摘要 |
<p>The process involves forming a spectral image of recorded sound. Filters defining a frequency band and energy band are selected. Formant components, which are fragments of windows constituted of points close to the frequencies and energies, are identified. A descriptor for each formant and a distance between two formants are calculated. The descriptor of one formant is compared with the descriptor of the other formant. An independent claim is also included for a tele-monitoring system having a system of identifying a specific sound in a noisy environment.</p> |