Analyzing the Prosody of Speech

McLean Hospital
posted on 01/20/2011

• request more info 
•   
•  LICENSE ONLINE

Psychiatric disorders such as autism, depression, schizophrenia, bipolar disorder, schizoaffective disorder, and anxiety disorders can be difficult to diagnose or to distinguish from each other. In many cases, individuals suffering from these disorders can exhibit altered patterns of speech. Human speech comprises content and prosody. Content of speech includes the information given by the words and sentences, while prosody adds valuable information to the content of speech by varying the rhythm, stress, and intonation of speech. Speech prosody can be analyzed from recorded or live speech, to analyze the speech data in view of features of the underlying prosody. The speech data can be characterized by various parameters that include, for example, the fundamental frequency of the acoustic signal (pitch), the amplitude of the signal (intensity/loudness), and the temporal behavior of the signal, i.e., the time periods during which the acoustic signal is present (utterance or syllable) and the time periods without any acoustic signal (pause). The invention is based on the discovery that a representation of speech data from different sources can be distinguished by separating the speech data into utterances and pauses according to one or more specific rules.
Potential Commercial Uses: The invention features methods of diagnosing psychiatric disorders by analyzing the prosody of speech in a patient. Using specific rules developed by the inventors, a patient’s speech can be analyzed in a recording or other reproduction of live speech, in order to detect specific patterns that are characteristic of disorders such as schizophrenia or bipolar disorder.
Patent Status: McLean Hospital and Massachusetts Institute of Technology have filed a pending U.S. patent application claiming this invention.
Licenses Available: McLean Hospital is offering a worldwide exclusive license to this technology.