Friday 31 December 2010

Computer program recognizes human emotions from conversation analysis

ScienceDaily (Nov. 8, 2010) — Researchers from the Universidad Politécnica de Madrid's Facultad de Informática have developed a computer program capable of human emotion recognition from automated voice analysis.

The application analyses the sound measurements of a conversation, output by another purpose-built program. Then, based on the rules described in the new application, it is able to identify the emotions hidden in an expression and determine whether the speaker is sad, happy or nervous. Even if the emotion is unclear, the application is able to specify how close the speaker is to each emotion in percentage terms. The application was presented by Susana Muñoz Hernández at the First International Conference on Fuzzy Computation, held in Madeira, Portugal, in 2009.

The application is based on a new tool called RFuzzy, implemented in the Prolog programming language. Prolog is able to represent and operate with what is known as fuzzy logic. Prolog is used primarily in artificial intelligence and expert systems applications.

RFuzzy is a programming support, which stands out for its ease of use and its expressivity. It is able to represent, handle and reason with subjective concepts like high, low, fast, slow, etc.

Apart from being applied to detect emotions in human conversations, RFuzzy has also been used to enhance robot intelligence. In this particular case, RFuzzy was used to program robots participating in the world robot soccer league (RoboCupSoccer), which has been held since 1996 with the aim of developing robotics and artificial intelligence. The experience was outlined in the book Robot Soccer, Inteh, Croatia, 2010, edited by Vladan Papic.

Being based on fuzzy logic, RFuzzy has important benefits for both conversation analysis and the enhancement of championship soccer robot behaviour. Its logical mechanisms are flexible and it leaves some margin of interpretation to the computer. The computer will then make the decision depending on a series of logical rules that take measurable parameters (volume, speech pitch and rate, position, speed or distance of the robot from the ball, etc.) as a reference.

The details of this research are to be published shortly in the INS- Information Science journal and can be consulted online at Science Direct.

Story Source:

The above story is reprinted (with editorial adaptations by ScienceDaily staff) from materials provided by Facultad de Informática de la Universidad Politécnica de Madrid.

Journal Reference:

  1. Susana Munoz-Hernandez, Víctor Pablos-Ceruelo, Hannes Strass. RFuzzy: Syntax, semantics and implementation details of a simple and expressive fuzzy tool over Prolog. Information Sciences, 2010; DOI: 10.1016/j.ins.2010.07.033

Note: If no author is given, the source is cited instead.

Disclaimer: Views expressed in this article do not necessarily reflect those of ScienceDaily or its staff.