In:
The Journal of the Acoustical Society of America, Acoustical Society of America (ASA), Vol. 26, No. 1_Supplement ( 1954-01-01), p. 137-138
Abstract:
Automatic devices that respond selectively to particular phonetic elements in the stream of speech have been built by Howland, Smith, and Davis, and an automatic digit-recognizer has been developed by Davis, Biddulph, and Belashek. The digit-recognizer responds accurately to spoken numerals, but mistakes other sounds for numerals. The word-recognizer described here is designed with the aim of responding always to a given word, never to any other word. It has the disadvantage of wording only on selected words, but this is not a disadvantage in the application for which it was intended: selecting telephone lines in response to key words from operators in information centers whose hands are too busy to push buttons. The word-recognizer combines Howland's phoneme-detector with a system of time delays and gates that is sensitive to a particular temporal sequence of speech sounds, For example, if the key word is “butterfat” (a particularly good word), a time delay is started every time an “uh” occurs. This delay lasts until the “er” is due and opens a gate to the “er” recognizer at that time. If the “er” comes on schedule, a second delay is started. It lasts until the “a” is due, etc. When all the contingencies are satisfied, the device connects the speaker to the appropriate communication line. The recognizer works best for the speaker for whom it was adjusted, but others can operate it with fair reliability.
Type of Medium:
Online Resource
ISSN:
0001-4966
,
1520-8524
Language:
English
Publisher:
Acoustical Society of America (ASA)
Publication Date:
1954
detail.hit.zdb_id:
1461063-2
Permalink