Abstract
Long-time spectra of signals of different radio broadcast programmes have been measured using the method of sample superposition (superposing signals of statistically independent sections of programme signals, recorded on tape, interpreted by ensembles with instruments of similar character). Spectra of signals of different ensemble groups (set of ensembles of equal composition of interpreters), having equal probabilities of occurrences, are calculated. Using programme policy statistics of three different transmitted programmes, the probabilities of occurrence of different ensemble classes (sets of ensembles, composed according to a rule) are calculated. Using these probabilities, the long-time-weighted spectra of the signals of three different radio programmes and the averages of these spectra are computed. The values are only slightly dependent on programme policy. A definition of the spectrally equivalent programme signal and a network for forming it are given.References
[1] I. B. CRANDALL, D. McKENZIE, Analysis of the energy distribution in speech, Phys. Rev. Ser., 3, 2, 221-232 (1922).
[2] L. J. SILIVAN, Speech power and its measurements, Bell Syst. Techn. Journ., 8, 646-661 (1929).
[3] H. K. DUNN, S. D. WHITE, Statistical measurements on conversational speech, JASA, 11, 278-288 (1940).
[4] N. R. FRENCH, J. C. STEINBERG, Factors govering the intelligibility of speech sound, JASA, 13, 90-113 (1947).
[2] L. J. SILIVAN, Speech power and its measurements, Bell Syst. Techn. Journ., 8, 646-661 (1929).
[3] H. K. DUNN, S. D. WHITE, Statistical measurements on conversational speech, JASA, 11, 278-288 (1940).
[4] N. R. FRENCH, J. C. STEINBERG, Factors govering the intelligibility of speech sound, JASA, 13, 90-113 (1947).