| 摘要 |
For the automatic recognition of signal sequences, such as for example speech signals and/or music signals, in each case envelope curves (rectifier 3) are formed from predetermined signals and from signals to be examined and these envelope curves are then band-limited to a narrow, low frequency range (e.g. 2 to 50 Hz). Timing excerpts (approx. 1.7 sec.) are if appropriate stored in a memory (6) as "masks" after digitalisation. When monitoring transmissions, the band-limited envelope curve signals of the signals to be examined are then buffered in a second memory (7) in sections of equal length (of approx. 1.7 sec. in length). Within the period between the arrival of two successive signals to be examined the contents of the buffer (7) are then compared (comparator 8) with all the masks of the first memory (6), a detection signal being generated in the case of a predetermined degree (e.g. 95%) of correspondence.
<IMAGE> |
