JP2004085963A - Navigation voice compensating device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve articulation of navigation voice by utilizing an existing navigation device without separately adding hardware such as a microphone and a microphone amplifier thereto. <P>SOLUTION: A microphone 11 is shared in voice recognition processing and articulation improving processing and either processing is performed by switching an operation mode by a mode switching section 14. Thereby the need of separately adding hardware such as the microphone and the microphone amplifier only for improving articulation is eliminated and cost an increase of the cost can be suppressed. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a navigation sound correction device, and is particularly suitable for use in a device for correcting the sound pressure level of a guidance sound of a navigation device according to ambient noise and the like.
[0002]
[Prior art]
2. Description of the Related Art In a navigation device for guiding a vehicle, a current position of the vehicle is detected, map data around the vehicle position is read from a recording medium such as a DVD, a map image is drawn on a display device, and a predetermined position of the map image is displayed. A vehicle position mark (own vehicle position mark) is overlaid on the drawing. Then, as the vehicle moves, the map image is scrolled so that the map information around the vehicle position can be understood at a glance.
[0003]
Such a navigation device has a function of setting a guidance route from a departure point to a destination, a function of displaying a guidance route on a map image, a function of giving a voice guidance of a turning direction at a guidance intersection on the guidance route, and the like. ing. That is, in this type of navigation device, a route with the smallest cost from the departure point to the destination is automatically searched for by using a map data to perform a simulation such as a breadth-first search (BFS) method or a Dijkstra method, and the search is performed. The obtained route is stored as a guidance route.
[0004]
While the vehicle is traveling, the guidance route is displayed on the map image so as to be distinguishable from other roads, and when the vehicle approaches within a predetermined distance from the guidance intersection on the guidance route, an intersection guide map (intersection enlarged view) And an arrow indicating the direction of travel at the intersection) are displayed to indicate which road should be run and in which direction the vehicle should travel at the intersection. Also, if the direction of travel at the intersection is a right turn, voice guidance such as "You are ahead right" is given, and if there is a considerable distance from the current vehicle position to the guidance intersection, "It will be a little while" Voice guidance is provided as appropriate.
[0005]
By the way, the above-mentioned guidance sound of the navigation apparatus (hereinafter referred to as navigation sound) is very audible when there are various noises (engine sound during running, road noise, wind noise, etc.) and car audio output sound. It becomes difficult. To solve such a problem, a system has been proposed in which predetermined signal processing is performed to enhance the clarity of the navigation sound (for example, Japanese Patent Application Laid-Open No. H11-166835).
[0006]
FIG. 7 is a block diagram showing a configuration of the conventional system. As shown in FIG. 7, in the conventional system, the audio sound output from the car speaker 1, the navigation sound output from the monitor speaker 2, and the running noise generated in the vehicle are all input from the microphone 3. The transfer function of the adaptive filter (ADF) 5 and the loudness compensation gain of the gain control unit 6 are variably controlled based on the audio signal input to the microphone, so that the transfer function is always appropriate regardless of the noise level. The navigation sound can be reproduced at the volume or the like.
[0007]
[Problems to be solved by the invention]
However, in the above-mentioned conventional system, it is necessary to use the microphone 3 for inputting audio sound, navigation sound and running noise in order to estimate the noise amount. Further, in this estimating unit, it is necessary to sequentially update the transfer function (filter coefficient) of the adaptive filter 5 in accordance with the amount of noise that is constantly changing in the vehicle interior. Therefore, the microphone 3 must always be in the active state.
[0008]
From the above, considering that the current navigation system is equipped with a function to improve the clarity of navigation voice, it is necessary to separately add hardware such as a microphone and a microphone amplifier for clarity improvement processing, which increases costs. Problem had arisen.
[0009]
The present invention has been made to solve such a problem, and the clarity of navigation sound can be improved by using an existing navigation device without separately adding hardware such as a microphone and a microphone amplifier. The aim is to be able to improve. It is another object of the present invention to realize this without deteriorating the performance of the clarity improvement processing.
[0010]
[Means for Solving the Problems]
In order to solve the above-described problems, in the navigation sound correction device of the present invention, the navigation device originally includes an utterance sound input microphone provided for the speech recognition process, and the navigation sound and the surrounding sound are included in the clarity improvement process. It is also used for inputting sound, and the operation mode is switched to one of the intelligibility improvement processing and the speech recognition processing to use the microphone input voice. Preferably, in the present invention, when the operation mode is switched to the speech recognition processing, control is performed so that the adaptive filter processing in the intelligibility improvement processing is stopped.
