An automatic pruning down of a wiretapped message or a recorded speech for specific words or phrases contained in it can alleviate much of labor of an intelligence agent. Given a reasonable success of recognition of a...
详细信息
ISBN:
(纸本)1932415513
An automatic pruning down of a wiretapped message or a recorded speech for specific words or phrases contained in it can alleviate much of labor of an intelligence agent. Given a reasonable success of recognition of a desired phrase in a recording, we have shown in this paper, we expect that automation and integration of our algorithms could lead to automatic recognition of a given phrase in a recording. In this paper a simple scheme of recognition of a desired phrase is described and the results are presented. The utterances are scanned for formants F1 and F2 of its vowels. Pratt or another good software can be used for this purpose. Having determined the first two formants for each vowel contained in the recordings, the vowels are identified using Peterson and Barney vowel loops. Successful matching of a string of vowels and vowel-sequence contained in the recording with the previously stored vowels and vowels-sequence of the phrase can determine if the phrase looked for in the utterances is present. The sequence of the vowel string is important. Once successful matching has been achieved, the entire message can then be listened to for verification. Consonants are not included for recognition. The results on specific phrase "Usama Bin Laden" spoken in a short utterance are encouraging.
The Directed Acyclic Word Graph (DAWG) is an efficient data structure to treat and analyze repetitions in a text, especially in DNA genomic sequences. Here, we consider the Compact Directed Acyclic Word Graph of a wor...
详细信息
ISBN:
(纸本)3540632204
The Directed Acyclic Word Graph (DAWG) is an efficient data structure to treat and analyze repetitions in a text, especially in DNA genomic sequences. Here, we consider the Compact Directed Acyclic Word Graph of a word. We give the first direct algorithm to construct it. It runs in time linear in the length of the string on a fixed alphabet. Our implementation requires half the memory space used by DAWGs.
A quantitative analysis of the widely recognized inefficiency of the SNOBOL 4 pattern matching algorithm is presented. The possibility of increasing the efficiency of patternmatching by special case processing is dis...
详细信息
A quantitative analysis of the widely recognized inefficiency of the SNOBOL 4 pattern matching algorithm is presented. The possibility of increasing the efficiency of patternmatching by special case processing is discussed and a new approach for string processing languages design along this line is proposed.
暂无评论