Method and device for supporting optimized acoustic adjustment of the human vocal tract

Abstract

Method for supporting the adjustment of acoustic properties of hollow bodies, in particular a vocal tract and thus the voice quality, wherein an acoustic wave field consisting of several frequencies or frequency bands is generated by at least one loudspeaker and / or by at least one waveguide connected to a loudspeaker at or in the mouth, wherein the distance of the sound outlet to the mouth is less than 0.8 * radius of the respective mouth opening, wherein frequencies in the range of 200 to 10,000 Hz are used, wherein the wave field is amplified or attenuated by adjusting the geometry of a user's vocal tract, wherein, for adjusting the geometry of a user's vocal tract, at least three adaptation frequencies, which correspond to at least three addressed vocal tract resonances and are immediately fed back to the user during use, are amplified or attenuated.wherein the resulting wave field is measured via at least one external microphone placed outside the mouth or waveguide and acoustically isolated from the sound source, wherein the distance from the microphone to the mouth is greater than 0.8 * radius of the mouth opening.

Description

1.1 Information on the technical field:

[0001] The invention relates to a training method and a training apparatus that allows the human vocal tract to be adjusted to an externally generated acoustic wave field and simultaneously measures the acoustic properties of the vocal tract. Furthermore, the invention can also be used to measure and adjust the acoustic properties of technical tubes or hollow bodies. 1.2 State of the art:

[0002] The impedance at the opening of hollow bodies, tubes, or tube systems is an important characterization parameter that allows conclusions to be drawn about the distribution of acoustic pressure conditions inside the tube or hollow body. While it is often possible to measure through the tubes to determine the acoustic transfer function, this is more difficult with tube systems open on one side and not possible in the case of the vocal tract. The following section will discuss the specific state of the art measurement technology for determining the acoustic properties of the human vocal cycle. Analysis methods for characterizing the human voice take up a large part of the technical literature. For example, US 4,862,503 A by M. Rothenberg from 1988 describes a device for analyzing the human voice. The focus here is not on improved adjustment of the vocal folds and vocal tract during use of the device.In US Pat. No. 8,641,637 B2, "Method of measuring an acoustic impedance of a respiratory system and diagnosing a respiratory disease or disorder or monitoring treatment of the same," acoustic impedance spectroscopy is used to characterize the respiratory system. Low frequencies are predominantly used. DE 101 31 823 A1 by Malte Kob, "Miniature arrangement for measuring acoustic impedance," describes an apparatus for characterizing the acoustic properties of the human vocal tract. In general, acoustic impedance spectroscopy aims to capture a wide wavelength range.

[0003] In US Pat. No. 4,326,416, the vocal tract is stimulated broadband via a spark gap within a tube. A microphone is installed inside the tube, and the response of a transiently emitted sound signal is measured. In this application, a direct connection between the mouth and the incoming tube is required to generate absolute measurement values. Furthermore, the phase of the signal flowing into the mouth varies randomly depending on the chosen excitation mode.

[0004] It is also known that low-frequency waves introduced via a loudspeaker interact with the vocal folds (Rothenberg et al.). Furthermore, introduced waves across a broad frequency spectrum are used to measure the acoustic impedance of the vocal tract (Rothenberg, Hanna et al.). In this method, different frequencies are introduced simultaneously or sequentially at or into the mouth via a so-called impedance tube. This tube is cylindrical in design to allow for the application of easily interpreted acoustic measurement techniques. Two or more microphones, or a microphone and a device for measuring the sound velocity, are attached to this tube (Hanna, Kob). The transfer function can be calculated from the intensity and phase of the measurement signal. This transfer function, together with a calibration of the entire system, enables a measurement of the acoustic impedance of the vocal tract.The method has also been described in a feedback mechanism for learning foreign language pronunciation while simultaneously using the voice. A measurement during saxophone playing was also developed by the same research group (Chen et al.). The acoustic impedance measurement is sensitive to disturbances, and the setup requires complex calibration and high-quality measurement microphones.

[0005] US Patent No. 8,641,637 B2 describes a device in which low-frequency sound signals are delivered to the mouth via a tube. The use of low frequencies requires large loudspeakers. The microphones are mounted within the waveguide, which is also used to introduce the signal.

[0006] Acoustic impedance was determined during voice use by introducing a noise signal and recording it with a microphone directly at the point of introduction (Vos et al. 2016). The excitation signal is modulated in such a way that a signal originally measured with the mouth closed can be reproduced. However, this does not allow for direct acoustic feedback for the user. In addition, direct excitation of the vocal tract is made more difficult by the acoustic filter used within the excitation waveguide, as an unobstructed flow of the acoustic signal into the mouth is only possible in an attenuated manner. Although the measurement is simpler than with an acoustic impedance tube, it is still susceptible to interference, so that only a select number of measurements can be evaluated (Vos et al.).

[0007] In DE 101 31 823 A1, in addition to a microphone for measuring sound pressure, a device for measuring the sound velocity at the mouth and in the immediate vicinity of the opening of a waveguide is proposed. Furthermore, a sine sweep is used for excitation. The defined distance from the mouth to the microphone is not mentioned. The use of adaptation signals for adjusting the human vocal tract is not mentioned, so the relative phase position of the adaptation signals cannot be described. A sweep signal is used to measure the acoustic properties of the vocal tract. However, this signal has the disadvantage that its onset is abrupt, and test subjects are easily affected by startle changes in their vocal tract settings. This change can occur in the range of 10-100 ms and is not visible with the sweep method, but it does alter the result. This means that additional and unnoticed errors can occur during the measurement.

[0008] The citation also refers to a small room in which an acoustic system is excited and a simultaneous measurement of pressure and velocity is performed. The sensors are positioned at the end of a tube with an exponentially decreasing cross-section, meaning the microphone is mounted at the end of the sound source; there is no spatial separation between the exit of the excitation signal and the microphone. Therefore, a simultaneous measurement of the acoustic impedance and the voice signal is fundamentally limited.

[0009] In addition, the two measuring devices are at risk of contamination directly at the sound outlet / mouth during singing, which can reduce measurement accuracy if particles or aerosols are transferred to the measuring devices. The use of a microphone or sound velocity sensor within the waveguide used for excitation does not allow for the measurement of the speech signal or sound signal of the user; this requires positioning outside the waveguide or the oral cavity.

[0010] In the journal PLOS One, Vol. 9, July 2014, B. Delvaux and D. Howard describe a system for measuring the impedance of the vocal tract using plastic models in the article A new method to explore the spectral impact of the piriform fossae on the singing voice: Benchmarking using MRI based 3D printed vocal tracts. The system uses a sweep signal (“exponential sine sweep”) that sweeps through the frequency range to be measured. In addition, the microphone used is located at the location of the glottis of a plastic model. This design is not suitable for use on living people. An application to humans using this measuring principle is therefore not considered. In addition, the use of the measuring procedure presented assumes that the opening of the glottis during singing fundamentally makes no contribution to the position of the vocal tract resonances, since the models only extend as far as the glottis and the sound pressure is measured directly at the glottis.