[0011]
According to the present invention configured as described above, it is not necessary to separately add hardware such as a microphone and a microphone amplifier only for the intelligibility improvement processing, and an existing navigation device is provided with a function for improving the intelligibility of navigation voice. It is possible to suppress an increase in cost when performing. In addition, the mode switching causes a sudden change in the level of the microphone input sound to the adaptive filter of the intelligibility improvement function. At that time, however, the adaptive filter processing is stopped. Receiving less.
[0012]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The navigation sound correction device according to the present embodiment monitors the sound pressure levels of engine noise, road noise, running noise such as wind noise, and audio sounds such as music, and appropriately adjusts the navigation sound according to these sound pressure levels. By giving a gain, the navigation sound can be easily heard.
[0013]
FIG. 1 is a block diagram illustrating a configuration of the navigation sound correction device according to the present embodiment. As shown in FIG. 1, the navigation audio correction device 100 of the present embodiment includes a monitor speaker 2, an adaptive filter (ADF) 5, a gain control unit 6, an equalizer 7, a microphone 11, a microphone amplifier 12, a talk switch 13, a mode switch. And a mode control unit 15.
[0014]
Note that, apart from the navigation sound correcting device 100, an audio sound reproducing device (not shown) and a sound recognition engine 20 are provided. The audio sound reproducing apparatus includes a reproducing unit (not shown) for reproducing various audio sources such as a CD, MD, DVD, and radio broadcast, and a car speaker 1, and an audio reproduced sound is output from the car speaker 1. . Further, the voice recognition engine 20 recognizes the speaker voice input through the microphone, and executes a command corresponding to the speaker voice to the navigation device.
[0015]
The microphone 11 is used for inputting the speaker's voice originally provided for the voice recognition processing of the navigation device, and is also used for the intelligibility improvement processing in the present embodiment. That is, not only the speaker's voice but also the audio sound output from the car speaker 1, the navigation voice output from the monitor speaker 2, and the running noise generated in the vehicle compartment are all input from the same microphone 11. The microphone amplifier 12 amplifies an audio signal input from the microphone 11.
[0016]
The talk switch 13 is a switch for instructing the start of voice recognition processing by utterance. That is, the voice input by utterance is accepted at the timing when the talk switch 13 is operated. The mode switching unit 14 includes a clarity improvement mode in which navigation voice, audio sound, and running noise are input from the microphone 11 to correct the navigation voice, and a voice recognition mode in which speech voice is input from the microphone 11 to perform voice recognition. The operation mode is switched to any one of.
[0017]
In general, the output of the navigation sound is performed in accordance with the current position during traveling, and it is not known at which timing the sound is output. Therefore, in order to improve the intelligibility of the navigation sound output at the undefined timing, the transfer function of the adaptive filter 5 must be constantly updated in accordance with the amount of noise that changes sequentially. Therefore, the mode switching unit 14 always operates the clarity improvement processing as a default operation mode, and switches to the voice recognition mode only when the talk switch 13 is turned on.
[0018]
When the mode is switched to the voice recognition mode, the uttered voice input from the microphone 11 is supplied to the voice recognition engine 20. Thereby, processing corresponding to the utterance command is executed. On the other hand, when the mode is switched to the intelligibility improvement mode, the navigation voice, audio sound, and running noise input from the microphone 11 are supplied to the subtractor 4. The subtracter 4 calculates an error by subtracting the navigation sound output from the adaptive filter 5 from the navigation sound, audio sound, and running noise input from the microphone 11, and extracts only the audio sound and running noise.
[0019]
The adaptive filter 5 includes an identification filter 21 and a voice correction filter 22, as shown in FIG. The identification filter 21 is a filter for identifying an impulse response of the acoustic system (a filter coefficient of the audio correction filter 22), and is an adaptive filter based on an LMS (Least Mean Square) algorithm or an N-LMS (Normalized-LMS) algorithm. Is used. The identification filter 21 operates so that the power of the error e (n) output from the subtractor 4 is minimized to identify the impulse response of the acoustic system.
[0020]
The voice correction filter 22 performs a convolution operation using the filter coefficient w (n) determined by the identification filter 21 and the navigation voice x (n) to be controlled, thereby converting the navigation voice x (n). On the other hand, the same transfer characteristics as those of the equalizer 7 are provided. The filtered navigation sound y (n) output from the sound correction filter 22 is supplied to the subtractor 4, and the error e (n) calculated here is fed back to the identification filter 21.
[0021]
The gain controller 6 converts the navigation sound after the filter control output from the adaptive filter 5 and the navigation sound to be controlled output from the navigation device based on the audio sound and the traveling noise output from the subtractor 4. An optimum gain to be added to the signal is calculated, and the calculated gain value is output to the equalizer 7.
[0022]
The equalizer 7 performs gain correction on the navigation sound. That is, by giving the gain calculated by the gain control unit 6 to the navigation sound input from the navigation device, the navigation sound is corrected. This correction is performed, for example, for each of a plurality of divided frequency bands.
