A finalized stretch of recognized speech.
Other plausible transcriptions, most likely first, excluding text.
The recognized text.
The words in text, with confidence and timing when the recognizer reported them.