[0011] The same authors used the method with a modified setup to measure vocal tract resonances during voice production (9th International Workshop in Florence, September 2-4, 2015). This method uses a sine sweep signal. The microphone is positioned directly at the waveguide output, meaning the distance from the microphone to the waveguide opening is identical. As described above, using a sine sweep has the disadvantage that its onset is abrupt and can lead to changes in vocal tract geometry. Its use for simultaneous adaptation of the oral cavity is not described, nor is it possible with the device.

[0012] US 2012 / 0136660 A1 describes an apparatus for estimating resonance frequencies from changes in vocal tract geometry to generate artificial speech signals. The aim is to measure vocal tract resonances, particularly in noisy environments, and to use them to transmit a speech signal via mobile phones more intelligibly than before. For this purpose, very quiet signals of a continuous noise signal from a loudspeaker are used, which are provided between 15 Hz and 20 kHz. The microphone is in Fig. 1 is placed at the level of acoustic excitation. There is no information about the geometry of the microphone and loudspeaker relative to the mouth. The acoustic excitation and measurement are combined as a single unit.

[0013] Influencing the vocal tract geometry is enabled indirectly through training in using the system and is aimed at speech recognition without phonation. Training is mentioned that relies on feedback from the measured values ​​for optimized speech recognition. The use of specific acoustically provided adaptation signals to adjust the human vocal tract is not mentioned, nor is the relative phase position of adaptation signals described. Thus, the user does not receive immediate feedback on changes in resonances or the amplification of adaptation signals simultaneously during use.

[0014] In pop music, the so-called "mouth-tube effect" is used. Acoustic signals—for example, from an electric guitar—are channeled into the mouth through a tube, and the timbre is varied by altering the geometries of the vocal tract. This approach aims for an artistic effect rather than for measuring and improving the acoustic properties of hollow bodies such as the vocal tract. 1.3 Criticism

[0015] There is no device or method with which the acoustic properties of the vocal tract, for example, can be optimized for acoustic resistance or a specific resonance frequency without using the voice, i.e. without active vibration of the vocal cords. In addition, measuring the acoustic properties of the vocal tract using impedance spectroscopy is very complex. Two usually expensive acoustic measuring instruments are required, which operate with an apparatus including complex calibration. In addition, impedance spectroscopy provides feedback with a slight time delay. Impedance spectroscopy does not provide any direct indication of how the setting of the vocal tract should be changed. Simple and effective biofeedback for the setting of the vocal tract including, but also independently of, phonation of vowels is not yet possible, or only with great equipment complexity.

[0016] In methods and devices for measuring the impedance of the vocal tract, the sound source is guided to the mouth via a tube. In addition, one or two microphones are usually placed inside the tube (impedance tube) or attached directly to the tube's output (Kob), in which case a sound velocity sensor is also included. This significantly limits usability, particularly since the signal measured at the location of the microphone does not correspond to the signal heard by the user. Furthermore, as with the use of microphones within a waveguide used for excitation, additional microphones are required if a vocal signal is to be recorded with high quality before or after the measurement.

[0017] In addition, the known methods for a quantitative measurement of the acoustic impedance require a constant distance between the opening of the cavity and the sound output opening of a loudspeaker or waveguide.

[0018] The present invention enables targeted adjustment of the vocal tract via biofeedback for efficient transmission of acoustic signals from vocal fold vibration, as well as for other applications where the shape of the vocal tract plays a role. This requires, in particular, precise adjustment of the phase position through the electrical control of the loudspeakers in order to adapt the phase position of the input signal to the natural resonance conditions within the vocal tract. 1.4 Description of the invention

[0019] The invention relates to a training method and a measuring apparatus in which the acoustic properties of a human vocal tract are measured and / or adjusted to specified frequencies or frequency bands. The following will specifically address the situation for measuring the vocal tract; the conditions also apply mutatis mutandis to the measurement and / or adjustment of other hollow bodies. In these cases, the term "mouth" in the description is to be considered a synonym for the opening of hollow bodies whose geometry can be flexibly adjusted.

[0020] The vocal tract has a highly flexible structure. Voice production originates from the vocal folds. The resulting signal enters the vocal tract. The resonances of the vocal tract modify the signal originating from the vocal folds, acting as an acoustic filter. The vocal tract is often described in the literature as a one-sided tube with a variable geometry. If a resonance is induced by the vocal folds, resonances are also induced in the vocal tract. The resonance condition is primarily determined at the exit of the vocal tract, i.e. at the mouth, by the transition of the sound waves into the outside space. Inside the vocal tract, the resonance conditions can be adjusted very flexibly. A high intensity of so-called"Standing" waves from the frequency spectrum of the frequencies provided by the voice occur when reflections at the mouth and within the vocal tract occur in a defined manner. All interference factors that prevent the free formation of standing waves can be described as loss factors or attenuation factors. Reducing these attenuation factors or loss factors facilitates vocal production and can make the vocal sound more pleasant.

[0021] The invention can be used to adjust positions of the vocal tract with a lower or higher loss factor. For this purpose, a wave field is provided in front of the mouth, which corresponds to the acoustic resonance conditions of the vocal tract resonances at the location of the mouth. These wave fields are actively amplified or attenuated by the user. If specific frequencies or frequency bands are offered, the user can actively amplify or attenuate these wave fields. Fig. 5, Fig. 7 and Fig. 10 shows, for example, that a resonance of the vocal tract amplified through the mouth can be maintained even when stimulated by the lungs, ie by a breathing sound or vocalization. Furthermore, the example shown in Fig. The experiment shown in Figure 4 shows that the loss factor or bandwidth of formants of the vocal tract is variable by adjusting the vocal tract and can be detected by the method or the apparatus used.

[0022] The invention offers the user the possibility of both measuring and specifically adjusting the acoustic properties of their own vocal tract or other acoustically resonating hollow bodies. In the case of hollow bodies, this application involves varying the geometry of the hollow body and adapting its acoustic properties to the specific wave fields available.

[0023] The signal or signals used to actively adjust the vocal tract are referred to below as "adaptation signals," and the corresponding frequency as "adaptation frequency," to distinguish them from the "measurement signals" presented in the next section, which enable the measurement of the acoustic properties of the vocal tract. "Adaptation signals" also refer to frequencies extracted from the measurement signal, which are made available to the user separately via a feedback mechanism and used to adjust the acoustic properties of the vocal tract, particularly via acoustic feedback.

[0024] The user adapts the vocal tract in such a way that the frequencies responsible for the carrying power of the voice are amplified. The training effect is achieved initially without using the user's own voice and only by amplifying or attenuating the resonance of the vocal tract in interaction with specific frequencies offered as a wave field through the mouth. Particularly helpful is the resonance setting of frequencies in a range that includes the second to fifth formants. These lie between approximately 700 and 4500 Hertz. However, the resonance setting is also helpful within the first formant (depending on positioning, between 200 and 1000 Hz). However, the higher formants up to the 10 kHz range also contribute to the carrying power of the voice and can be addressed separately.

