Electric artificial larynx

Abstract

To provide an electric artificial larynx that can be worn around a neck to reproduce a voice that is hands-free and close to a specific flesh voice of a user, etc.SOLUTION: In an electric artificial larynx using LPC residual waves, a circuit 20 uses two different oscillator (narrow-range oscillator 11 and wide-range oscillator 12) to generate speech that is hands-free and similar to natural voice.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an electrolarynx. [Background technology]

[0002] Healthy people produce sounds by vibrating the vocal cords when their exhaled air passes through the larynx (glottis), and produce various speech sounds by changing the frequency components of the sounds with articulatory organs such as the tongue, jaw, and lips. The sounds produced by the vibration of the vocal cords are called "laryngeal primary sounds" because they are the basis of speech.

[0003] In contrast, laryngectomees who have had their larynx removed are normally unable to speak at will, but if the articulatory organs remain, they can produce artificially created sounds in the mouth in place of laryngeal primary sounds, or by sending sounds into the mouth, they can speak, albeit imperfectly.

[0004] One of the devices that artificially creates sounds to replace laryngeal primary sounds is the electrolarynx. The electrolarynx mechanically or electromechanically generates sounds to replace laryngeal primary sounds, and guides the sounds into the oral cavity through neck vibrations, etc., to assist the laryngectomee in speaking. Some electrolarynxes of this type are equipped with a diaphragm and a voice coil motor that vibrates the diaphragm to generate sounds in the oral cavity or guides the sounds into the oral cavity, and change the generated sounds by changing the period and strength of the vibration of the diaphragm.

[0005] For example, the electric artificial larynx disclosed in Patent Document 1 has a memory unit that stores laryngeal original sound data calculated by linear predictive analysis from specific real voice data, a control unit that controls the period and / or peak value of the laryngeal original sound data in response to the user's operation or biometric information, and a sound output unit that outputs sound based on the sound data controlled by the control unit, thereby providing an electric artificial larynx that can reproduce a voice that is close to the specific real voice of a user, etc.

[0006] Furthermore, in the electric artificial larynx disclosed in Patent Document 2, it is possible to provide a mechanism that can improve the naturalness of the voice or impart individuality by driving the voice in response to a series of pulse signals obtained by applying small changes in the period and / or duty ratio to the waveform of the pulse signal that is the basis for generating sound.

[0007] Furthermore, in the electric artificial larynx disclosed in Non-Patent Document 1, a vibrator separated from the electric artificial larynx main body is fixed to the underside of the jaw with a mounting device and can be operated with a remote-controlled finger switch, armpit switch, or foot switch, making it possible to provide a remote-controlled hands-free artificial larynx mechanism that eliminates physical constraints during conversation. [Prior art documents]

[0008] [Patent Document 1] Patent No. 6403448 [Patent Document 2] Patent No. 4940408 [Non-Patent Document 1] Densei Co., Ltd. Commercialization of a hands-free artificial larynx operated by a remote control. Project to promote the development of devices to support the independence of people with disabilities, FY2011, Report Summary of the Invention [Problem to be solved by the invention]

[0009] The voice of a healthy person is composed of a periodic component that controls the fundamental frequency and aperiodic components that control the rest. On the other hand, the electrolarynx disclosed in Patent Document 1 only takes into account the periodic component, and although it improves the clarity of the voice, it remains a mechanical voice output.

[0010] Furthermore, in the electric artificial larynx disclosed in Patent Document 2, the period and duty ratio of the pulse signal are obtained based on the sound waveform of the voice recorded from a healthy subject or a patient before laryngectomy, etc., so that it is possible to give minute changes that are physiologically acceptable to the listener and to reproduce a sound that is closer to the sound produced by the human vocal cords.

[0011] However, even the electro-artificial larynx disclosed in Patent Document 2 does not take into account the effects of resonance in the vocal tract on the output sound, and therefore has the problem of being unable to reproduce a voice that is close to the specific natural voice of a user, etc.

[0012] Furthermore, both Patent Document 1 and Patent Document 2 have the problem that the device must be held in the hand and pressed against the throat when in use.

[0013] Although Non-Patent Document 1 provides a hands-free type, there is a problem in that the vibration sound is not processed to reproduce a specific voice close to a human voice.

[0014] The vibrators used in Patent Documents 1 and 2 and Non-Patent Document 1 have a limited resonant frequency between 60 Hz and 200 Hz, and therefore have the problem that they cannot generate high-frequency vibration sounds suitable for women and children.

[0015] The present invention has been devised against this background, and has an object to provide a hands-free electro-artificial larynx that can reproduce a voice that is close to the natural voice of a specific person, such as a user. [Means for solving the problem]

[0016] The electroartificial larynx of the present invention has two transducers, one for wide range and one for narrow range, and creates vibration sounds using an algorithm to make them sound more similar to the human voice, and is operated using a voice storage device such as a smartphone.

