Abstract: An optical head assembly for use with an optical disk apparatus includes a magnetic circuit block having permanent magnets each being elongate in a track traversing direction and a yoke for forming a magnetic circuit which conducts a magnetic field generated by the permanent magnets. The yoke includes a first through hole extending in a focusing direction, magnetic gaps located at opposite sides of the first through hole in the track traversing direction and individually accommodating the permanent magnets, and a second through hole for admitting the light beam issuing in the track traversing direction to guide it to the first through hole. The first through hole and magnetic gaps each has a cross-section relatively long in the track traversing direction. A beam deflector is loosely fitted in the first through hole for deflecting in the focusing direction the light beam which is incident thereto from the second through hole. The beam deflector is supported by a carriage.

Abstract: This speech recognition system uses a converted-spectrum for comparison of unknown and prestored reference signals. The comparison distance is modified by a weighting factor W to emphasize differences. A straight line approximation of the short term spectrum provides a datum against which the original spectrum is differenced to provide the converted spectrum, and provide deviation signs which are used to modify an initial prestored weighting factor P to form W.

Abstract: Speech recognition with time warp is simplified by finding a certain portion of a word whose time duration is the same for all speakers. In comparing an unknown speech with a reference speech, the time duration of an unknown speech is coincided with the time length of a reference speech with the two processes. According to the invention, an element vector of a speech is classified to the first portion and the second portion. The former is a consonant and co-articulation which couples the two sounds, and the latter is a vowel. The length of the first portion is almost independent from a speaker, and the length of the second portion depends upon a speaker. Therefore, the present invention matches the first portion of an unknown speech with that of the reference speech directly without changing the time length. Next, the sample elements in the second portion of the unknown speech is linearly matched with that of a reference speech. Thus, excellent recognition is obtained using a simple calculation.