[0025] The position found by adjusting the vocal tract to amplify the wave field corresponds to an improved adjustment of the vocal tract for using the voice or playing wind instruments. At the same time, it is also possible to change the position of specific formants for the correct pronunciation of foreign languages. This is particularly useful through training with two frequencies or frequency ranges that correspond to the desired formants. The invention therefore supports the user in bringing the vocal tract into an improved position for these applications. Through practice, a favorable setting can be found time and again. The method thus enables the active adjustment of predefined positions of the vocal tract.

[0026] Acoustic resonators exhibit natural vibrations.

[0027] The amplification of the adaptation frequency(ies) and / or the frequency band(s) by the vocal tract can be associated with an attenuation or a quality factor of the resonator. Amax~1α~f0Bw=Q

[0028] Where A is the amplitude of the resonator, i.e., here, the vocal tract, α is the loss factor, f0 is the resonator frequency, (Bw) is the bandwidth, and Q is the quality factor. In order for the adaptation signal to interact well with the resonance conditions of the vocal tract, a high sound velocity is required in front of the mouth. According to the generally accepted model of the coupling of a driving signal with a forced oscillation, both a maximum in the amplitude ( Fig. 3 and Fig. 4), as well as a phase shift of -90 degrees compared to the excitation as a resonance condition in Fig. 3 and Fig. 4. The externally applied adaptation signal acts as the driving force, and the resonator is also excited by the driver. In the case of resonance, a maximum amplification is measured through the vocal tract, and the phase between the adaptation signal and the amplified oscillation shifts to approximately -90 degrees according to the formula tan φ=2αωω02−ω2

[0029] With the angular velocity ω = 2πf of the driver and the phase shift φ between the driver and the resonator.

[0030] This phase relationship also applies to the antiresonance setting. A gain of 1 occurs when the angular velocities between the driver and the resonator are sufficiently far apart. According to the model of a damped oscillation, the resonance condition, determined by the phase shift between the excitation signal without resonance and with resonance, can be generated at different amplitudes. A higher amplitude can be directly linked to a reduction in damping. This lower damping, generated at specific frequencies, simultaneously corresponds to a narrower oscillation bandwidth or an increased quality factor of the resonator.Adapting acoustic properties, for example, of the vocal tract with a view to optimizing resonance at one or more given frequencies can therefore reduce losses and lead to a clearer voice and / or better acoustic coupling of the vocal tract to the tube system of wind instruments. At the same time, clearly defining the formants across their narrow bandwidth increases speech intelligibility, or rather, the intelligibility of sung vowels. Both the amplitude and phase of a resonance setting between the excited wave field and the resulting amplification must be taken into account.

[0031] The vocal tract can be described acoustically using a tube model. The resonance conditions of a tube closed at one end are described by the radiation impedance at the tube opening. This can be simplified to: αL<<1; αk<<1 through: zm0ρ0cS=αL−i cos kL sin kLsin2kL+(αL)2cos2kL

[0032] (Zm0: radiation impedance at the aperture, L: length of the tube and k = 2π / λ; α: loss factor). The complex resistance and the real part are high when kL = n * π. Under these conditions, the radiation is small (antiresonance). At the resonance frequency, the imaginary part of the resonance disappears at kL=(2n+12)∗π with a pressure node at the open end and a pressure antinode at the closed end. For the sound velocity, the relationship is exactly the opposite: a node is found at the closed end and an antinode at the open end of the tube. Under resonance conditions, the amplitude is largely determined by the loss factor α. The amplitude under resonance conditions can therefore be related to various loss mechanisms, such as those that can arise from unfavorable wall contact or reflection losses.

[0033] From the above, it follows that a resonator such as the vocal tract, when excited with a phase position corresponding to the resonance condition, i.e., with a high acoustic velocity at the open end, responds acoustically comparable to an excitation at the closed end. Therefore, coupling the sound waves with a pressure node (high acoustic velocity) at the mouth is particularly advantageous for stimulating the vocal tract, similar to the human voice via the vocal cords.

[0034] If only frequencies related to resonances (and no anti-resonances) of the vocal tract are offered, an amplification of the resulting signal can be directly related to a reduction of the damping factor.

[0035] Excitation can therefore be achieved with an acoustic signal, which has its own radiation impedance. In order not to change the resonance properties of the vocal tract through the interaction, this impedance should be small compared to the impedance of the vocal tract. This is generally the case when coupling occurs via the air and with resonance properties that result in a high sound velocity at the transition between the mouth and the driver, i.e. the phase position of the stimulating signal should correspond to the resonance conditions of the vocal tract at the coupling location. Coupling will be particularly effective when the driving sound source is located within the final correction element of the vocal tract, i.e. within a distance of approximately 0.8*r; where r is the radius of the opening of the hollow body to be measured. This can be achieved, for example, by a loudspeaker located close to the mouth.Since no additional resonators such as waveguides are used in this case, the phase can be specified by adjusting the electrical signal used to excite a loudspeaker. In particular, the phase relationship of different frequencies, especially overtones, can be specified.

[0036] Ideally, the entire signal from the sound generator can interact with the vocal tract. However, even with strong shielding of the sound generator, residues of the excitation signal remain that do not interact with the vocal tract. These wave fields make measuring the quality factor more difficult. In particular, small changes in the distance between the mouth or the opening of a cavity and the stimulating signal can interfere with the determination of the quality factor or the bandwidth of a resonance in a cavity or the vocal tract. The proportion of these interfering signals can be determined by evaluating amplitude and phase. The phase of the stimulating acoustic signal is preferably specified relative to the phase of the electrical signal driving a loudspeaker.

[0037] At the recording location, which is determined by the position of a recording microphone, both the acoustic wave field that interacts with the vocal tract and other stray signals are recorded together. A reliable measurement of the bandwidth is nevertheless possible using at least one waveform, such as a sinusoidal oscillation. However, this requires determining the portion of the sound wave that does not interact with the vocal tract. To enable this determination, the geometry of the acoustic system can be varied during the measurement of the phase and amplitude of the overall signal in such a way that the resonance frequency of a hollow body or vocal tract is crossed relative to the exciting frequency.For example, if the frequency of the coupling wave field is 900 Hz, the resonance conditions can be adjusted by changing the shape of the vocal tract, with a resonance frequency of around 900 Hz or up to an antiresonance frequency of around 900 Hz. The results of such a measurement are shown in . Fig. 3 (only setting a resonance) and Fig. 2 (adjustment of antiresonances as well). From the shift in the phase position and simultaneous measurement of the amplitude, the portion of the sound wave that does not interact with the cavity can be identified. Fig. 4 When coupling a fixed frequency of 1300 Hz, both the establishment of a resonance (maximum amplitude around a -90 degree phase shift) and an antiresonance (minimum amplitude around a -90 degree phase shift) can be observed. Both curves converge at smaller phase shifts. If only the resonating signal is measured, a phase shift of -180 degrees is measured in the case of a high excitation frequency and a low resonance frequency.

