You might initially assume that the units would be individual words. But this is not the case, since in almost all languages words flow into each other smoothly during speech. Nor are letters appropriate units, as they are pronounced differently depending on the word and the level of emphasis. Other difficulties involved in the understanding of speech include variations in pitch, depending on the mood and voice of the speaker, accents, dialects, and different speeds. All of these factors have a significant impact on the sound of the spoken word, resulting in very different patterns from person to person, even though the spoken sentence is identical.
The brain apparently copes effortlessly with all this when interpreting language, and works at a tremendous speed. In fact, at a normal rate of speech, we perceive up to 14 speech signals per second. Another fascinating fact is that, if the rate of speech increases to 60 signals per second, the content becomes even easier for us to understand.
Even modern computer technology cannot compete. To date, there is no voice recognition program that recognizes spoken language as rapidly as the human brain.