[0017] Narrow-range vibrators are the type used in conventional EL devices, and their volume is stable in the range of 60Hz to 200Hz. The vibrator generates vibrations by moving its axis up and down due to the Lorentz force generated by the coil and permanent magnet. The principle is that when a current is passed through the voice coil and a magnetic circuit generates a magnetic flux, a thrust is generated according to Fleming's left-hand rule, which pushes the axis up and strikes the diaphragm. When the current is turned off, a reaction force is generated and the axis moves away from the diaphragm. Repeating this action generates the vibration sound.

[0018] The wide-range vibrator is manufactured by Foster Electric Co., Ltd. and has the same structure as JP 2020-199506. The principle of generating vibration sound is the same as the narrow-range vibrator, but it is designed to stably output a wide frequency band from 55 Hz to 1000 Hz. However, the output volume is smaller than that of the narrow-range vibrator.

[0019] The algorithm for generating vibration sounds is such that in the frequency band between 60Hz and 200Hz, the narrow-band oscillator outputs periodic components of vibration sounds, and the wide-band oscillator outputs non-periodic components, with the fundamental frequency of a specific sound being the boundary at 200Hz, which is the resonant frequency band of the narrow-band oscillator. In the range above 200Hz, the narrow-band oscillator does not operate, and the wide-band oscillator outputs periodic components.

[0020] The above-mentioned periodic components refer to vibration sounds that are the amplified fundamental frequency and harmonic components of the LPC residual wave of specific audio data.

[0021] The non-periodic components mentioned above refer to vibration sounds in which only the non-periodic indexes are extracted from specific audio data using software called WORLD. Effect of the Invention

[0022] According to the present invention, it is possible to provide an electro-artificial larynx that can be worn around the neck, providing a hands-free experience and reproducing a voice that is close to the natural voice of a particular user, etc. [Brief description of the drawings]

[0023] [Figure 1] This figure explains a system including an electric artificial larynx attachment unit 10 according to the first embodiment of the present invention, an electric artificial larynx circuit unit 20 that provides vibration sound to the electric artificial larynx attachment unit 10, a vibration sound storage device 30 that stores the vibration sound for providing the vibration sound to the electric artificial larynx circuit unit 20, and an original sound generation device 40 that creates periodic and non-periodic components of the vibration sound that is the source of the voice obtained by performing a predetermined process such as linear predictive analysis on the specific human voice data of a user, etc., when the fundamental frequency of the specific human voice data of a user, etc. is in the range of 60 Hz to 200 Hz. [Diagram 2] This diagram illustrates a system including an electric artificial larynx attachment unit 10 according to the first embodiment of the present invention, an electric artificial larynx circuit unit 20 that provides vibration sound to the electric artificial larynx attachment unit 10, a vibration sound storage device 30 that stores the vibration sound for providing the vibration sound to the electric artificial larynx circuit unit 20, and an original sound generation device 40 that performs predetermined processing such as linear predictive analysis on the specific human voice data of a user, etc., to create periodic components of the vibration sound that is the source of the obtained voice, and provides the periodic components to the vibration sound storage device 30, when the fundamental frequency of specific human voice data of a user, etc. is greater than 200 Hz. [Diagram 3] FIG. 2 is a front view of the configuration of the attachment part 10 of the electric artificial larynx when attached. [Figure 4] FIG. 2 is a diagram showing the configuration of the attachment part 10 of the electric artificial larynx as viewed from behind when attached. [Diagram 5] 1 shows the internal structure of a narrow-area transducer 11. [Figure 6] This is a diagram of the frequency spectrum waveform of a naturally uttered voice from 700 Hz to 2000 Hz. [Figure 7] This is a diagram of the frequency spectrum waveform from 700 Hz to 2000 Hz of the voice waveform produced by a conventional electrolarynx. [Figure 8] FIG. 13 is a diagram showing the frequency spectrum waveform from 700 Hz to 2000 Hz of a voice waveform produced by the electrolarynx of the present invention using a vibration sound generated by the generation algorithm of the present invention. [Figure 9]FIG. 1 is a diagram showing a frequency spectrum waveform from 700 Hz to 2000 Hz of a voice waveform produced by the electrolarynx according to the present invention using a vibration sound of a simple LPC residual wave. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0024] Hereinafter, an embodiment of the present invention will be described. Note that the present invention is not limited to the following embodiment.

[0025] (First embodiment) FIG. 1 shows a system including an attachment unit 10 of an electric artificial larynx, a circuit unit 20 of the electric artificial larynx that provides vibration sound to the attachment unit 10 of the electric artificial larynx, a vibration sound storage device 30 that stores the vibration sound to be provided to the circuit unit 20 of the electric artificial larynx, and an original sound generation device 40 that performs predetermined processing such as linear predictive analysis on specific real voice data of a user, etc., to create periodic and non-periodic components of the vibration sound that is the source of the voice obtained, and provides these components to the vibration sound storage device 30. The original sound generation device 40 is used when generating a male voice with a fundamental frequency of 60 Hz to 200 Hz.