[0038] The amplitude at which the measured or extrapolated curves of the amplitude intersect as a function of the phase shift of resonance and antiresonance can be interpreted as the acoustic signal that corresponds to a gain of approximately one. If the intensity of the coupled wave at the location of the microphone is the same as the phase-shifted signal of a resonating signal, the magnitude of the maximum measured phase shift is reduced. For example, a value related to the bandwidth can be determined using the following method: Modifying an adaptation signal with a view to particularly intensive amplification, if necessary.also to an attenuation, recording the amplitude and phase of one or more adaptation frequencies, evaluating an amplitude / phase relationship to determine a measured gain factor of one, convolving the measured values ​​at maximum amplitude with stored data (calibration data) to determine a gain factor that can be directly related to the bandwidth or the quality factor. The necessary calibration data is obtained frequency-dependently from known tube models that are ideally adapted to the geometry of a typical vocal tract. With appropriate comparative measurements or using comparison curves already determined from experience, a quality factor or a bandwidth of the formant can also be deduced in a simplified manner from the maximum amplitude of the measured signal.In addition, to determine the quality factor, the amplitudes and phase pairs can also be recorded and compared with a previously saved characteristic curve. The described method makes it possible to position a microphone at a defined distance from the sound outlet and simultaneously extract information about the bandwidth.

[0039] If this approach is used for multiple frequencies simultaneously, for example, as they occur within a frequency band, the measurement can be made even simpler and / or more accurate. For example, if the vocal tract is tuned from a lower resonant frequency to a higher one, a phase shift is first measured at a given frequency around 900 Hz, followed by one at 910 Hz. From the relative position of the phase shifts at the two frequencies of known separation, the bandwidth of the vocal tract resonance can be directly determined. This is done based on the equations for tuning the phase and amplitude resonance and / or in combination with empirical data obtained from comparative measurements of specimens with known acoustic properties.Such an evaluation is best performed by software already installed on the device in use, which evaluates the amplitude and phase according to the procedure described above. The following steps are advantageous: Providing electrical signals of one or more frequencies, transmitting electrical signals to a loudspeaker, coupling the resulting acoustic signal to the vocal tract, measuring the resulting acoustic response at a specified distance from the sound source to a microphone, analyzing the phase and amplitude of one or more frequencies of the acoustic signal in comparison to the electrically adjusted parameters while varying the shape of a vocal tract, convolving the amplitude and phase values ​​with a correction function from a previous measurement and / or from the phase shift of two or more frequencies, preferably within a vocal tract resonance. As an alternative to measuring the phase shift, the time-varying amplitude of two or more frequencies within a vocal tract resonance can also be used.The accuracy of the evaluation of acoustic data can thus be further supported by varying the acoustic properties of the vocal tract during the measurement.

[0040] A particularly good coupling between the sound outlet and the vocal tract opening is achieved when the area of ​​the sound outlet is smaller than that of the mouth opening. This also allows for free breathing and only minimally alters the resonance conditions of the vocal tract. Therefore, the sound outlet should have an area of ​​2 cm. 2not exceed. It is also beneficial if the geometry of the sound outlet is adapted to that of the mouth. This can be achieved by a rectangular shape of the sound outlet. For example, for coupling into the vocal tract, a sound outlet of 8 × 2 mm is a good value. At the same time, with a rectangular arrangement, the flat side can rest on the lip. In general, the length to width ratio of the sound outlet should not be greater than 2:1 when used on the vocal tract, which roughly corresponds to the length to width ratio with a relaxed, open mouth. If other hollow bodies are used for the measurement, different maximum dimensions result in relation to the opening of the hollow bodies.

[0041] The transducer can be inserted directly into the mouth or into the area between the lips, into the vestibule, or into the oral cavity, where it generates the wave field. It can also be introduced into the mouth or into the mouth from an external sound source outside the mouth via a waveguide. The waveguide can be rigid or flexible. A flexible tube is more ergonomically advantageous. This tube can be tapered towards the end to allow for comfortable insertion of the waveguide into the mouth. If the signal is introduced into the mouth, a tube diameter of less than 3 cm, or preferably less than 1 cm, should be used to avoid restricting jaw flexibility during exercise.

[0042] This allows areas of resonance, as well as anti-resonance, in the vocal tract to be addressed. The frequency of the wave field used as feedback for resonance adjustment (adaptation frequency) is generally above 200 Hz. In general, the frequency and intensity of both the adaptation signal and the measurement signal should be selected or specified so that only minimal mechanical interaction with the tissue occurs. The higher the frequencies of the acoustic wave field, the less the interaction with the body and thus the strain on the tissue and vocal cords, especially with frequent use of the training instrument. At frequencies above 200 Hz, the wall of the vocal tract already functions predominantly reflective.

[0043] The acoustic signals preferably have a maximum sound pressure level (SPL) of 80 dB SPL at the mouth, which is well below the pain threshold of 130 dB SPL. If a sound pressure of 80 dB SPL or higher is used, it is advantageous to limit the duration of use in accordance with medical experience regarding safe sound exposure. The invention can already have such a control mechanism integrated into the control of the sound generators. In addition, attenuation of the acoustic signal at the subject's ear can be beneficial. This can be achieved using passive or active hearing protection, for example, as used in occupational safety devices. Active or passive ear protectors can be used. Using hearing protection also attenuates other disturbing external noises, and the required intensity of the signal used can be further reduced.A high sound pressure at the mouth can be helpful for a short time, as this allows the resulting sound pressure to be felt within the vocal tract. A variable change in the sound pressure at the mouth is also helpful. The amplitude can be expressed as a rising or falling sound field.

[0044] Defined wave signals such as sine, triangle, sawtooth or square wave can be used for stimulation. By selecting the signal, the overtone spectra of the acoustic wave field with or for introduction into the vocal tract are offered to the user together with the fundamental frequency to create resonance. The user can then either amplify individual frequencies in a targeted manner or use a combination of different frequencies simultaneously for amplification. To adjust an acoustic wave field, the rise times of the square wave, for example, can be shifted symmetrically or asymmetrically or adapted to the acoustic needs of the respective user. If, for example, a specific overtone range is to be amplified and coupled to another formant at the same time, the signal can offer overtones in different frequency ranges for amplification by adjusting the vocal tract.For example, with a fundamental frequency of 600 Hz and a particularly prominent third and fourth overtone at 1800 Hz and 2400 Hz, respectively, the vocal tract can be adjusted simultaneously for all three frequencies, with the frequencies located within different formants. It is also advantageous if the user is able to select different waveforms and use them for the introduction to the vocal tract.

[0045] The fundamental frequency of the signal or signals used should be below 3500 Hz, preferably between 600 and 2500 Hz. The overtones may well lie in a higher frequency range, but are defined and adjustable via the phase relationship to the fundamental tone. This setting can be integrated into the device's control signals. These relative phase positions are advantageously based on the phase relationships that occur within the vocal tract. It is therefore advantageous to adjust the phase relationship setting for exciting resonances in the vocal tract according to the variation in phase position within the formats to be excited. This setting can already be integrated into the device without the need for interaction with the user.

