Use of Neural Networks in the Formation of a High-Quality Smoothed Audio Signal


Relevance. High-quality smoothing of sound during its passage in local networks and in stereo systems allows for the transmission of sound over wireless networks with virtually no delays. The basis for this sound remains controversial. Its decoding is possible both on the receiving device and on the transmitter. At the same time, the processors can provide decoding in real time with automatic adjustment by algorithms. Neural algorithms can be used both on the basis of signal sequence and on the parallel use of receivers.

Purpose. The purpose of this study the learning a diagnostic method that combines the analysis of several indicators, which will significantly increase the probability of detecting a malfunction in sound transmission or individual nodes.

Methods. In the process of the study the used of exact algorithms and non-exact algorithms. Exact algorithms are divided into linear programming techniques and dynamic programming techniques. Linear programming methods include: Brute Force method, Branch and Bound method, Gomori method and other.

Results. The authors show that in this regard, the problem of high-quality sound transmission using neural algorithms is reduced to the problem of an optimal transportation project and is solved by optimising the local sound route. The equilibrium distribution of transport routes should be built into the equipment itself and the audio decoding protocol.

Conclusions. The use of a sound receiver with neural algorithms makes it possible to identify the most difficult areas that cause sound delay or distortion. Practical application of the study is possible in conditions of dynamic provision of sound filling of the room to create the effect of presence and volumetric absorption of sound, if necessary

Keywords: algorithm, function, optimised route, fuzzification, sound transport