System and method of text zoning

A method of improving quality of automated speech analytics, comprising:

• receiving audio data in an electronic medium;
• parsing the audio data into a plurality of overlapping temporal audio segments of predetermined segment duration, wherein a time duration of overlap of segments is a predetermined overlap length;
• decoding the overlapping temporal audio segments using a language model to produce a digital transcript file of the audio data having a plurality of audio segments;
• zoning the transcript file into meaning units, wherein the zoning comprises:
• identifying a plurality of utterances within the transcript file by applying at least one of an acoustic model and a linguistic model to the digital transcript file;
• determining that one of the plurality of utterances is a separate meaning unit if the one of the plurality of utterances has fewer than a predetermined number of words and that the one of the plurality of utterances contains more than one separate meaning unit if the one of the plurality of utterances has more than the predetermined number of words;
• parsing data associated with utterances having more than the predetermined number of words into data windows of predetermined word length;
• and
• calculating for each data window, a probability of the data window including a meaning unit boundary by applying an N-Gram linguistic model to the data window;
• and
• applying automated speech analytics to the meaning units.