[0023]
The mode control unit 15 controls the mode switching unit 14 according to ON / OFF of the talk switch 13 so as to switch the operation mode between the intelligibility improvement mode and the voice recognition mode. The mode control unit 15 also stops the adaptive filter 5 from adapting the transfer function when switching to the speech recognition mode. Then, at the stage when the speech recognition has been completed (the navigation device has set the flag of recognition end), control is performed so that the calculation of the adaptive filter 5 is restarted.
[0024]
As described above, in the intelligibility improvement mode, the subtractor 4 subtracts the estimated value of the navigation sound obtained by the adaptive filter 5 from the sound (navigation sound, audio sound, running noise) input from the microphone 11. By feeding back the difference value to the adaptive filter 5, a filter coefficient having a transfer function amplitude characteristic as shown in FIG. 3 is always calculated.
[0025]
In this state, when the operation mode is switched from the intelligibility improvement mode to the voice recognition mode in response to the operation of the talk switch 13, the voice level input from the microphone 11 to the subtractor 4 becomes zero. Conversely, when returning from the voice recognition mode to the intelligibility improvement mode, a large-level voice is suddenly input from the microphone 11 to the subtractor 4.
[0026]
As described above, when the operation mode is switched, the difference value calculated by the subtractor 4 fluctuates greatly. At this time, if the calculation of the adaptive filter 5 is continuously performed, the filter coefficient values calculated by the adaptive filter 5 are significantly different or diverge. As a result, the performance of separating navigation sound and surrounding sound (intelligibility improvement performance) is adversely affected.
[0027]
On the other hand, in the present embodiment, the adaptation process of the transfer function by the adaptive filter 5 is stopped during the speech recognition mode, and the adaptation process is restarted when returning to the intelligibility improvement mode. As a result, even if the audio input level to the subtractor 4 fluctuates greatly due to the switching of the operation mode, the transfer function is not estimated by the large fluctuation value at that time. Therefore, it is possible to avoid a disadvantage that the estimation accuracy of the transfer function is deteriorated (that is, the estimation error of the adaptive algorithm is increased).
[0028]
In the speech recognition mode, the adaptive processing of the transfer function by the adaptive filter 5 is temporarily stopped, but the influence of this is small. That is, in the current navigation device, it is recommended to perform the utterance operation “in a natural posture with the steering wheel held” in order to correctly recognize the voice. For this reason, a large change in the amount of ambient noise is unlikely to occur within a few seconds in the voice recognition mode, and there is no large difference in the amplitude characteristics of the transfer function before and after the switching of the operation mode.
[0029]
FIG. 4 is a diagram illustrating a difference between amplitude characteristics of a transfer function obtained before and after switching of the operation mode. In FIG. 4, the horizontal axis indicates frequency, and the vertical axis indicates the difference between transfer functions before and after switching. As shown in FIG. 4, there is no large difference between the amplitude characteristics of the transfer function before and after the switching of the operation mode. Therefore, even if the transfer function adaptation processing is temporarily stopped during the voice recognition mode, the transfer function estimation accuracy hardly deteriorates.
[0030]
FIG. 5 is a flowchart showing the operation when the talk switch 13 is pressed. In FIG. 5, when the talk switch 13 is pressed, the mode control unit 15 stops adaptation processing (update of filter coefficients) of the transfer function in the adaptive filter 5 (step S1). Specifically, a stop control signal is output to adaptive filter 5. Thereafter, the mode control unit 15 controls the mode switching unit 14 to switch the operation mode to the voice recognition mode (Step S2).
[0031]
Next, the navigation device outputs a signal sound to start accepting the input of the uttered voice (step S3), and then executes a voice recognition process (step S4). At the time of this voice recognition processing, cancellation of the voice recognition processing by pressing the talk switch 13 is also accepted. Next, the navigation device determines whether the voice recognition has been completed (step S5). If the voice recognition has not been completed, the voice recognition processing in step S4 is continued.
[0032]
On the other hand, when the voice recognition processing is completed, the mode control unit 15 controls the mode switching unit 14 to switch the operation mode to the intelligibility improvement mode (step S6). At this time, the adaptive processing of the transfer function in the adaptive filter 5 (update of the filter coefficient) is restarted. Specifically, the output of the stop control signal is stopped. After that, the command whose voice has been recognized in step S4 is executed (step S7). The reason why the operation mode is switched before the command is executed is to start a time-consuming transfer function calculation as soon as possible.
