METHOD AND SYSTEM FOR MEASURING AND QUANTIFYING USER'S STRESS LEVELS THROUGH VOICE SIGNAL ANALYSIS

FIELD OF INVENTION

The present invention relates to a method and system for measuring and quantifying user's stress levels. More specifically, the present invention relates to method and system for quantifying user's stress by assigning scores/weight-ages/numeric values for individual attributes through analysis of user's voice attributes, and then using proprietary algorithm to come up with a total stress score/weight-age/numeric value which is an algebraic combination of individual attribute-level scores, said scores/weight-ages/numeric values assigned to each voice parameter are based on correlation studies with Cortisol levels in the body. Significant correlation between the two at the initial stage helped validate the methodology. Advantageously, the method provides non-biased, consistent scoring of stress and even an inexpensive way for encouraging preventive healthcare system.

DESCRIPTION OF PRIOR ART

Lie detection devices, i.e. devices which measure psychological stress as an indicator of deception, are commonly used in the fields of law enforcement and the military and occasionally in the private sector. The oldest type of lie detection device is known as the polygraph which measures changes in a person's body associated with the stress of deception, including alterations in heart rate, breathing, and electrodermal activity. In a polygraph examination, the subject is required to be "wired" to the examination equipment in order to record the various physiological changes. Later advances in the field of lie detection technology have focused on the detection of stress in human speech in an attempt to produce a portable lie detection device which could be used in the field. Voice stress detection devices provide an advantage over the traditional polygraph in that they do not require that the subject be "wired," and are thus a non-invasive means of truth detection. Thus, the roots of voice analysis are found in the history of lie detection. Recent research such as Detecting Stress in Unconstrained Acoustic Environments using Smartphones, by Hong Lu (of Intel Labs) et al, Speech analysis under stress by Patil Rajesh Namdeo and Amit Kumar Mishra of MIT, Aurangabad, India in Apr 2012, Voice Stress Analysis of L.J.M. Rothkrantz, P. Wiggers, J.W.A. van Wees, R.J. van Vark; Data and Knowledge Systems Group of Delft University of Technology, Voice monitoring to measure emotional load during short-term stress by Peter Wittels, Bernd Johannes, Robert Enne, Karl Kirsch, Hanns-Christian Gunga, and Voice monitoring to measure emotional load during short-term stress; Peter Wittels, Bernd Johannes, Robert Enne, Karl Kirsch, Hanns-Christian Gunga has also shown that stress can be rather accurately evaluated from voice analysis.

But, very little effort was made to establish exact patterns or to provide a formula for determining stress level, which leads to considerable confusion to the evaluation process and creating many errors. Further, in prior art for voice stress analysis systems, the determination of stress is made empirically by a human examiner. A problem with the prior art systems is that since the judgment is made purely by visual observation, the personal bias of the examiner may enter into the scoring of the voice pattern. Also, since the results are based on individual human judgment, they can be influenced by, for example, the quality of training and the cumulative field experience of the individual examiner. This will naturally create inconsistencies in the scoring between individual examiners. US7321855 relates to a computer-implemented method of assigning a numeric score to a voice pattern sample of a human subject, wherein the score is indicative of the psychological stress level of the human subject. A verbal utterance of a human subject is converted into electrical signals to provide a subject wave pattern. The pattern is quantified and compared with known voice pattern characteristics which exhibit a sequential progression in the degree of blocking in the pattern, wherein each of the known voice patterns is assigned a numerical value range.

WO2003081578 relates to a method for detecting, measuring, or monitoring the presence or absence of at least one emotion in a subject from a speech sample obtained from the subject which comprises extracting at least one feature from the speech sample, assigning the speech sample a score using a weighted frequency band scoring scheme, and comparing the score with a general reference model or a control. Therefore, there is a need for solving the drawbacks of the prior art voice analysis systems by a computer software-implemented scoring system which utilizes an algebraic algorithm to analyze all the voice attributes and generate a score. Moreover, the computer software-generated score is an inexpensive method to provide non-biased, consistent scoring of user's stress. Further, there is a need to make such an offering available over any mobile phone be they simple feature phones or smartphones, especially when a large mass of users in the world have no smart phones.

