JPS636600A - Nasal detector - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a nasal sound detection device for detecting nasal sound, which is the sound emitted from the nostrils when airflow flows through the nasal passages.

[Conventional technology]

In recent years, various training devices have been developed to train people with language disabilities, especially hearing-impaired people, to speak and speak. by the way,
When vocalizing, sounds emitted from the mouth and sounds emitted from the nostrils are emitted. However, hearing-impaired people tend to nasalize vowels when they pronounce them. Nasalization significantly impairs vowel pronunciation, making the sound extremely difficult to hear.

Japanese does not have nasal vowels (nasalized vowels), so when training speech-impaired people to learn vowels, it is necessary to clearly distinguish between ``nasals'' and ``vowels.'' For this reason, it is necessary for vocalization and speech training devices to visually display nasal sounds, especially nasalized vowels, but the conventional vocalization and speech training devices described above have the following features: (1) nasal sounds; Detection using software using spectral characteristics.

(2) Something that extracts and detects the components of nasal sounds emitted from the nasal cavity in some way.

etc.

[Problem that the invention seeks to solve]

By the way, in the case of (1) mentioned above, the detection efficiency is not very high, and if you try to increase the accuracy, the amount of calculation increases rapidly, causing the problem that it is not suitable for real-time display. In the case of 2), it is necessary to wear a small vibrometer on the nose to detect the vibration of the nose that accompanies the nasal sound. Therefore, wearing a small vibrometer is cumbersome and undesirable from a training point of view, and it also causes problems with the circuit. Problems arose, such as complication.

[Means for solving problems]

The present invention solves the above-mentioned problems and provides a nasal sound detection device that has a simple configuration and facilitates vocalization and speech training that eliminates nasal sounds. A microphone that makes a sound, a means for calculating the ratio of the sound volume to the nasal sound volume from the signals from the two microphones, and a display that displays a cursor length corresponding to the ratio obtained by the means. This is done by a nasal sound detection device comprising:

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to the drawings.
In the figure, 1 is a headband worn on the head of the trainee, 2 is a mounting plate fixed to the headband 1 so that it can be attached freely, and 3 is a sound source from the mouth fixed below the mounting plate 2. 4 is an arithmetic microphone fixed to the lower bent portion 2a of the mounting plate 2 and collects sounds from the nose; 5 is a trumpet-shaped microphone placed over the arithmetic microphone 4; It is a sound collector.

When the headband 1 described above is worn on the trainee's head, the everyday microphone 3 is located at the trainee's mouth, and
A sound collector 5 is positioned under the nostril of the trainee as shown in FIG. Therefore, the daily use microphone 3 detects sounds coming from the trainee's mouth, and the calculation microphone 4 detects nasal sounds via the sound collector 5.

Next, a device for detecting nasal sounds using the two microphones 3 and 4 will be explained with reference to FIG. In addition, the third
In the figure, 3 and 4 are the aforementioned microphones.

6.7 is an amplifier that amplifies the signals from the respective microphones 3 and 4; 8 is a rectifier that rectifies the amplified signal from the amplifier 6;
Reference numeral 9 designates a low-pass filter that passes only low-frequency components from the signal from the rectifier 8. Since energy is concentrated in the low-frequency range in the spectrum of nasal sounds, only the nasal components are passed through the low-pass filter 9. Extract. 1O is an analog-to-digital converter that digitally converts the signal passed through the low-pass filter 9 into 8-bit nasal sound data.

11 is a detector that extracts the audio waveform and amplitude obtained through the microphone 3 and the amplifier 7; 12 is the detector 1;
This is an analog-to-digital converter that digitally converts 1-bit audio data into 12-bit audio data. Reference numeral 13 denotes a control circuit including an arithmetic unit for obtaining the ratio between the nasal sound data and voice data, that is, nasal sound data/voice data (hereinafter referred to as B/V).

14 is an output circuit for displaying the output signal from the control circuit 13 on a display 15 such as a CRT.

