JP2004133051A - Voice input device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To stabilize the signal level of an inputted input voice signal in a voice input device. <P>SOLUTION: In a voice input device 1 which is provided with a voice amplifying circuit 13 which amplifies the input voice signal by changing a gain according to a gain control signal and a gain control circuit 16 which generates a gain control signal from the signal level of the amplified input voice signal and which outputs the gain control signal to the voice amplifying circuit 13, the gain control circuit 16 is provided with a time constant switching control signal generating means 33 which generates a time constant switching control signal for switching a time constant and a time constant switching means 30 which switches the time constant of the circuit for generating the gain control signal according to a time constant switching control signal to be inputted. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a voice input device, and more particularly, to a voice input device that stabilizes a signal level of an input voice signal.
[0002]
[Prior art]
With recent advances in speech recognition technology, applications for inputting speech signals to personal computers and the like have increased dramatically. However, in order for a personal computer or the like to accurately recognize a human voice signal, the signal quality of the input voice signal is sufficiently improved by improving the S / N ratio of the input voice signal, stabilizing the input level, separating environmental sounds, and the like. Need to be increased. In particular, since the signal strength of a voice signal spoken by a person fluctuates due to individual differences between speakers and the distance between the speaker and the microphone, it is very difficult to input this voice signal to the microphone with a stable signal strength. Difficult. If the signal level of the input speech signal is not supplied stably, no matter how much the speech recognition engine is improved, the overall speech recognition rate cannot be increased.
[0003]
Therefore, it is common practice to use a circuit called a compressor to keep the signal level of an audio signal having a variable signal strength input from a microphone constant. This compressor consists of a variable gain amplifier and an audio level detection circuit. When the audio signal intensity is low, the amplifier gain is increased, and when the audio signal intensity is high, the amplifier gain is decreased. It operates to maintain a constant level of audio signal.
[0004]
FIG. 4 is a block diagram showing a general configuration of a voice input device using a conventional compressor. In the audio input device 101 shown in FIG. 1, an audio signal input from a microphone 11 connected to a microphone plug 12 is amplified by an audio amplification circuit 13 having a variable gain, and is converted into a digital signal by an A / D conversion circuit 14. Are output to the CPU 15 as a control device. The gain control circuit 16 includes a diode 17, a charge resistor 19, a capacitor 20, and a discharge resistor 21. Among these circuit elements, the diode 17 detects a change in the output level of the audio amplification circuit 13. Further, a circuit constituted by the elements of the charging resistor 19, the capacitor 20, and the discharging resistor 21 holds the audio signal level detected by the diode 17. The output of the gain control circuit 16 composed of these circuit elements becomes a gain control signal for controlling the gain of the audio amplification circuit 13. Note that the audio amplifier circuit 13 corresponds to an amplifier of variable gain of the compressor function described above, and the gain control circuit 16 corresponds to an audio level detection circuit of the compressor function.
[0005]
The output of the CPU 15 is transmitted to a host 23 equipped with a voice recognition engine or the like when the talk switch 22 is turned on. The capacitor 20 holds the peak value of the gain control signal, the charging resistor 19 determines the charging time of the gain control signal charged in the capacitor 20, and the discharging resistor 21 discharges the gain control signal charged in the capacitor 20. Determine the time.
[0006]
Next, the operation of the gain control circuit 16 will be described. The audio signal input to the microphone 11 is amplified by the audio amplification circuit 13 and input to the gain control circuit 16. At this time, the capacitor 20 is charged via the diode 17 and the charging resistor 19. The signal charged in the capacitor 20 is discharged through the discharge resistor 21. Since the discharge resistor 21 is normally set to a larger value than the charge resistor 19, it takes time for the charge charged in the capacitor 20 to be discharged. , And the gain control circuit 16 functions as a peak hold circuit. As described above, the gain control circuit 16 follows a sudden change in the audio signal, quickly transmits the change as a control signal to the audio amplifier circuit 13, and when there is no rapid change in the audio signal, In order to limit the distortion of, the operation is performed to keep the gain as stable as possible.
[0007]
[Problems to be solved by the invention]
However, according to the audio input device using the above-mentioned prior art compressor, the compressor works to keep the peak of the input audio signal constant, but on the other hand, when the power is turned on, when the microphone plug is inserted, the talk switch is turned off. When an excessive signal (noise) is input, for example, when the button is pressed, the gain of the audio amplifier circuit 13 is extremely lowered, and the recovery of the gain is delayed, so that the audio amplifier circuit 13 cannot be used until the gain returns.