[0046] Low-frequency components do not interfere with the functionality of the invention; they can even be used to make the signal more pleasant for the user. For example, the external acoustic signal can be adjusted to the user's natural speaking frequency. Likewise, overtones of individual tones, sounds, noise, or musical chords can be used as a feedback signal. For example, if a resonance frequency of 840 Hz is to be used as the target frequency for the optimized setting, a sound composed of a tone with a fundamental frequency of 210 Hz can be used, and the third overtone at 840 Hz can be used for the resonance adjustment through the vocal tract. In addition, other tones or noises can be mixed in.

[0047] At least two of the following technical procedural steps are carried out: selection or determination of frequencies or frequency bands; when using multiple frequencies, a selection or determination of the phase relationship between them; electrical control of a loudspeaker with frequencies or frequency bands, whereby when using multiple frequencies these should be within a specified phase relationship to one another; excitation of the air column of the vocal tract by a sound source which corresponds to a volume source; measurement of a resulting acoustic signal; evaluation of the changes in amplitude and phase resulting from the interaction of the acoustic wave field with the vocal tract, whereby the temporal change in the amplitude / phase relationship resulting from a change in the vocal tract during the measurement can also be included.the recording and storage of further acoustic, particularly vocal, signals before, during, and after the stimulation of the vocal tract with a wave field; an evaluation of the measured acoustic signals according to parameters such as volume, bandwidth of one or more formants, one or more quality factors of resonances and / or antiresonances as a function of time and / or frequency; the representation of the measured acoustic parameters, which can also be comparative; as well as the representation of a temporal change in acoustic parameters.

[0048] All or some of the above steps are advantageously integrated into a computer program and / or permanently installed on a computer that controls a suitable apparatus.

[0049] Fig. Figure 1 shows a typical setting of a user who is already accustomed to using the device. An arrangement comparable to Fig. 15 with two flexible waveguides for sound transmission to the mouth and signal pickup to a microphone. The two ends of the waveguides had an inner diameter of 6 and 4 mm, respectively, and were loosely placed on the user's lips. The signal was fed at a voltage of approximately 6 mV directly from the headphone output of a mobile device to a Visaton BF 32 broadband loudspeaker and is barely audible through the mouth without resonance adjustment. Sinusoidal signals with frequencies of 201 Hz and 803 Hz were used for excitation. The microphone signal was fed via a preamplifier to a lock-in amplifier (reference frequency: 803 Hz), and only the signal at 803 Hz was read out. Fig. 1, only a slight signal is detectable at 7 seconds when the acoustic signal is activated. Only after 10 seconds does the signal become clear when the mouth position is varied and the resonance conditions are adjusted. The variation in mouth / tongue / throat and neck position is directly detectable from the electrical signal and can be associated by the user with these movements.

[0050] It is advantageous to use different adaptation frequencies within a frequency range. For example, frequencies of a fundamental oscillation of 800, 850 and 900 Hz of the aforementioned electrical control signals to the loudspeaker can be combined to enable broader coverage of a resonance range in the vocal tract. The frequencies within a frequency range can be very close to one another, for example at 5 Hz intervals. This can be achieved, for example, by combining frequencies of 800 Hz, 805 Hz, 810 Hz,..., up to 830 Hz into a frequency band. The width of such a frequency band is advantageously in the range of the bandwidth of the formants. This has the advantage that the adaptation signal, as a defined frequency band, is perceived as more natural by the user. At the same time, an evaluation of the resonance of the individual frequencies in the frequency band is possible for the direct determination of the quality factor orthe bandwidth of the vocal tract at the adaptation frequency. At the same time, the precise position of the adaptation frequency can also be achieved within a frequency band by more strongly emphasizing the amplitudes of individual frequencies within the frequency band. In this case, frequencies in the middle of the frequency band can be advantageously emphasized, while the frequencies at the edges have a greater role to play in enabling the measurement of the bandwidth and making the signal more pleasant. In general, the necessary flexibility can be achieved with a bandwidth of the frequency bands below 400 Hz. The targeted use of lower bandwidths of the frequency bands can also be advantageous for the application. The bandwidth can be determined both by determining the amplitude distribution and by determining the phase relationship.

[0051] If the adaptation signal is accompanied by an additional oscillation, the bandwidth can also be determined from the fluctuation in amplitude versus frequency. Fast and small fluctuations are preferable, as in this case the settings of the vocal tract are not changed or only slightly by the user. For example, with an adaptation frequency of around 800 Hz and an assumed bandwidth of 50 Hz, a small variation of plus / minus 10 Hz can already result in a measurable change in intensity and thus serve to determine the bandwidth. With an oscillation of plus / minus 25 Hz in this example, a drop in sound pressure of 3 dB can already be expected. The oscillation of the adaptation signal can occur constantly. However, it can also occur in bursts after a hold time at a stable adaptation frequency or adaptation frequencies at time intervals, for example within a time period of 100 ms.

[0052] The number of adaptation frequencies or frequency bands used is based on the human formants. Therefore, it is advantageous to offer several wavebands simultaneously, which can be amplified simultaneously. One adaptation frequency or frequency band is usually sufficient to perceive a noticeable effect. If multiple adaptation frequencies or bands are used, five of these bands are usually sufficient. In these cases, supporting the feedback by recording with a microphone and generating acoustic, visual, or haptic feedback is helpful.

[0053] Fig. Figure 2 shows a typical sound level measurement after adjusting the adaptation signal by changing the resonances of the vocal tract over time at frequencies of 700 Hz, 900 Hz, and 1300 Hz, and during normal breathing through the mouth. The scale bar corresponds to a gain of 20 dB. The curves were recorded consecutively, with the distance from the microphone to the mouth being 20 cm. A clear amplification of the acoustic signal is visible due to the resonance adjustment of the vocal tract. The measurement setup corresponds to a setup comparable to Fig. 16.

[0054] When the vocal tract is set to resonate, a phase shift of -90° relative to the acoustic phase of the signal generated by the loudspeaker can be observed. This phase position, like the amplitude, can therefore be used to determine the resonance setting of the vocal tract relative to the adaptation signal. As shown in Fig. 4, both signals - the excitation and the amplification or attenuation by the vocal tract - are measured by a microphone. The phases of both signals overlap to form a resulting wave field with a phase position that lies between the two signals. This reduces the measured phase shift, depending on the relative amplitude of the adaptation signal and the signal modified by the vocal tract. If both signals are equally intense, a reduction in the phase shift of 30° can be expected. It can therefore be helpful to position a microphone in such a way that the signal from the vocal tract is predominantly measured. If the acoustic wave field is introduced directly into the mouth, acoustic separation is already provided by the position of the microphone outside the mouth. If the microphone and sound generator orIf the output of a waveguide is mounted at a comparable height, this can be achieved by placing an acoustic barrier between the microphone and the exciting wavelength. This barrier is preferably made of an acoustically insulating material. It can also serve to define the geometric relationships between the exciting wave field and the position of a microphone. If a barrier is used, the width of an acoustic barrier between the sound outlet of a loudspeaker and the opening of a microphone should preferably be at least half the length of the final correction term of the sound outlet and at most twice the length of the final correction term of the sound outlet. The barrier can also be designed as a wall of a sound generator's sound outlet extended on one side.

