The book presents new effective digital algorithms for processing audio and tonal signals characterized by high natural redundancy. The redundancy is fully preserved in headers of compressed audio frames in MPEG-1 Layer 2 format. Extensive computer modeling proved that it is possible to detect and correct all catastrophic errors in three fields of a frame header. It was suggested that a Hamming error detection code with parity bits be used on the encoder end and transmitted over the existing low-speed channel to the decoder. By correcting errors in frame headers one can significantly reduce requirements to the signal-to-noise ratio of satellite digital channels used for radio or TV audio broadcasting.
Strong correlation between digital samples of an audio signal presented in the non-linear PCM format makes detectable errors in the most significant bits of sample codes perceived as ‘crackling’. To make sure that no double threshold excess occurs (i.e. no two disturbances of opposite signs exceed the threshold values set for the first signal derivative), the distorted sample is replaced with the previous one.
The notion of interval "spectrum" is introduced for intervals between successive points where digital samples change sign. Knowing the sign bit of the PCM code, one can roughly estimate the width of the signal’s frequency spectrum and distinguish between music, speech and signaling tones.
The book demonstrates feasibility of quick adaptation of sampling rate to the content of broadcast programs.
An adaptive speech compression algorithm based on time-domain compression technique is analyzed. By using this algorithm, one can ensure recognition of isolated words with speech compression rate as high as 30 while maintaining acceptable quality of sound.
The proposed error correction algorithms are described in detail in the Appendix. As a bonus, the software package implementing those algorithms along with 12 compressed audio files that represent different types of content can be downloaded for free from
https://www.dropbox.com/sh/s6ll5nxyga6vrnu/AACsuq8ehSK67adZZoctDD_-a?dl=0.
The software allows simulating introducing, detection and correction of three types of catastrophic errors in the headers of compressed sound frames in MPEG-1 Layer-2 format.
This book is addressed to researchers, practicing engineers and graduate students interested in adaptive digital processing of telecommunication signals with natural redundancy.
Irina S. Brainina, Doctor of Technical Sciences, Professor of the Department of Theoretical Radio Engineering, Povolzhsky State University of Telecommunications and Informatics, Samara city, former city Kuibyshev. After graduating the Electrotechnical Institute of Communications she entered the postgraduate course of the Leningrad Electrotechnical Institute of Communications. In 1969 she defended her thesis and obtained the degree of Candidate of Technical Sciences. In 2006 she defended her doctoral dissertation, was awarded the scientific degree of Doctor of Technical Sciences. She is the author of three monographs, one of which was published in Russia and two in the United States. She published more than 30 scientific articles and received more than 50 patents for inventions. Areas of academic interest are: applied analysis of emissions of random processes, of interference-resistant digital signal processing using their natural redundancy.