[0008]
FIG. 5 is a signal waveform diagram showing a state of a microphone input, a gain control circuit output, and an audio amplifier circuit output due to noise when a microphone plug is inserted. In the figure, when the microphone 11 is inserted into the microphone plug 12, as shown in the microphone input of the figure, an excessive signal (noise) different from that at the time of the normal signal is generated. , A control signal which rises rapidly as shown in FIG. This control signal operates so as to fall slowly (for several tens of seconds) even after the microphone plug insertion noise has disappeared, for the above-mentioned reason. Therefore, for several tens of seconds, the output of the audio amplification circuit 13 becomes small, and the audio input device 101 cannot be used.
[0009]
Now, it is conceivable to reduce the time constant of the discharge of the gain control circuit 16 in order to shorten the time until the gain returns. However, if the response of the gain control becomes faster, the amplitude is controlled to a constant amplitude up to the normal strength of the utterance, and as a result, there is a problem that an unnatural sound which is hard to hear is produced.
[0010]
The present invention has been made in view of the above, and an object of the present invention is to provide a sound input device that stabilizes a signal level while maintaining a natural state of an input sound signal even when an excessive signal is input. And
[0011]
[Means for Solving the Problems]
In order to solve the above-described problems and achieve the object, in the audio input device according to the present invention, an audio amplifier circuit that changes a gain according to a gain control signal and amplifies an input audio signal, A gain control signal that is generated from the signal level of the input audio signal, and a gain control circuit that outputs the gain control signal to the audio amplification circuit. A time constant switching control signal generating unit that generates a constant switching control signal; and a time constant switching unit that switches a time constant of a circuit that generates the gain control signal in accordance with the input time constant switching control signal. And
[0012]
According to the present invention, the time constant switching control signal generating means generates a time constant switching control signal for switching the time constant, and the time constant switching means to which the time constant switching control signal has been input, generates the time constant switching control signal. The circuit operates to switch the time constant of the circuit that generates the gain control signal according to the control signal.
[0013]
In the voice input device according to the next invention, the time constant switching control signal generating means generates a time constant switching control signal for switching a time constant at a predetermined time immediately after power-on.
[0014]
According to the present invention, the time constant switching control signal generating means can generate the time constant switching control signal for switching the time constant at a predetermined time immediately after the power is turned on.
[0015]
In the audio input device according to the next invention, the time constant switching control signal generation means detects that the microphone is inserted into the microphone plug, and switches the time constant at a predetermined time after the detection. It is characterized by generating a signal.
[0016]
According to the present invention, the time constant switching control signal generating means can detect that the microphone has been inserted into the microphone plug, and can generate the time constant switching control signal for switching the time constant at a predetermined time after the detection.
[0017]
In the audio input device according to the next invention, the time constant switching control signal generation means detects that the talk switch has been pressed, and outputs a time constant switching control signal for switching the time constant at a predetermined time after the detection. It is characterized by generating.
[0018]
According to the present invention, the time constant switching control signal generating means can detect that the talk switch has been pressed, and can generate the time constant switching control signal for switching the time constant during a predetermined time after the detection.
[0019]
In the audio input device according to the next invention, the pulse width of the time constant switching control signal generated when a microphone is inserted into a microphone plug or the time constant generated when a talk switch is pressed. The pulse width of the switching control signal is smaller than the pulse width of the time constant switching control signal generated at a predetermined time immediately after the power is turned on.
[0020]
According to the present invention, the pulse width of the time constant switching control signal generated at a predetermined time immediately after the power is turned on is generated when the microphone plug is inserted or the time constant switching control generated when the talk switch is pressed. By setting the predetermined pulse width longer than each of the signal pulse widths, the voice input device can be effectively stabilized.
[0021]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of a voice input device according to the present invention will be described in detail with reference to the drawings. It should be noted that the present invention is not limited by the embodiment.
[0022]
FIG. 1 is a block diagram showing a configuration of a voice input device according to the present invention. In the audio input device 1 shown in FIG. 1, an audio signal input from a microphone 11 connected to a microphone plug 12 is amplified by an audio amplification circuit 13 having a variable gain, and is converted into a digital signal by an A / D conversion circuit 14. It is output to CPU 15 which is a control device. The output of the CPU 15 is transmitted to a host 23 equipped with a voice recognition engine or the like when the talk switch 22 is turned on. The talk switch 22 is a switch for preventing an unintentional audio signal from being output to the host 23. That is, the sound signal is transmitted to the host 23 when the talk switch 22 is turned on by the speaker, and is not transmitted to the host 23 in other cases.
