Speech ClippingJuly 11th Fact-of-the-Day
Consonants contain the majority of the intelligence information in speech (the portion that makes it possible to understand what someone is saying), but vowels contain the majority of the power. In fact, vowels typically have about four times the peak amplitudes of consonants. The primary value of vowels is in making speech sound natural so speakers can be identified. If speech is processed to further reduce the amplitudes of consonants, relative to vowels, a speaker's voice remains easily recognizable, but it becomes very difficult to understand what is being said. However, if speech is electronically processed to greatly reduce the amplitudes of vowels, relative to consonants, there is little loss of intelligibility, but a big loss in perceived quality and it becomes very difficult to identify speakers. Because vowels have about four times the amplitudes of constants, speech clipping is a simple method of reducing the amplitudes of vowels relative to consonants. 6 dB of speech amplitude can be clipped off with only a very slight reduction in intelligibility. If the clipped signal is used to modulate an AM transmitter the average sideband power is increased approximately 6 dB (four times). (This method is not effective with SSB, because audio peaks and corresponding SSB envelope peak shapes are different.) ©2005 Martek International All rights reserved.
Search other ham radio sites with Ham Radio Search