[0055] A measurement can also be performed with a microphone located further away from the mouth. It is advantageous that this microphone largely picks up the signal that is also available as a feedback signal through the ear. This is achieved by a position that lies outside the final correction term of 0.8 cm * r, where r is the radius of the mouth opening. While the coupling of the adaptation signal therefore preferably occurs within the final correction term, measuring the acoustic signal outside the final correction term, similar to perception at the ear, is advantageous.

[0056] It is also possible to record the amplitude and phase of the input signal without resonance formation, ie with the mouth closed, and to use it as a comparison value to calculate or estimate the output impedance of the vocal tract.

[0057] Fig. Figure 3, for example, shows the adjustment of the resonance amplitude in relation to the phase of a sinusoidal adaptation signal. The curves were recorded during a 15-second adjustment period with nasal breathing, and the adaptation frequency was 900 Hz. The loudspeaker excitation signal was compared with the microphone signal using a lock-in amplifier. The acoustic waves were fed in and out via a waveguide both for introducing the sound field to the mouth and for collecting the signal from the mouth to the microphone. No phase shift and an amplitude of zero correspond to the initial position with the mouth closed. The amplitude of the amplified signal is particularly intense at -90°, while a phase shift beyond this is accompanied by a reduction in amplitude.

[0058] Fig. Figure 4 shows the relationship between phase and amplitude of vocal tract excitation at different frequencies using a single sinusoidal adaptation signal. The parameters were recorded using a lock-in amplifier over a period of approximately 30 seconds. Breathing occurred through the open mouth. The sound generator was an iPhone 5s speaker / buzzer. Zero degrees of phase shift corresponds to a loudspeaker signal with the mouth closed. In this setup, the microphone picks up the adaptation signal at a distance of 20 cm and passes it on to the amplifier input. Shown are Fig. Figure 4 shows the results of the adjustment compared to the adaptation frequencies between 700 and 2000 Hz. It is clear that the resonance setting, measured by the frequency shift, can be achieved at different amplitudes. If the resonance is achieved with a higher amplitude compared to the adaptation frequency, this corresponds to a lower attenuation at the corresponding frequencies, a narrower bandwidth, or a higher quality factor. It is obvious that lower attenuation of the resonance has a positive effect on speech intelligibility and / or vocal carrying capacity.

[0059] Breathing through the mouth / or exclusively through the nose significantly changes the acoustic properties of the measured area of ​​the vocal tract (compare Fig. 3 and Fig. 4). If the contact between the throat and mouth is closed by raising the tongue, only the oral cavity is addressed. An adjustment of the resonance frequency of the vocal tract induced by a change in the adaptation frequency causes, in particular, an expansion of the mouth volume and a lowering of the larynx. Breathing through the mouth allows the entire acoustically relevant area, from the mouth and throat to the larynx, to be captured. This is particularly useful for exercises to improve vocal function or for exercises to improve the resonance of wind instruments.

[0060] In addition, by analyzing the phase relationship simultaneously with the intensity analysis, the bandwidth of the resonance or the quality factor can be determined. According to a phase shift between the adaptation signal and the resulting amplified signal (compare Fig. 3, Fig. 4) a high amplitude can be seen with a phase shift of -90 degrees. However, the mixing between the excitation amplitude and the amplified signal via the resonance must also be taken into account. If both signals are recorded simultaneously, the phases mix according to the summation of both signals, resulting in a deviation of the measured signal of -90 degrees. Thus, with the same signal component, the exciting frequency (compared to that amplified by the vocal tract) corresponds to a phase shift of -50 degrees. This value can vary slightly depending on the stability of the excitation signal radiation when the mouth is open compared to closed and is best determined empirically.At the same time, the influence of the excitation signal on the measured phase shift can be achieved by acoustically isolating the excitation signal from the microphone, for example, via a waveguide from the mouth / oral cavity to the microphone or an acoustic barrier between the excitation and the microphone. This has the advantage of predominantly recording the amplified signal, thus further attenuating the influence of the direct loudspeaker signal.

[0061] Fig. Figure 5 shows a spectrum of breath sounds after prolonged adjustment of the vocal tract to 700, 900, 1300, and 2000 Hz during breathing through the mouth, respectively. These evaluations are also possible while adjusting the vocal tract to an adaptation signal. While the upper part of the figure shows the microphone signal of the breath sounds, the lower part shows an analysis of the signal analogous to the representation. Fig. 5. The arrows indicate the position of the adaptation frequencies used. The use of the adaptation frequencies allows for a defined position of the resonances of the vocal tract. In addition, other resonances can be identified, which can be evaluated separately or provide information for further optimization when using new adaptation signals. For example, both resonances can be amplified jointly by two superimposed adaptation signals in the range of the resonance setting thus generated and – if necessary, with the specification of a specific ratio to each other – by using the training device.

[0062] After adapting the vocal tract geometry to the resonance specifications using one or more external frequencies without using one's own voice, an immediate change in the tonal characteristics can be measured, even with the use of the voice, and can also be heard or felt directly by the user. The same applies to other applications such as playing wind instruments, learning foreign languages, or training muscles or muscle groups in the neck / head / chest area.

[0063] Fig. Figure 6 shows the analysis of a male user's sung vowel "a" recorded with a microphone, with the frequencies represented as an overtone spectrum after a Fourier analysis (analyzed using the Wavesurfer program). The sung note corresponds to a fundamental frequency of around 140 Hz. The recording was made before and after using a square wave signal of 900 Hz as an adaptation signal. This electrical signal was transmitted to a loudspeaker to generate an acoustic wave field. The distance from the loudspeaker membrane to the mouth was 4 cm. The wave field was specifically amplified by the user by changing the position of the tongue, jaw, and palate, as well as the position of the larynx. The exercise was carried out for 60 seconds. The range shown is from 17 Hz to 10 kHz. The fundamental tone and the first overtones changed less as a result of the exercise than the higher-frequency ranges.The increase in intensity is clearly evident in the spectral range from about 1200 Hz, but especially in the range of the singer's formants. Furthermore, components above 4000 Hz can only be generated after the acoustic wave field has been initiated and the resulting adaptation of the vocal tract by the user.

[0064] If cavities are measured, the geometry of the cavities can be changed so that the acoustic properties of the cavities are achieved at defined frequencies and / or frequency bands. Analogously, the resonances of the vocal tract can be adjusted without using one's own voice and without vibration of the vocal folds, corresponding to normal voice use, i.e., without so-called phonation. It is particularly important that the resonances developed after such an adjustment without phonation (when breathing through the mouth) are also retained when the vocal folds are used. This applies both to the sounds produced during exhalation and to spoken or sung sounds after a practice phase according to the invention. Fig. Figure 7 shows a spectrogram of a sung tone of a male subject after prolonged adjustment in relation to specific adaptation frequencies (sine signals) of 700, 900 and 1300 Hz. The acoustic wave field was measured using an arrangement comparable to Fig. 12 and with an efficient coupling of the sound velocity through the mouth. The adjustment was made using the invention without a vocal signal, in Fig. Figure 7 evaluates the subsequently recorded vocal signal. The user did not actively produce a vowel, meaning, unlike the other spectrograms shown, during phonation after adjusting the vocal tract, the position of the tongue, palate, jaw, larynx, lips, etc., found during the feedback phase was maintained as much as possible. The amplification of the adaptation frequencies after adjusting the vocal tract is clearly observable, even during vocal stimulation of the vocal tract by the vocal folds.

