摘要 |
An audio source (100) is amplified (102), filtered (104) and subsequently digitized (108) so that a Fourier transform may be performed by a digital signal processor (112). The frequency components of interest are then formed into histograms over a time period on the order of 5 minutes of speech sampled every 16 ms. The histogram and recognition of the audio source are performed by a computer (114) driven by an appropriately programmed algorithm. |