OBJECTS OF INVENTION

One or more of the problems of the conventional prior art may be overcome by various embodiments of the present invention. It is the primary object of the present invention to provide method and system for measuring and quantifying user's stress levels. It is another object of the present invention to provide method and system for measuring and quantifying stress levels in human body through analysis of voice attributes using computer -implemented software with an algorithm. It is another object of the present invention to provide method and system for quantifying user's stress by assigning stress scores/weight-ages/numeric values for individual attributes through analysis of user's voice attributes, and then using the algorithm to come up with a total stress/ weight-age/numeric value score which is an algebraic combination of individual attribute-level scores. It is another object of the present invention, wherein the analysis and evaluation of voice attributes includes but not limited to pitch variance, energy variance, spectral slope, fO mean, jitter and loudness. It is another object of the present invention, wherein the total stress score/weight-age/numeric value is sent to a user device. It is another object of the present invention, wherein the system assigns and sends an avatar representation explaining the level of total score/weight-ages/numeric values attained by the user.

SUMMARY OF INVENTION

Thus according to the basic aspect of the present invention there is provided a method for measuring and quantifying user's stress levels through voice signal analysis comprising: Collecting user's voice sample through a phone call or one or more voice receivers; Breaking the user's voice sample in myriad of parameters of voice attributes; Assigning scores/weight-ages/numeric values to each voice attribute for the collected user's voice sample by computer - implemented software; Calculating the assigned stress scores/weight-ages/numeric values using an algebraic equation involving all the voice attributes; Combining all individual voice attributes scores to total stress score/weight-age/numeric value; Communicating the total stress score/weight-age/numeric value to a user device; and Assigning and sending an avatar representation explaining the level of total score/weight-age/numeric value attained by the user, wherein the scores/weight-ages/numeric values assigned to each voice attribute are based on correlation studies with Cortisol. It is another aspect of the present invention, wherein validation is done for the initial set of users by correlating with Cortisol hormone released from the users during stress. It is another aspect of the present invention, wherein the voice attributes includes but not limited to pitch variance, energy variance, spectral slope, fO mean, jitter and loudness.

It is another aspect of the present invention, wherein the avatar representation is classified into zones in accordance to the level of total score/weight-age/numeric value attained by the user. It is another aspect of the present invention, wherein the zones include but not limited to calm, good stress, medium stress and high stress. Another aspect of the present invention is directed to provide a system for measuring and quantifying user's stress levels through voice signal analysis comprising: One or more voice receivers; Central processor; Memory in communication with the central processor; and User device, wherein the voice receiver is a microphone, which collects user's voice sample, wherein the user's voice sample is broken in myriad of parameters of voice attributes, wherein the central processor that includes software with an algorithm assigns scores/weight-ages/numeric values to each voice attribute for the collected user's voice sample, wherein the scores/weight-ages/numeric values assigned to each voice attribute are based on correlation studies with Cortisol, wherein the assigned scores/weight-ages/numeric values are calculated using an algebraic equation involving all the voice attributes and combined to give total stress score/weight-ages/numeric values, and wherein the total stress score/weight-ages/numeric values is sent to the user device by the memory in communication with the central processor. It is another aspect of the present invention, wherein the user device is a mobile phone. It is another aspect of the present invention, wherein validation is done for the initial set of users by correlating with Cortisol hormone released from the users during stress. It is another aspect of the present invention, wherein the voice attributes includes but not limited to pitch variance, energy variance, spectral slope, fO mean, jitter and loudness.

DETAILED DESCRIPTION OF THE INVENTION

The present invention as discussed hereinbefore relates to a method and system for measuring and quantifying user's stress levels. More specifically, the present invention relates to method and system for quantifying user's stress by assigning stress scores for individual attributes through analysis of user's voice attributes. According to one aspect of the present invention, a method for measuring and quantifying user's stress levels through voice signal analysis comprises collecting user's voice sample through a phone call or one or more voice receivers; breaking the user's voice sample in myriad of parameters of voice attributes; assigning scores/weight-ages/numeric values to each voice attributes for the collected user's voice sample by computer - implemented software, said scores/weight-ages/numeric values assigned to each voice attribute are based on correlation studies with Cortisol; calculating the assigned stress scores/weight-ages/numeric values using an algebraic equation involving all the voice attributes; combining all individual voice attributes scores/weight-ages/numeric values to total stress score/weight-age/numeric value; communicating the total stress score/weight-age/numeric value to a user device; and assigning and sending an avatar representation explaining the level of total score/weight-age/numeric value attained by the user. Significant correlation between the two at the initial stage helped validate the methodology. Advantageously, the method provides non-biased, consistent scoring of stress/weight-ages/numeric values and even an inexpensive way for encouraging preventive healthcare system.

