Data processing: speech signal processing – linguistics – language – Speech signal processing – Application
Reexamination Certificate
2000-05-02
2002-05-07
Dorvil, Richemond (Department: 2741)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Application
C704S270100, C704S275000, C704S231000, C704S251000
Reexamination Certificate
active
06385582
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a man-machine system having speech recognition functions, and more specifically, to a man-machine system in which a user can input desired instructions in a simple manner at the user side, and in which desired processes can be performed properly in accordance with the user instructions at the man-machine system side.
Although the concept of man-machine systems initially concerned a system (device) constructed for enhancing the respective advantages of human and computers, nowadays it is said that this concept also embraces systems which facilitate relationships between human and more general machines (machines in a broader sense) as well as computers.
Man-machine systems, such as systems equipped with a speech recognition device in which a speaker (user) can instruct (command) its intention through voice input, are known. For example, a navigation system using a GPS (Global Positioning System) cruising scheme for automobiles is known. In this navigation system, users can designate a destination, etc., through voice input.
When a user pronounces a desired destination, this navigation system speech-recognizes the destination; searches a driving path from the present location to the destination; and displays the searched driving path on a map through a display device.
For example, the navigation system recognizes the destination designated by the user by performing predetermined steps shown in FIG.
13
. Suppose that the user wishes to know a driving path to “Meguro station”, which is the destination. First, a voice synthesizer in the speech recognition device generates a synthesized sound of “Please enter the name” in order to request the user to voice-input (pronounce) a specific destination name. If the user pronounces “Meguro station” in response, the speech recognition device extracts the characteristics of the voice of “Meguro station”, and temporarily stores the extracted characteristic parameters D
1
in a memory part or the like. That is, at the first step, the speech recognition device only extracts the characteristics of the voice of “Meguro station” without performing final-stage recognition.
Next, at the second step, the voice synthesizer generates a synthesized sound of “Please enter a genre” in order to request the user to pronounce a genre, which is a higher level concept than the specific destination the user desires.
If the user pronounces “train station name” in response, the speech recognition device extracts the characteristics of this voice of “train station name” to generate the corresponding characteristic parameters D
2
. Further, the speech recognition device compares the characteristic parameter D
2
with recognition reference vocabularies in a recognition word dictionary which has been pre-installed in the speech recognition device, and selects a recognition reference vocabulary LD
2
which is most similar to the characteristic parameter D
2
, thereby conducting speech recognition of the voice of “train station name” pronounced by the user.
Next, at the step
3
, the voice synthesizer generates a synthesized sound of “Please enter a prefecture name” to request the user to pronounce a region name.
If the user pronounces “Tokyo” in response, the speech recognition device extracts the characteristics of this voice of “Tokyo” to generate the corresponding characteristic parameters D
3
. Further, the speech recognition device compares the characteristic parameter D
3
with recognition reference vocabularies in the recognition word dictionary, and selects a recognition reference vocabulary LD
3
which is most similar to the characteristic parameter D
3
, thereby conducting speech recognition of the voice of “Tokyo” pronounced by the user.
Next, at the step
4
, among recognition reference vocabularies in the recognition word dictionary, the speech recognition device narrows down recognition reference vocabularies to the ones belonging to the categories of the recognition reference vocabularies LD
2
and LD
3
. Further, the speech recognition device compares the characteristic parameter D
1
With the narrowed-down recognition reference vocabularies to select a recognition reference vocabulary LD
1
which is most similar to the characteristic parameters D
1
, thereby conducting speech recognition of the voice of “Meguro station” pronounced at the first step.
That is, at the first step where the lower level concept of the name “Meguro station” is pronounced, it is in general difficult to identify the recognition reference vocabulary LD
1
corresponding to “Meguro station”, which exists within the region the user desires.
Because of this difficulty, the characteristic parameters D
1
of the pronounced voice of “Meguro station” are first stored in the memory part. Then, at the second through fourth steps, a searching range for recognition reference vocabularies in the recognition word dictionary is narrowed down by receiving voices of the genre and region name from the user. Then, by comparing the characteristic parameters D
1
with the thus narrowed-down recognition reference vocabularies, the recognition reference vocabulary LD
1
corresponding to “Meguro station” is relatively easily identified.
Finally, based upon the selected recognition reference vocabularies LD
3
and LD
1
, a synthesized sound of “It is ◯◯◯ in &Dgr;&Dgr;&Dgr;, isn't it?” is generated to provide the user with the recognition result. That is, when the recognition reference vocabularies LD
3
and LD
1
are properly recognized as “Tokyo” and “Meguro station”, respectively, the synthesized sound of “It is ◯◯◯ in &Dgr;&Dgr;&Dgr;, isn't it?” becomes a synthesized sound of “It is Meguro station in Tokyo, isn't it?”, and is presented to the user as such.
Thus, the speech recognition device merely selects the recognition reference vocabularies LD
1
to LD
3
from the recognition word dictionary as the vocabularies most similar to the respective words pronounced by the user. Accordingly, there is an inevitable possibility that “Meguro station” is wrongly recognized as “Mejiro station”, or “Tokyo” is wrongly recognized as “Kyoto”, etc., in the case where the user's pronounced voice was not clear or in some other circumstances. If such misrecognition occurs, a synthesized sound of “It is Mejiro station in Kyoto, isn't it?” would be presented to the user. Thus, the synthesized sound is generated based on recognition reference vocabularies LD
3
and LD
1
in order to ask for the user confirmation of the recognition results, as described above.
If the user determines that correct speech recognition is performed by hearing this synthesized voice thus presented, the user pronounces “search start”, for example. Then, the speech recognition device recognizes this, and the navigation system receives a confirmation instruction and searches a driving path from the current location to the Meguro station in Tokyo. The navigation system then displays the searched driving path on a map through a display device.
On the other hand, if the user determines that the recognition is wrong, the user indicates so by pronouncing “return”. Receiving that instruction, the speech recognition device restarts speech recognition, and repeats the speech recognition until it receives the instruction of “search start” from the user with respect to re-presented recognition result.
As explained above, the navigation system possesses a superior functionality in that it enables conversational operations by the combination of a speech recognition device and a voice synthesizer.
Also, because the user is lead to pronounce words, which become keywords, in the order which matches the user's thought characteristics, the system provides the user with improved convenience. In other words, in designating the desired destination, the user designates the most specific destination (Meguro station in the example above), and then designates its genre and region name where that destination exits. Thus,
Arent Fox Kintner & Plotkin & Kahn, PLLC
Dorvil Richemond
Nolan Daniel A.
Pioneer Corporation
LandOfFree
Man-machine system equipped with speech recognition device does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Man-machine system equipped with speech recognition device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Man-machine system equipped with speech recognition device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2883614