METHOD FOR ENLARGING THE BANDWIDTH OF A NARROW-BAND FILTERED SPEECH SIGNAL

摘要

The narrow-band filtered speech signal is evaluated in relation to the frequency components above a threshold frequency such that a spectral structure is calculated from the narrow-band speech signal time sections, each narrow band voice signal time section is classified as a voiced and/or unvoiced sound, first supplements exhibiting a spectral structure are produced in order to enlarge the narrow-band voice signal in relation to the above sound-type classification thus performed, second additions exhibiting a spectral structure are combined in order to enlarge the narrow-band voice signal based on generally known methods in order to evaluate the statistic properties of the narrow-band voice signal, the two additions and the narrow band spectral structure are combined in such a way that an enlarged spectral structure respectively occurs and a broadband expanded voice signal time section is produced on the basis thereof before a broadband enlarged voice signal is produced from the individual broadband enlarged voice signal time sections.