The Investigation of a Technique for the Identification of Several Major Languages

The problem of automatically identifying a language from a spoken sample has, until recently, had very limited attention. Whenever speech exists, it can be categorized, according to the source of excitation of the vocal tract, into one of the following four groups: voiced nonfricative, voiced fricative, unvoiced fricative and no speech. The identification technique makes use of a measurement of the time the waveform spends within each categorized group of each language. The fraction of time the speech spends within each group is expected to vary for each language. The results indicate that for Chinese, English, German and Japanese, their time fractions for each corresponding group are distributed within some intervals which overlap greatly