[0023]
The gain control circuit 16 includes a gain control signal generation circuit 18 including a diode 17, a charging resistor 19, a capacitor 20, and a discharge resistor 21, a time constant switching unit 30 including a transistor 31 and a resistor 32, and a time constant switching control signal. Generating means 33. The gain control signal generation circuit 18 and the time constant switching means 30 are connected in parallel, and output a gain control signal to the audio amplification circuit 13 under the control of the time constant switching control signal generation means 33. The diode 17 detects a change in the output level of the audio amplifier circuit 13. The capacitor 20 holds the peak value of the gain control signal, and the charging resistor 19 determines the charging time of the gain control signal charged in the capacitor 20. The discharge resistor 21 and the resistor 32 of the time constant switching means 30 determine the discharge time of the gain control signal charged in the capacitor 20.
[0024]
The time constant switching control signal generating means 33 receives a signal output from the microphone plug 12 when the microphone 11 is inserted and a signal detected when the talk switch 22 is pressed, and turns on the power. To detect when it is started. When these are detected, that is, when the microphone 11 is inserted, the talk switch 22 is pressed, or the power is turned on, the time-constant switching control signal generating means 33 is connected to the transistor 31 of the time-constant switching means 30. And outputs a control signal (control pulse) having a predetermined time width for conducting the control signal.
[0025]
Next, the operation of the gain control circuit 16 will be described in detail. First, the audio signal that has passed through the microphone 11 and has been amplified by the audio amplification circuit 13 is input to the gain control signal generation circuit 18 of the gain control circuit 16. At this time, the capacitor 20 is charged via the diode 17 and the charging resistor 19. Here, the time constant determined by the charging resistor 19 and the capacitor 20 determines a time called “attack time”, and when a large-level audio signal is input, the dynamic range is cut off. It is set to a small value (for example, a few milliseconds) so that there is no possibility. On the other hand, a time constant determined by the capacitor 20 and the discharge resistor 21 of the gain control signal generation circuit 18 determines a time called a “recovery time”, and is set to a value as large as possible (in order to obtain a natural sound). For example, several tens of seconds).
[0026]
FIG. 2A is a time chart showing an example of a state when the power is turned on, a talk switch is operated, and a microphone plug is inserted and removed (High side is on (or in), Low side is off (or out)), and FIG. 5 is a time chart showing respective pulse signals output by the time constant switching control signal generating means 33 in the state of FIG. As shown in FIG. 3B, when the power is turned on and the talk switch is operated, when each of them changes from off to on, and when the microphone plug is inserted and removed, when it changes from out to in (that is, when the microphone plug is inserted), the normal Is output.
[0027]
The time constant switching control signal generating means 33 constantly monitors the state of the power supply, the talk switch 22 and the microphone plug 12, and performs a predetermined time immediately after the power is turned on, a predetermined time immediately after the talk switch 22 is turned on, and The pulse shown in FIG. 2 is output to the time constant switching means 30 at a predetermined time immediately after being inserted into the plug. At this time, the transistor 31 conducts to form a ground, the discharge resistor 21 and the resistor 32 are connected in parallel, and the value of the above-mentioned recovery time is reduced. Since the resistor 32 is set to a smaller value than the discharge resistor 21, the voltage charged in the capacitor 20 is discharged through the resistor 32 in a very short time.
[0028]
FIG. 3A is a waveform diagram showing an example of a signal waveform input to the audio amplification circuit 13 when the power is turned on, when the talk switch is operated, and when the microphone plug is inserted and removed, and FIG. FIG. 3C is a time chart showing the state of the talk switch and the microphone plug, and FIG. 4C is a time chart showing each pulse signal output by the time constant switching control signal generating means 33 in the state of FIG. 4 is a time chart showing a waveform of a gain control signal output from the gain control circuit 16. As shown in FIG. 4C, while the time constant switching control signal is being output from the time constant switching control signal generating means 33 (on period), the discharging resistor 21 of the gain control signal generating circuit 18 and the time constant The resistance 32 of the switching means 30 is connected in parallel, and the recovery time can be reduced. On the other hand, while the time constant switching control signal generation unit 33 does not output the time constant switching control signal (OFF period), the resistor 32 of the time constant switching unit 30 is disconnected from the discharge resistor 21 of the gain control signal generation circuit 18. Recovery time is increased.
[0029]
By the way, the pulse width of the time constant switching control signal output from the time constant switching control signal generating means 33 is determined by the noise immediately after power-on, immediately after the talk switch 22 is turned on, or immediately after the microphone 11 is inserted into the microphone plug 12 This is the time required for discharging the electric charge charged in the capacitor 20 by the excessive signal with the discharge characteristic obtained by the time constant switching means 30. On the other hand, it takes time for the system to stabilize due to the effect of noise at the time of power activation. Therefore, as shown in FIG. 3, the pulse width of the time constant switching control signal output when the power is turned on needs to be set longer than the pulse width output when the talk switch is turned on or the microphone is inserted into the microphone plug. .
