En este artículo se propone un esquema de modulación digital denominado Modulación Particionada de 4 Estados en 3 Dimensiones (4P-3D). Dicho esquema divide el mensaje binario en bloques de 2 bits y representa cada bloque mediante un vector real de 3 componentes. Los puntos de la constelación coinciden con los vértices de un tetraedro regular centrado en el origen. El conjunto base de la modulación está formado por tres funciones de una familia wavelet ortonormal. Los análisis teóricos y los resultados de las simulaciones confirman que la modulación 4P-3D tiene un mejor desempeño frente al ruido que la modulación por amplitud de pulso binaria (2-PAM) sobre canales con ruido blanco aditivo Gaussiano (AWGN). Además, la conformación de la señal modulada a partir de wavelets hace que los errores de sincronismo en la detección de las muestras tengan un efecto prácticamente despreciable en la probabilidad de error de bit. Un aspecto desfavorable del esquema 4P-3D es la reducción de la eficiencia espectral respecto a la modulación 2-PAM.

Introduction: This review article was prepared as part of a graduate thesis at Universidad del Cauca in 2017. It sought to find the most appropriate methods for detecting voice fundamental frequency to be implemented in real environments. This is part of a solution to improve the communication of people with hearing disabilities and include them in society, since most of the proposals only aim to improve the communication channel in which the hearing-impaired individual is the transmitter. Methodology: An updated review of the literature was carried out, based mainly on scientific articles published in the last five years.  For the inclusion of articles, a systematic mapping was performed on the different methods for detecting voice fundamental frequency. Results: the phenomena considered by the various algorithms to define the environment range from noise and interference to reverberation; the performance of the algorithm depends on the quality of the recorded audio, which is observed in the variations obtained which depend on the database used; up to two different fundamental frequencies can be detected. Conclusions: Novel methods have been implemented to make the detection of voice fundamental frequency more efficient; however, there is still much work to be done in this area.