[0065] To generate the sound field, primarily electrical signals are used, which are then transmitted to a sound generator. The sound generator can be a conventional loudspeaker, a loudspeaker such as those used in mobile devices (e.g., tablets or mobile phones), or a ferroelectric sound generator, i.e., a so-called piezo loudspeaker. Loudspeakers already used in smartphones are also suitable for the training device. In this case, the smartphone can also be embedded in an acoustically insulating environment that leaves the loudspeaker output free as a sound opening. If this type of loudspeaker is incorporated into a separate device, encapsulation of the loudspeaker and / or acoustic isolation of the loudspeaker is advisable.

[0066] The adaptation signal and the measurement signal can be delivered to the mouth via a single transducer or two separate transducers. The use of two separate transducers may be useful if the measurement signal or the adaptation signal have different output impedance requirements for the transducers at the mouth, and / or if one or both signals are delivered to the mouth via waveguides.

[0067] In addition, additional frequency bands or frequencies can be superimposed on an adaptation signal, or switched on or off (measurement signal). These frequencies simplify an acoustic analysis of the acoustic adjustment of the vocal tract. Broadband measurement signals are suitable for this purpose, e.g. noise signals that cover a broad spectrum from approximately 200 Hz up to 20 kHz, with the particularly relevant frequency range being between 200 Hz and 5 kHz. White or, better yet, pink noise can be used for this, taking the relevance of the different frequencies into account for the evaluation. Deterministic noise with a stable phase relationship between the individual frequencies is more suitable. The more clearly the individual frequencies are defined and the more distant they are from one another, the better the measurement signal can be evaluated and differentiated from other noise sources.At the same time or in addition, the noise can be adjusted by integrating the measurement signal over a period of time, preferably a flexible one. Integration times of over 100 ms can significantly reduce the noise spectrum and, in principle, enable a frequency resolution of up to 10 Hz after Fourier transformation. With an integration time of one second, a resolution of just one Hertz can be achieved.

[0068] To make the evaluation more flexible, a number of different measurement signals can be selected. If a measurement signal is synthetically generated from different individual oscillations as deterministic noise, the spacing of the individual frequencies determines the measurement accuracy with respect to the frequency. Conversely, a better signal-to-noise ratio can be achieved with a greater spacing of the frequencies. The sampling rate can be adapted to the needs of the analysis depending on the frequency range. This is helpful for precise assignment of frequencies during subsequent analysis and for noise suppression of the overall signal. For example, in the range from 200 Hz to 1200 Hz, a spacing of 10 Hz between the individual sine frequencies can be used (i.e., 200 Hz, 210 Hz, 220 Hz, etc.).), while in the range from 1200 Hz to 2000 Hz a spacing of 20 Hz between the sine waves of the composite signal is used and in the higher frequency range a spacing of 30 Hz is selected. The ranges can also transition smoothly from one resolution to another. At the same time, the intensity of the measurement signal used can be adapted across the frequency range. For example, a higher intensity in the particularly relevant frequency range can make the measurement more precise, while lower intensities are used in the less interesting ranges, particularly the higher frequency components above 5 kHz. In general, a lower intensity is used at higher frequencies, as is the case with so-called pink noise, for example. If particularly precise frequency sampling is required, the frequencies should be closer together, particularly in the particularly relevant frequency range.This can be configured to achieve resolution within a very narrow frequency distribution, even using random noise. If a deterministic signal is used as the measurement signal, the spectral analysis is advantageously tuned to the frequency maxima of this measurement signal.

[0069] In the case of deterministic noise, the phase relationship between the measurement signal and the signal amplified by the vocal tract, or the phase position of the measurement signals relative to each other, is also known and can be used to improve the signal-to-noise ratio. If an adjustment is made using selected frequencies of the measurement signal reflected back to the user via feedback as an adaptation signal, the phase relationship can also be specifically used or evaluated.

[0070] The ratio between the adaptation signal and the measurement signal should ideally be designed such that the adaptation signal and its variation due to changes in the resonance conditions within the vocal tract are still clearly audible compared to the measurement signal. This generally means that the oscillation of the adaptation signal is attenuated by no more than 21 dB compared to the measurement signal, typically no more than 12 dB. However, at low intensities, the adaptation frequency can also be visualized or fed back optically or haptically as a feedback signal after recording and further processing of the signal via a microphone and appropriate data processing. Haptic feedback, for example, through shakers or vibrators attached to the chest and / or neck and / or wrist, enables coupling to established vocal pedagogy.Vibrating elements are preferably placed in locations where, according to established vocal pedagogy, a vibration sensation can be felt with a favorable setting, and frequencies are used that can also be felt during singing. This links the adjustment of the vocal tract using the training method described here with the established kinesthetic perception used by vocal teachers as biofeedback. Evaluations or displayed intensities obtained from the adaptation signal or the measurement signal are beneficial for effective biofeedback. These can be made available to the user visually via a display. When using multiple adaptation frequencies, the relationship between the frequencies and / or their individual intensities can be shown.

[0071] The amplification of different frequencies can also be used for so-called formant tuning to amplify specific overtones during singing. At the same time, an analysis of the measurement signal already provides clues as to which direction the formant setting should be adjusted compared to the adaptation signal. In so-called format tuning (Sundberg 2013), the formants are adjusted to the overtones of the notes to be sung by changing the vocal tract. For example, if a tenor sings a high A (440 Hz fundamental frequency), a resonance adjusted by the vocal tract around 880 Hz can amplify the first overtone, and a resonance adjusted around 2.1 kHz can amplify the fifth overtone, thus making the voice more carrying and colorful. The influence on the timbre can also be individually tailored.By recording a vocal signal with a microphone and / or recording the adaptation signals / measurement signals, an optimized setting can be repeatedly retrieved and compared with the current training status. The phase relationships of the adaptation signal amplified by the vocal tract can also be used to evaluate resonance intensity, as defined phase relationships of the various frequencies allow for separation from random noise or interference.

[0072] Fig. Figure 8 shows an example of the temporal variation of the adaptation signal and the superposition of an adaptation signal with a measurement signal. The evaluation of Fig. 8 was performed using the program Wavesurfer, version 1.8.8.p4. A dark color corresponds to a higher sound pressure at the corresponding frequency or time, while a lighter color represents a lower amplitude at the corresponding frequency or time. While the measurement signal (here a deterministic noise signal with closer spacing of the superimposed oscillations at lower frequencies and somewhat larger at higher frequencies) in Fig. 8 was kept stable, the adjustment signal (adaptation signal) was initially kept constant at 600 Hz and then varied from 600 to 2000 Hz. This was followed by a hold phase at 2000 Hz. In this case, the adaptation of the vocal tract to the adaptation signal was achieved via the ear or via a visual representation of the adaptation signal, the amplitude of which was changed by the vocal tract. At the same time, the measurement signal can be recorded and / or further processed and / or visualized and, with sufficiently fast data processing, provides information about the acoustic properties of the vocal tract. However, an evaluation can also be carried out in such a way that the adaptation signal is initially operated without a measurement signal to adjust the vocal tract and only then is the measurement signal added or replaced by the adaptation signal.