Next, a training method for removing nasal sounds from vocalization based on the above circuit will be explained with reference to the flowchart of FIG.

Now, when the subject utters a voice, the voice from the mouth and the nasal sound from the nose are detected by microphones 3 and 4, and amplifiers 6 and 7 are detected, respectively.
is amplified to the level set in . And amplifier 6
The output from the microphone 4 is waveform-shaped by a rectifier 8, and a low-pass filter 9 removes unnecessary information from the information taken in from the microphone 4.

Next, the output from the low-pass filter 9 is converted into 8-bit nasal sound data by the A/D converter IO, and the data is sent to the control circuit 13.
Enter.

On the other hand, the amplitude of the output waveform obtained through the microphone 3 and the amplifier 7 is converted into audio data of 12 beep/2 by the A/D device 12 and input to the control circuit 13.

When the two data mentioned above are input to the control circuit 13,
The 118 circuit 13 performs control as shown in the flowchart of FIG. That is, in step a), set the number of frames N according to the preset training time, then, for example, extract the formants when 1 frame = 25.6 tasec = 256 samples (sampled at 10 KHz), and calculate the magnitude of the nasal sound. B, calculate B/V at the sound volume V (step b). Then, it is determined whether or not the first formant frequency F1 and the second formant frequency F2, which are the results of the calculation, are within the displayable range. If not, it is waited until a displayable formant frequency arrives, and the display is possible. If so, proceed to the next step (step c).e When a displayable formant frequency arrives, B/V, which is a number representing the degree of nasalization, is calculated (step d).

Furthermore, the length of the cursor is L=4XINT (B/V)+4 (step e).

The result of this calculation is displayed with a cursor on the display 15. At this point, it is determined whether or not the cursor protrudes from the upper and lower boundaries of the display 15. If it does, the display address is set so that the cursor is on the border. (step g), and if it does not protrude, proceed to the next step (step f).0 Next, determine whether only one cursor is displayed, and if one or more cursors If the cursor is displayed, erase the previous cursor (step i); if only one is displayed, proceed to the next step (step h). Then, the first formant F1 and the second formant F2 are displayed on the display 15.
(step j). Next, the number of frames N according to the training time is calculated as N=N-1 (step k
), the training continues until N becomes O, and ends when it becomes 0 (step Il).

To explain the specific training method, as shown in FIG. When the subject pronounces "A", the length of the "A" sound pronounced by the subject is displayed on the display 15 as the length of the cursor 16. The pronunciation is closer to the sounds of "tech" and "o" than the standard, and the amount of nasal sounds is proportional to the length of the cursor 16, so there is a considerable amount of nasal sound. This means that it has become Therefore, the test subject practices speaking while looking at the display 15 so that the cursor 16 overlaps with the letter "A" on the display 15, and also so that the length of the cursor 16 becomes shorter.

〔Effect of the invention〕

As described above, the present invention collects sounds from the mouth and sounds from the nose using microphones, and expresses the ratio of the sounds from the mouth to the nasal sounds on a display with different cursor lengths. Furthermore, since the state of pronunciation is also displayed on the display, it is possible to practice pronunciation and to practice pronunciation that does not include nasal sounds.

[Brief explanation of drawings]

The figures show an example of the position of the nasal sound detection device according to the present invention, FIG. 1 is a perspective view of the device for collecting sounds from the mouth and nasal sounds, and FIG. 2 is a part of the device for collecting the same nasal sounds. front view, 3rd
FIG. 4 is a block diagram, FIG. 4 is a flowchart, and FIG. 5 is a front view showing the display state of the display. 3.4...Microphone, 15...Display, 16.
··cursor. Figure 1 2nd cause

Claims (1)

[Claims] a microphone that collects sounds from the mouth; a microphone that collects nasal sounds; a means for calculating the ratio of the sound volume to the nasal sound volume from signals from the two microphones; A nasal sound detection device comprising: a display that displays the length of a cursor corresponding to the ratio.