to reschedule a delivery and the new date.