[0026] Second embodiment FIG. 2 shows a system including an attachment unit 10 of the electric artificial larynx, a circuit unit 20 of the electric artificial larynx that provides vibration sound to the attachment unit 10 of the electric artificial larynx, a vibration sound preservation device 30 that stores the vibration sound to be provided to the circuit unit 20 of the electric artificial larynx, and an original sound generation device 40 that performs predetermined processing such as linear predictive analysis on specific real voice data of a user, etc., to create periodic components of the vibration sound that is the source of the obtained voice, and provides the periodic components of the vibration sound to the vibration sound preservation device 30, and is used when generating a female voice with a fundamental frequency greater than 200 Hz.

[0027] 3 is a front view of the wearing part when worn, and FIG 4 is a rear view of the wearing part when worn. It is composed of a narrow-area transducer 11, a wide-area transducer 12, a band 13, and a buckle 14.

[0028] (Principle of narrow band transducer) The narrow-band vibrator in Figure 5 is used in the EL "Yourtone" by Densei Co., Ltd. When current is passed through the voice coil and a magnetic flux is generated by the magnetic circuit, a thrust force is generated according to Fleming's left-hand rule, which pushes the shaft up and strikes it against the diaphragm. When the current is turned off, a reaction force is generated and the shaft moves away from the diaphragm. By repeating this process, vibration sound is generated.

[0029] For the principles and structure of the wide-area transducer, please refer to JP 2020-199506. EXAMPLES

[0030] An embodiment of the present invention will be described below. In this embodiment, "Hello" was recorded using natural speech, a conventional electrolarynx, and the electrolarynx of the present invention. A frequency spectrum was obtained by applying a Fourier transform to the voice extracted for 0.1 seconds from the start of the phoneme of the vowel "a" in "ha" of "hello." The recording environment used was a personal computer (Panasonic, Let's Note, CF-5V), a microphone (Audio Technica, ATR1100x), a microphone terminal USB conversion cable (Planex, PL-US35AP), and Praat software.

[0031] Figures 6 to 8 show frequency spectra in the range of 700 Hz to 2000 Hz for natural speech, a conventional electrolarynx, and the electrolarynx of the present invention. The horizontal axis indicates frequency, and the vertical axis indicates sound pressure level. Since the first formant of the vowel "a" is approximately 780 Hz and the second formant is approximately 1240 Hz, the spectrum was plotted in the range of 700 Hz to 2000 Hz. The characteristics of the first and second formants that were not seen in the speech produced by the conventional electrolarynx were seen in the speech produced by the electrolarynx of the present invention. In other words, it can be seen that the speech produced by the electrolarynx of the present invention is closer to natural speech.

[0032] Figure 9 shows a voice waveform in which a simple LPC residual wave is used as a vibration sound in the electrolarynx according to the present invention. It can be seen that the first formant is particularly clear in the vibration sound created by the generation algorithm according to the present invention shown in Figure 8. From this, it can be seen that the vibration sound created with the characteristics of "a" is more clearly seen than the LPC residual wave, and it can be said that using an LPC residual wave with the fundamental frequency and its overtones raised by ±10 Hz by 12 dB as a vibration sound is novel and effective. [Explanation of symbols]

[0033] 10...Electronic artificial larynx attachment part 11...narrow-range vibrator, 11A...diaphragm, 11B...shaft, 11C...rubber membrane, 11D...voice coil, 11E...permanent magnet, 11F...magnetic circuit 12...Wide-range transducer 13. Band 14…Buckle 20... Electric artificial larynx circuit section, 21... Bluetooth audio amplifier, 22... Charging / boosting board, 23... Lithium ion polymer battery, 24... Periodic component of vibration sound, 25... Non-periodic component of vibration sound 30…Vibration sound storage device 40... Vibration sound generating device, 41... Periodic component of vibration sound, 42... Non-periodic component of vibration sound, 43... Recording medium, 43A... Recorded voice data, 44... Recording microphone

Claims

1. 1. An electric artificial larynx comprising a system including a mounting part having two different transducers, a circuit part for providing vibration sound to the mounting part, a vibration sound storage device for storing the vibration sound to provide the vibration sound to the circuit part, and an original sound generating device for creating vibration sound and providing it to the vibration sound storage device, wherein one of the transducers is a narrow-band transducer that outputs periodic components of the vibration sound in a frequency band of 60 Hz to 200 Hz, and the other is a wide-band transducer that outputs non-periodic components of the vibration sound in the range of 60 Hz to 200 Hz and periodic components of the vibration sound in a range higher than 200 Hz.

2. An electric artificial larynx as described in claim 1, characterized in that two vibrators are arranged in parallel on a band and wrapped around the throat, thereby inputting vibrations for voice in a hands-free manner.

3. An electroartificial larynx as described in claim 1 or claim 2, characterized in that an output close to a human voice is obtained by using voice synthesis technology based on a pre-recorded human voice.

4. An electric artificial larynx as described in any one of claims 1 to 3, characterized in that the voice synthesis technology uses an algorithm that generates periodic and non-periodic components of vibration sound.

Images