[0030]
In addition, by generating a time constant switching control signal having a pulse width as short as possible immediately after the talk switch 22 is pressed or according to the characteristics of an excessive signal after the microphone plug is inserted, the voice input device 1 can be operated early. It can be set to a usable state.
[0031]
In this embodiment, the recovery signal time is reduced by turning on the transistor 31 of the time constant switching means 30 with a control signal from the time constant switching control signal generating means 33. However, a transistor is used. The present invention is not limited to this, and it is only necessary that the electric charge stored in the capacitor 20 can be discharged in a short time.
[0032]
As described above, according to this embodiment, the time-constant switching control signal generating means is configured to generate the time-constant switching control signal for switching the time constant at a predetermined time immediately after the power is turned on. The time required to return can be shortened, and the signal level can be stabilized while maintaining the natural state of the input audio signal.
[0033]
Further, according to this embodiment, the time constant switching control signal generation means detects that the microphone is inserted into the microphone plug, and generates a time constant switching control signal for switching the time constant at a predetermined time after the detection. With this configuration, the time until the gain returns can be shortened, and the signal level can be stabilized while maintaining the natural state of the input audio signal.
[0034]
Further, according to this embodiment, the time constant switching control signal generating means detects that the talk switch has been pressed, and generates a time constant switching control signal for switching the time constant at a predetermined time after the detection. With this configuration, the time until the gain returns can be shortened, and the signal level can be stabilized while maintaining the natural state of the input audio signal.
[0035]
【The invention's effect】
As described above, according to the present invention, when an excessive signal is input, the gain control signal generated in response to the time constant switching control signal generated by the time constant switching control signal generating means causes the gain to return until the gain returns. Can be shortened, so that the signal level can be stabilized while maintaining the natural state of the input audio signal.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a voice input device according to the present invention.
FIG. 2A is a time chart showing an example of a state when power is turned on, a talk switch is operated, and a microphone plug is inserted and removed (High side is on (or in), Low side is off (or out)); 6B is a time chart showing respective pulse signals output by the time constant switching control signal generating means in the state of FIG.
3A is a waveform diagram showing an example of a signal waveform input to the audio amplifier circuit when the power is turned on, when a talk switch is operated, and when a microphone plug is inserted and removed, and FIG. It is a time chart which shows the state of a power supply, a talk switch, and a microphone plug, (c) is a time chart which shows each pulse signal which a time constant switching control signal generation means outputs in the state of (a), (d) 3 is a time chart showing a waveform of a gain control signal output from the gain control circuit.
FIG. 4 is a block diagram showing a general configuration of a voice input device using a conventional compressor.
FIG. 5 is a signal waveform diagram showing states of a microphone input, a gain control circuit output, and an audio amplifier circuit output due to noise when a microphone plug is inserted.
[Explanation of symbols]
1, 101 Audio input device 11 Microphone 12 Microphone plug 13 Audio amplification circuit 14 A / D conversion circuit 16 Gain control circuit 17 Diode 18 Gain control signal generation circuit 19 Charge resistor 20 Capacitor 21 Discharge resistor 22 Talk switch 23 Host 30 Time constant switching Means 31 Transistor 32 Resistance 33 Time constant switching control signal generating means

Claims (5)

An audio amplifier circuit that changes the gain according to the gain control signal and amplifies the input audio signal;
A gain control circuit that generates a gain control signal from the signal level of the amplified input audio signal, and outputs the gain control signal to the audio amplification circuit;
In a voice input device comprising:
The gain control circuit,
Time constant switching control signal generating means for generating a time constant switching control signal for switching the time constant,
Time constant switching means for switching a time constant of a circuit that generates the gain control signal in accordance with the input time constant switching control signal,
A voice input device comprising: 2. The voice input device according to claim 1, wherein the time constant switching control signal generating means generates a time constant switching control signal for switching a time constant at a predetermined time immediately after power-on. 3. The time-constant switching control signal generating unit detects that a microphone has been inserted into a microphone plug, and generates a time-constant switching control signal that switches a time constant at a predetermined time after the detection. The voice input device according to claim 1. The time constant switching control signal generating means detects that the talk switch has been pressed, and generates a time constant switching control signal for switching the time constant at a predetermined time after the detection. The voice input device according to any one of the above. The pulse width of the time constant switching control signal generated when a microphone is inserted into a microphone plug or the pulse width of the time constant switching control signal generated when a talk switch is pressed is a predetermined value immediately after power-on. The voice input device according to any one of claims 1 to 4, wherein each of the pulse widths is smaller than a pulse width of the time constant switching control signal generated at time.

Images