According to another aspect of the present invention, the computer software-implemented system for measuring and quantifying user's stress levels through voice signal analysis comprises one or more voice receivers like microphone; central processor; memory in communication with the central processor; and user device such as mobile phones. The system is embedded with measuring and quantifying stress level software with an algorithm. The voice receiver collects user's voice sample. The user's voice sample is broken in myriad of parameters of voice attributes. The central processor that includes computer - implemented software with algorithm assigns scores/weight-ages/ numeric value to each voice attributes for the collected user's voice sample, said scores/weight-ages/ numeric value assigned to each voice attribute are based on correlation studies with Cortisol levels in the body. The system assigned scores/weight-ages/ numeric values are calculated using the proprietary algorithm to come up with a total stress score/weight-age/numeric value which is an algebraic combination of individual attribute - level scores. The voice attributes includes but not limited to pitch variance, energy variance, spectral slope, fO mean, jitter and loudness and are combined to give total stress score/weight-age/ numeric value.

The total stress score/weight-age/numeric value is sent to the user device by the memory in communication with the central processor. The system assigns and sends an avatar representation explaining the level of total score/weight-ages/numeric value attained by the user. For illustration, It is well-known that the hormone Cortisol is a good bio-marker for stress, in the body. Validation by correlating with Cortisol hormone released from the human body during stress is done for an initial sample of users. The stress scores/weight-ages/numeric values are obtained from an initial set of users by collecting voice samples, said scores/weight-ages/numeric values assigned to each voice attribute based on correlation studies with Cortisol. Significant correlation between the two at the initial stage helped validate the methodology. To know stress level, the user calls the respective service number through his/her mobile phone. The user receives multiple computer generated questions to his/her mobile phone. The user answers reply through his/her voice for each question. For every user/caller, the voice sample is processed by computer - implemented software program and a score/weight-ages/numeric values based on algorithm is calculated and a total stress score/weight-age/numeric value is sent to the user/caller device either by SMS or through a mobile application running in the smartphones.

The system assigns and sends an avatar representation explaining the level of total score/weight-age/numeric value attained by the user. The avatar representation is classified into zones in accordance to the level of total score/weight-age/numeric value attained by the user. The zones include but not limited to calm, good stress, medium stress and high stress. The user/caller receives the total score/weight-age/numeric value instantly, after the user's answer to the queries. It is to be understood that the foregoing objects are exemplary and explanatory only and are intended to provide further explanation of the invention as claimed.

WE CLAIM:

1. A method for measuring and quantifying user's stress levels through voice signal analysis comprising: Collecting user's voice sample through a phone call or one or more voice receivers; Breaking the user's voice sample in myriad of parameters of voice attributes; Assigning scores/weight-ages/numeric values to each voice attribute for the collected user's voice sample by computer - implemented software; Calculating the assigned stress scores/weight-ages/numeric values using an algebraic equation involving all the voice attributes; Combining all individual voice attributes scores to total stress score/weight-age/numeric value; Communicating the total stress score/weight-age/numeric value to a user device; and Assigning and sending an avatar representation explaining the level of total score/weight-age/numeric value attained by the user, wherein the scores/weight-ages/numeric values assigned to each voice attribute are based on correlation studies with Cortisol.

2. The method as claimed in claim 1, wherein validation is done for the initial set of users by correlating with Cortisol hormone released from the users during stress.

3. The method as claimed in claim 1, wherein the voice attributes includes but not limited to pitch variance, energy variance, spectral slope, fO mean, jitter and loudness.

4. The method as claimed in claim 1, wherein the avatar representation is classified into zones in accordance to the level of total score/weight-age/numeric value attained by the user.

5. The method as claimed in claim 4, wherein the zones includes but not limited to calm, good stress, medium stress and high stress.

6. A system for measuring and quantifying user's stress levels through voice signal analysis comprising: One or more voice receivers; Central processor; Memory in communication with the central processor; and User device, wherein the voice receiver is a microphone, which collects user's voice sample, wherein the user's voice sample is broken in myriad of parameters of voice attributes, wherein the central processor that includes software with an algorithm assigns scores/weight-ages/numeric values to each voice attribute for the collected user's voice sample, wherein the scores/weight-ages/numeric values assigned to each voice attribute are based on correlation studies with Cortisol, wherein the assigned scores/weight-ages/numeric values are calculated using an algebraic equation involving all the voice attributes and combined to give total stress score/weight- age/numeric value, and wherein the total stress score/weight-age/numeric value is sent to the user device by the memory in communication with the central processor.

7. The system as claimed in claim 6, wherein the user device is a mobile phone.

8. The system as claimed in claim 6, wherein validation is done for the initial set of users by correlating with Cortisol hormone released from the users during stress.

9. The system as claimed in claim 6, wherein the voice attributes includes but not limited to pitch variance, energy variance, spectral slope, fO mean, jitter and loudness.