A speaker independent algorithm is given which automatically detects the most steady-state portion of a vowel (vowel center) from continuous speech. The algorithm first extracts the segments each of which contains a vowel and, if present, pre- and/or post-vocalic liquids and semivowels, and then locates the most steady-state portion of the segment. An advantage of the algorithm is its ability to distinguish the nasal and the intervocalic liquid and semivowel segments without relying upon the formant frequencies which have been used in most of the previous work of vowel segment detection procedure. This results in a computationally simple algorithm. The test on ten sentences spoken by each of two males and two females resulted in score of 93.2% correct vowel center localization.