It was the objective of the study reported here to analyze a variety of Asian dialects for the purpose of discovering spectrum distribution differences that may be correlated to the dialects. The dialects subjected to the analysis were North (10), Central (10), and South (10) Vietnamese; Cantonese (5), Mandarin (5), and Fukienese (4) Chinese; Cambodian (15) and the highland tribal dialects of Jari (3), Bahnar (3), Sedang (3), Rade (3), Cham (3), Koho (3), and Hre (11). The number of individuals sampled in each dialect is shown by the digit in the parentheses. Confidence in the results should be rated accordingly. All samples were taken from male talkers. Spectrum analysis was performed by means of a 31-channel 1/3-oct band filter bank over the frequency range from 20 to 20 000 Hz, scanned at a rate of 60 Hz. The speech source data were collected on high-quality portable tape recorders and these, after appropriate editing to remove unwanted background, were played into the 1/3-oct band spectrum analyzer. The spectra were then digitized and transferred to an IBM 1410 computer for analysis for mean spectrum distribution for each dialect. Voiced and unvoiced events were segmented into separate groups and separate spectra for each of these classes thereby obtained. For the summary analysis, all spectrum scans of the voiced type and unvoiced type were individually averaged to produce an average voiced event spectrum and an average unvoiced event spectrum for each dialect. A tally of the number of spectra for each excitation class for each dialect also permitted the estimation of the ratio of voiced to unvoiced excitation that is characteristic of each dialect. The summary analysis showed that, generally, the voiced event mean spectrum differed only in minor detail as a function of dialect. The same was found true of the unvoiced event mean spectrum. The most variable parameter appeared to be the percentage of time occupied by voiced excitation. This parameter varied from 76% for Mandarin-Chinese to 90% for Cambodian, Jari, and Hre. [Dr. Campanella was employed at Melpar, Inc. at the time of performance of the study reported in this paper. The work was performed under Contract Booze-Allen Corp. subcontract 5-66-1]