Identificación de Emociones en la Voz.

Currently based applications processing of spoken language are very common. In this line of research we can be found speech and speaker recognition, voice synthesis among others. The work presented here is on as a machine can identify the emotions from a human speaker. The model used consists of three blocks, the first block is responsible for the signal pre-processing, the tasks are carried out are filtering, and optional pre-emphasis an algorithm for detecting voice. The second module is responsible for extracting features of the speech signal, in this work will be done using a spectral analyzer as is the Warping using specific the MFCC (Mel Frequency Cepstral Coefficients). The third one module is responsible for qualifying which uses the algorithm K-Means. The evaluation is done through an array of so-called confusion which are measured successes of the system, for trainning and test are used two databases one in Spanish (SES) and the other in German (EmoDB). [ABSTRACT FROM AUTHOR]