RESUMO
Long-term speech spectra from Finnish and English are analysed by a computer-based method which offers a rapid and accurate way of measuring these spectra from recorded material (single words, list of words, music, etc.). The signal processing scheme is illustrated. The speech spectra of a single Finnish word and the summation spectra from 50 phonetically balanced English and Finnish words are presented. A significant difference in the distribution of speech power is shown between. English and Finnish which is thought to be due to the difference in the vowel/consonant ratio between these languages. The value of the method as an aid to the comparative audiometric study of clinical findings and diagnoses between different languages is discussed.