Discriminability of the Quality of Amplitude-Compressed Speech In an earlier experiment on intelligibility of amplitude-compressed speech, subjects could not hear a difference between noncompressed speech and speech under some conditions of compression. Therefore, compression conditions were determined in which the quality of the two types of speech could be distinguished. When speech average level was 10 dB ... Research Article
Research Article  |   December 01, 1984
Discriminability of the Quality of Amplitude-Compressed Speech
 
Author Affiliations & Notes
  • Igor V. Náblek
    University of Tennessee, Knoxville
Article Information
Research Articles
Research Article   |   December 01, 1984
Discriminability of the Quality of Amplitude-Compressed Speech
Journal of Speech, Language, and Hearing Research, December 1984, Vol. 27, 571-577. doi:10.1044/jshr.2704.571
History: Received February 9, 1982 , Accepted July 2, 1984
 
Journal of Speech, Language, and Hearing Research, December 1984, Vol. 27, 571-577. doi:10.1044/jshr.2704.571
History: Received February 9, 1982; Accepted July 2, 1984

In an earlier experiment on intelligibility of amplitude-compressed speech, subjects could not hear a difference between noncompressed speech and speech under some conditions of compression. Therefore, compression conditions were determined in which the quality of the two types of speech could be distinguished. When speech average level was 10 dB above a masking noise, compression ratio (CR) was equal to 2.5, and the attack time (Ta) was 3 ms, the release time (Tr) had to be shorter than 120 ms to achieve discrimination by trained normal-hearing subjects. With longer attack times and/or higher compression ratios, the critical value of release times increased. Thus, the range in which the discrimination was observed also increased (for CR = 5 and Ta = 10 ms, the critical Tr was 360 ms). The discrimination of our hearing-impaired subjects was much worse than that of the normal-hearing subjects. For example, speech processed with CR = 10, Ta = 1 ms, and Tr = 10 ms could be distinguished from the noncompressed by only 50% of the impaired subjects.

Order a Subscription
Pay Per View
Entire Journal of Speech, Language, and Hearing Research content & archive
24-hour access
This Article
24-hour access