| US 7,487,084 B2 | ||
| Apparatus, program storage device and method for testing speech recognition in the mobile environment of a vehicle | ||
| Andrew Aaron, Ardsley, N.Y. (US); Subrata K. Das, Granite Springs, N.Y. (US); and David M. Lubensky, Danbury, Conn. (US) | ||
| Assigned to International Business Machines Corporation, Armonk, N.Y. (US) | ||
| Filed on Jul. 31, 2002, as Appl. No. 10/210,667. | ||
| Claims priority of provisional application 60/341504, filed on Oct. 30, 2001. | ||
| Prior Publication US 2003/0236672 A1, Dec. 25, 2003 | ||
| Int. Cl. G10L 15/00 (2006.01) | ||
| U.S. Cl. 704—231 [704/270; 704/275; 704/251; 704/228; 704/233] | 19 Claims |

| 1. An apparatus for testing speech recognition in a new vehicle, said apparatus comprising:
a laptop computer;
a speaker arrangement which propagates a speech output based on a known text previously recorded by a human; wherein the text
is comprised of a set of commands of interest, said speech output being stored digitally in a laptop computer;
a testing arrangement adapted to test the acceptability of installed speech recognition systems in vehicles while the vehicles
are being operated on a roadway at speeds of 0, 30, and 60 miles per hour, wherein the acceptability of a particular vehicular
speech recognition system is based upon a comparison of pre-specified standards of recognition accuracy and signal-to-noise
ratio values with a recognition accuracy value and a signal-to-noise ratio value produced by the particular vehicular speech
recognition system, based on the text recognized from the speech output and the testing arrangement is located separate from
the vehicle being tested, the propagated speech output being transmitted to the testing arrangement via a cellular transmission
unit located within the vehicle being tested;
wherein said speaker arrangement is configured to simulate the propagation of a human voice and is calibrated with respect
to an audio input of an installed speech recognition system in a vehicle such that the speech output is propagated at the
same pressure and distance from a microphone used to record the human speech;
wherein the speech output comprises three recordings of the human voice, one recording having been made while the human wore
a headset playing noise typical of an automobile operating on a roadway at a speed of 0 miles per hour, one recording having
been made while the human wore a headset playing noise typical of an automobile operating on a roadway at a speed of 30 miles,
and one recording having been made while the human wore a headset playing noise typical of an automobile operating on a roadway
at a speed of 60 miles per hour;
wherein the recording having been made while the human wore a headset playing noise typical of an automobile operating on
a roadway at a speed of 0 miles per hour is played during the test when the vehicle is being operated on a roadway at a speed
of 0 miles per hour;
wherein the recording having been made while the human wore a headset playing noise typical of an automobile operating on
a roadway at a speed of 30 miles per hour is played during the test when the vehicle is being operated on a roadway at a speed
of 30 miles per hour;
wherein the recording having been made while the human wore a headset playing noise typical of an automobile operating on
a roadway at a speed of 60 miles per hour is played during the test when the vehicle is being operated on a roadway at a speed
of 60 miles per hour; and
wherein the apparatus is used to test every one-hundredth car that leaves an assembly line.
|