[0033]
FIG. 6 is a flowchart showing the operation of the adaptive filter 5. FIG. 6 shows an example in which the LMS algorithm is applied to the adaptive filter 5. In FIG. 6, the adaptive filter 5 includes data necessary for calculation (a navigation sound x (n) to be controlled in the intelligibility improvement processing, a subtractor 4) in response to an interrupt at a predetermined time interval 1 / Fs. The error e (n) or the like fed back from the above is fetched (step S11).
[0034]
Then, in the voice correction filter 22 of FIG. 2, a convolution operation is performed using the filter coefficient w (n) determined by the identification filter 21 and the input navigation voice x (n) to be controlled. Thus, the navigation voice y (n) after the filter control is obtained (step S12).
[0035]
Next, the adaptive filter 5 determines whether the talk switch 13 is turned on, that is, whether it is in the voice recognition mode (step S13). During the voice recognition mode, a stop control signal (flag) is sent from the mode control unit 15 to the adaptive filter 5, so that the state of the talk switch 13 can be determined by looking at this flag.
[0036]
If the talk switch 13 is not active, the filter coefficient is updated in the identification filter 21 as usual (step S15). On the other hand, when the talk switch 13 is active, the interrupt processing ends without performing the update processing of the filter coefficient in the identification filter 21. That is, during the speech recognition mode, the filter coefficient of the adaptive filter 5 is handled as w (n + 1) = w (n), and only the convolution operation is performed during that time.
[0037]
As described above in detail, in the present embodiment, the microphone 11 is shared between the voice recognition process and the intelligibility improvement process, and the operation mode is switched according to the operation of the talk switch 13. As a result, it is not necessary to separately add hardware such as a microphone and a microphone amplifier only for the intelligibility improvement processing, and it is possible to suppress an increase in cost when the voice intelligibility improvement function is installed in an existing navigation device. .
[0038]
Further, in the present embodiment, during the operation in the voice recognition mode, the adaptive processing of the transfer function in the adaptive filter 5 is stopped. Thus, even if the level of the microphone input sound to the adaptive filter 5 changes abruptly by switching the operation mode, the abrupt level fluctuation does not adversely affect the estimation accuracy of the transfer function. Can be avoided.
[0039]
As described above, according to the present embodiment, both the intelligibility improvement processing and the voice recognition processing can be performed simultaneously without deteriorating the performance using one microphone 11 and one microphone amplifier 12.
[0040]
Note that the above-described embodiment is merely an example of a concrete example for carrying out the present invention, and the technical scope of the present invention should not be interpreted in a limited manner. That is, the present invention can be embodied in various forms without departing from the spirit or main features thereof.
[0041]
【The invention's effect】
As described above, the present invention shares the same voice input device for the voice recognition process and the intelligibility improvement process, and switches the operation mode to perform any process. This eliminates the need to separately add hardware such as a microphone and a microphone amplifier, thereby suppressing an increase in cost.
[0042]
According to another feature of the present invention, the adaptive filter processing in the intelligibility improvement processing is stopped during the operation in the voice recognition mode, so that the level of the microphone input voice suddenly changes according to the mode switching. However, the abrupt level fluctuation does not adversely affect the estimation accuracy of the transfer function, and the performance of both the intelligibility improvement processing and the speech recognition processing is reduced by using one speech input device for speech recognition. Can be processed without the need.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a configuration example of a navigation sound correction device according to an embodiment.
FIG. 2 is a diagram illustrating a configuration of an adaptive filter.
FIG. 3 is a diagram illustrating amplitude characteristics of a transfer function in an adaptive filter.
FIG. 4 is a diagram illustrating a difference between amplitude characteristics of a transfer function before and after switching operation modes.
FIG. 5 is a flowchart illustrating an operation when a talk switch is pressed.
FIG. 6 is a flowchart showing an operation of the adaptive filter.
FIG. 7 is a block diagram showing a configuration of a conventional navigation sound correction device.
[Explanation of symbols]
4 Subtractor 5 Adaptive filter 6 Gain control unit 7 Equalizer 11 Microphone 12 Microphone amplifier 13 Talk switch 14 Mode switching unit 15 Mode control unit 21 Identification filter 22 Voice correction filter 100 Navigation voice correction device

Claims (2)

A navigation sound correction device configured to improve the clarity of the guidance sound based on the sound pressure level of the guidance sound of the navigation device and the sound pressure level of the ambient sound that can be heard at the listening position of the guidance sound. And
Voice input means for inputting the guidance voice and the surrounding sound used for the intelligibility improvement processing of the guidance voice and the uttered voice used for the voice recognition processing from the same voice input device;
A navigation sound correction device comprising: a mode switching unit that switches an operation mode to one of the clarity improvement processing and the speech recognition processing. 2. The navigation voice according to claim 1, further comprising control means for controlling to stop the adaptive filter processing in the clarity improvement processing when the operation mode is switched to the voice recognition processing. Correction device.

Images