JPS6330640B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a plosive sound extraction device for extracting plosive sounds.

Conventionally, there is no known method for reliably extracting plosive sounds from the characteristics of sounds picked up by a microphone using a normal method.

The present invention provides a plosive sound extraction device that detects not only the sound pressure information of voices accompanying speech but also oral air flow information using a microphone, and processes the signals to extract plosive sounds reliably, easily, and in real time. The purpose is to

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the plosive sound detection device of the present invention.

In the figure, reference numeral 1 denotes a microphone that detects the oral airflow and sound pressure of the voice accompanying speech, and is placed, for example, in front of the speaker's mouth, with the windshield of the condenser microphone removed, so as to receive the oral airflow.

Reference numeral 2 denotes a low-pass filter that passes only the low frequency components of the output signal of the microphone 1, and is composed of, for example, an active filter with a cutoff frequency of 200 Hz and an attenuation slope of about 12 dB/octave. 3 is a high-pass filter that passes only the high frequency components of the output signal of microphone 1; for example, the cutoff frequency
It consists of an active filter with a frequency of 300Hz and an attenuation slope of approximately 12dB/octave. Reference numeral 4 denotes a power detection circuit for detecting the power of the output signal of the high-pass filter 3, which is constructed using, for example, a square circuit or the like, but it can also be replaced with a full-wave or half-wave rectifier circuit. Reference numeral 5 denotes an integrating circuit that integrates the output signal of the power detection circuit 4, and for example, a first-order delay circuit or the like is used. A processing circuit 6 reduces the output of the low-pass filter when the output of the integrating circuit 5 becomes large, and is composed of, for example, a voltage-controlled variable gain amplification circuit or a division circuit. In the extreme, a gate circuit can also be used.

Figures 2 b to g and Figures 3 j to q show the signal waveforms of each part of the plosive extractor shown in Figure 1. Figure 2 is for plosive sounds, and Figure 3 is for voiceless fricatives. Each case is shown below. Furthermore, Fig. 2a and Fig. 3i are the waveforms of the sound pressure of the voice detected by a microphone in the usual manner when a plosive and a voiceless fricative are uttered, respectively, Fig. 2h and Fig. 3q
show the oral airflow velocity waveforms in front of the mouth when plosives and voiceless fricatives are uttered, respectively.

In FIGS. 2 and 3, a and i represent representative examples of the audio waveforms of plosives and voiceless fricatives, respectively;
b, j are the output signals of the microphone 1 in FIG. 1, c, k are the output signals of the low pass filter 2, d, l are the output signals of the high pass filter 3,
e, m are the output signals of the power detection circuit 4, f, n
is the output signal of the integrating circuit 5, and g and p are the processing circuit 6.
represent the output signals of, respectively.

In the above configuration, plosive sounds, for example,
When a sound |pu| containing |p| is uttered, the microphone 1 placed in front of the mouth outputs a waveform as shown in FIG. 2b. This output waveform is a superposition of the output generated by the deflection of the microphone's diaphragm due to the oral air flow h and the output generated by the sound pressure a based on the air vibration at the time of bursting.

Here, the main component of the output due to oral airflow is generally
Experiments have confirmed that this has a lower frequency than the main component of the output due to air vibration sound. Therefore, the waveform c that is the output of the low-pass filter 2 is
It can be considered that it is based on the oral air flow during plosive speech.

In general, the oral airflow velocity during plosive utterances exhibits a large pulse shape with a sharp rise compared to the slow, small oral airflow during non-plosive utterances. Correspondingly, the output of the low-pass filter 2 also typically The waveform becomes a pulse like waveform c, and is clearly distinguished from the output of the low-pass filter 2 which hardly occurs during normal non-plosive speech.

However, voiceless fricatives among non-plosives,
In particular, when |Φ| is uttered, a fast and perturbing oral airflow is generated, making it difficult to distinguish the output waveform of the low-pass filter 2 from that of a plosive. In other words, Fig. 3 shows the case where an unvoiced fricative, especially |Φu|, is uttered, and the output of the low-pass filter 2 reaches a considerable size as shown in the waveform k in response to the oral airflow velocity waveform q, resulting in a plosive. It becomes difficult to distinguish the plosive sound from the waveform c in FIG.

On the other hand, as a result of experiments, the following points became clear. That is, low-pass filter 2 in the case of plosives
The output of is generated immediately after the rupture begins, whereas the low-pass filter output for voiceless fricatives, especially |Φ|, occurs much later than the beginning of the fricative.
Most of the fricatives occur near the center of the fricative section, and in the case of plosives, the output of the high-pass filter 3 occurs in a short period after the plosive point, as shown in Figure 2d, but on the other hand, voiceless fricatives, especially |Φ In the case of ｜, the friction occurs over the entire range as shown in Figure 3, and this component is not only due to the sound pressure of the voice due to friction, but also due to the deflection of the microphone's vibrating membrane by the oral airflow. It grows and reaches a considerable size.

Therefore, the outputs of the power detection circuit 4 when a plosive sound and a voiceless fricative sound are uttered are as shown in FIG. 2e, respectively.
The outputs of the integrating circuit 5 are as shown in FIG. 3 m, and the outputs of the integrating circuit 5 are as shown in FIG. 2 f and FIG. 3 n, respectively. That is, the output of the integrating circuit 5 at the time when a large output is produced in the low-pass filter 2 is much larger in the case of a voiceless fricative than in the case of a plosive. Therefore, by using the output of the integrating circuit 5, the output of the low-pass filter is processed by the processing circuit 6, so that when the output of the integrating circuit 5 is large, the output of the low-pass filter 2 is small. For fricative sounds, especially |Φ|, the outputs of the processing circuit 6 are as shown in FIG. 2g and FIG. 3p, respectively. As is clear from the figure, a much larger pulse output is obtained from the processing circuit 6 in the case of plosives than in the case of voiceless fricatives, and plosives can be accurately distinguished from voiceless fricatives in real time. An extraction device can be realized.

As is clear from the above description, the plosive sound detection device according to the present invention integrates the power of relatively high frequency components in the microphone output, controls the relatively low frequency components in the microphone output by the integrated output, This controlled low-frequency component is used to extract plosive sounds, making it possible to reliably extract plosive sounds.

Further, according to the present invention, it is only necessary to use one microphone as a sensor and the circuit is simple, so that the apparatus can be easily constructed and plosive sound extraction can be performed in real time.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of a plosive sound detection device according to the present invention, and FIGS. 2 and 3 are waveform diagrams of various parts in FIG. 1. 1...Microphone, 2...Low pass filter, 3
...High pass filter, 4...Power detector, 5...Integrator circuit, 6...Processing circuit.

Claims (1)

[Claims] 1. A microphone arranged in front of the mouth to receive oral airflow, a low-pass filter that passes only the low-frequency components of the output signal of the microphone, and a high-pass filter that passes only the high-frequency components of the output signal of the microphone; a power detection circuit that detects the power of the output signal of the high-pass filter; an integration circuit that integrates the output signal of the power detection circuit; and processing that reduces the output of the low-pass filter when the output of the integration circuit becomes large. What is claimed is: 1. A plosive sound extraction device, comprising: a plosive sound extraction device comprising: a plosive sound extraction device;