Processing speech signals in voice-based profiling

#A human voice sound is produced by a physical mechanism that involves air-flow from the lungs of the individual through the structures that comprise the vocal tract of the individual. The production of the human voice sound is a complex nonlinear process that depends on the shape of the human vocal tract. The vocal tract shape is dependent on a bone structure, body size, race, age, and so forth of an individual. The relative shape of the vocal tract resonant cavities, the vocal tract tissue structure and density, and other factors such as lung capacity (controlling air-flow dynamics through the vocal tract), inertia of movement of the vocal apparatus etc. can also contribute to the sound of the human voice. The combination of these complex factors result in voice sounds that can be unique to the speaker.

#This document describes a detection system front end server that routes a speech signal to one or more processing devices that generate a response to the speech signal; and a segmentation server that includes the one or more processing devices and that processes the speech signal to segment the data into a plurality of segments, with each segment representing a portion of the speech signal, and with the segmentation server further performing operations comprising: for each segment, generating a feature vector comprising data indicative of one or more features of the portion of the speech signal represented by that segment; and determining whether the feature vector comprises data indicative of one or more features with a threshold amount of confidence; and for each of a subset of the generated feature vectors, processing data in that feature vector to generate a prediction of a value of a profile parameter, with the subset comprising one or more feature vectors determined to have one or more features with a threshold amount of confidence, wherein the segmentation server generates and transmits an output responsive to machine executable code that generates a visual representation of the prediction of the value of the profile parameter, and wherein the detection system front end server uses the output responsive to the machine executable code to remotely update a display of a client device that submitted a request to present the visual representation of the prediction of the value of the profile parameter.

#In some examples, the output comprises the value of the profile parameter and forensic profile data through one or more application interfaces. In some examples, processing data in the feature vector to generate a prediction comprises selecting a predictor algorithm, based on the data indicative of the one or more features with the threshold amount of confidence, for processing the data in each of the subset of the generated feature vectors. In some examples, the segmentation server executes a first selected prediction algorithm on detection data processed in accordance with a second selected prediction algorithm, wherein the detection data represents two or more features having a predetermined correlation or a predetermined dependency between the two or more features, the two or more features each having a threshold amount of confidence.

#FIG. 1 shows an example diagram of a data processing system.

#FIG. 2 shows an example diagram of a segmentation server.

#FIG. 3 shows an example diagram of data flow in the segmentation server.

#This document describes a system and a method for concurrently deducing various profile parameters from a speech signal. A data processing system 100 uses the various profile parameters to generate a forensic profile of the speaker and their environment. A segment of speech can be defined as a voiceprint, regardless of lexical or semantic content, or completeness. In some examples, profile parameters include a set of one or more of bio-relevant facts, such as physical, physiological, or medical facts; socio-personal facts such as behavioral, demographic, or sociological facts; and environmental facts including information about the surroundings concurrent with the speech signal production and equipment used to generate, transmit or record the speech signal. The data processing system 100 deduces the profile parameters to generate a forensic profile.

#The data processing system 100 allows for forensic profiling from disguised speech signals and speech signals that are degraded or obscured. In an example, predictive mechanisms described herein can deduce the profile parameters associated with a speech signal in relation to a delayed recording, or predictive profiling.

#The data processing system 100 can include a concurrent estimation of profile parameters associated with a speech signal. The concurrent estimation includes an analysis of the speech of the speech signal, as well as the non-speech portions of any speech signal to deduce the acoustic environment and instrumentation used to generate, transmit or record the speech signal.