[0073] Fig. Figure 9 shows the change in the measurement signal with the mouth open and after adjusting the vocal tract in relation to a resonance of the adaptation signal. Fig. 9 are compared to the shot with the mouth closed ( Fig. 8) The vocal tract's tuning in terms of resonances (darker areas = amplification of the available measurement signal) and anti-resonances (brighter areas = attenuation of the measurement signal) can be clearly seen. It is clearly visible how, on the one hand, the resonance setting of the vocal tract adapts to the adaptation signal, and how the acoustic properties of the vocal tract can be recorded from the tuning relative to the adaptation signal. Thus, at the starting frequency of the adaptation signal at 600 Hz, a resonance in the range from approximately 600 Hz to 900 Hz is primarily evident, while above this frequency a narrow frequency range of anti-resonance around 1.2 kHz can be seen. In contrast, the resonance areas are less pronounced when the vocal tract is tuned at 2 kHz relative to the adaptation signal. At the same time, the anti-resonance visible when tuned relative to an adaptation frequency below approximately 1300 Hz is no longer present.

[0074] In addition, the adaptation signal in Fig. Figure 8 is a good example of how the adaptation frequency can be varied during use of the method or device. Overall, different frequencies help in the measurement of the acoustic properties of the vocal tract. In this case, different frequency ranges can be acoustically controlled and evaluated. This change in frequency can be Fig. 8 are composed of time segments. A variation of the amplitude can also be a temporal component. Overall, both signals that are not changed over time and signals that change over time, such as the sweep in Fig. 8. This temporal change may, for example, include a linear or non-linear sweep, an oscillation, a step function, or a combination of such sections as a temporal sequence.

[0075] When evaluating the measurement signal, the spectrally broken down sound pressure, such as in Fig. 9 shown by the signal with the mouth closed, as in Fig. 8 shown, divided. Fig. Figure 10 shows a possible representation of the acoustic properties of the vocal tract from a measurement with a corresponding measurement signal. The ordinate shows the ratio of the open-mouth sound pressure divided by a closed-mouth reference signal.

[0076] It is in Fig. 9 clearly shows that, for example, due to the temporal change in the acoustic conditions according to the specified sweep of the adaptation signal, both resonances and anti-resonances are passed through at different frequencies. This measurement also enables the evaluation of the quality factor via the resulting amplitudes and phase shifts between the electrical signal exciting a loudspeaker and the measurement signal from a microphone. This evaluation is possible not only across the frequency range of the adaptation signal, but across all ranges in which a change in the acoustic signal occurs due to the changed acoustic conditions of the vocal tract, i.e. in the above example, across the entire range between approximately 600 Hz and approximately 5000 Hz.

[0077] From the examples given in Fig. The resonances determined in Figure 9 are passed through various resonances and antiresonances, for each of which a specific frequency-dependent loss factor can be determined. This is reflected in the diagram by an overall lower amplification of the acoustic signal in specific wavelength ranges. For example, in Fig. 9 a relatively higher loss factor can be seen in the range of 1300 Hz.

[0078] After such or a similar measurement by means of a frequency sweep or by successively offering adaptation frequencies, the achievable bandwidth or the quality factor can be determined, stored, evaluated or displayed as a function of the frequency.

[0079] A Fourier analysis of the signal compared to a reference - recorded with the mouth closed - is, as already mentioned, in Fig. 10. The ratio of the measurement signal with the mouth open to the same signal with the mouth closed essentially corresponds to the acoustic impedance of the vocal tract. If the geometries of the stimulating signal to the mouth are known or constant, the absolute acoustic impedance can be determined by analyzing the measurement signal. For this purpose, the device is calibrated, for example, using test specimens with defined acoustic properties such as resonators (tubes open on one side or both sides, Helmholtz resonators, etc.). The absolute determination of the acoustic impedance is particularly useful for enabling comparison with different speakers / singers, etc., and / or for training speech recognition for spoken or sung sounds. At the same time, larger shifts in the overall setting of the vocal tract during a speaking or singing training phase can be tracked over a longer period of time.

[0080] It is clear in Fig. 10 The strong resonance around 800 Hz can be seen when the vocal tract was set to this frequency as an adaptation signal, whereas when the vocal tract was set to an adaptation signal of 1300 Hz, the resonance characteristics were significantly shifted into the higher frequency range. Furthermore, other maxima are also visible due to the setting, which provide clues to the acoustic properties of the vocal tract without phonation. The frequency ranges amplified using the voice are in a comparable range. This means that by measuring the acoustic properties of the vocal tract without using the voice, the resonance possibilities under phonation can also be trained.

[0081] Fig. Figure 11 shows a spectrogram of a sung signal after adjusting the vocal tract to an adaptation frequency of 600 or 1600 Hz respectively. The adjustment was made without using the voice. After adjusting to the signal, the note C was phonated and the resulting signal was converted into a frequency representation using a Fourier transformation. In both cases, after adjusting without phonation, the same vowel was phonated and the measured vocal signal was evaluated. Because a vowel was phonated, the position of the formants deviates from the adaptation frequency. However, after adjusting the vocal tract to the adaptation frequency, a clear change in timbre can be seen despite phonation on the same vowel. This shows that the attitude towards specific signals influences the overtone spectrum and the carrying capacity of the voice when producing vowels.

[0082] After adjusting the vocal tract to an adaptation frequency of around 600 Hz, a strong resonance can be detected up to around 900 Hz, which drops sharply down to around 1600 Hz and only reaches a maximum again at 2.4 kHz. The voice sounds dark, but is sustained due to the high proportion of overtones in the frequency range around 2.4 kHz. In contrast, after adjusting the vocal tract to an adaptation frequency of 1.6 kHz, the frequencies up to 900 Hz are supported less strongly, but the frequency range in the overtones is more significantly expanded. This voice sounds significantly brighter in comparison. Therefore, the timbre of the voice can be adjusted by adjusting the vocal tract.

[0083] Since in the invention both signals - the stimulation and - if the signal is amplified by a resonant setting of the vocal tract - also the signal amplified by the vocal tract - are incorporated into the biofeedback, it is helpful to have as little radiation of the acoustic signal as possible. If a waveguide is used, it should therefore be acoustically shielded. This can be omitted if the loudspeaker is placed close to the mouth. If a loudspeaker is used, a distance of less than 1 / 4 of the wavelength of the adaptation frequencies used is sufficiently small to be able to couple well into the vocal tract in accordance with the resonance conditions of a cavity or vocal tract. This corresponds either to a loudspeaker directly in front of the user's mouth or to a design with a short waveguide.When the transducer is inserted directly into the mouth, the excitation signal is further attenuated for the user, making the resonance effects even more audible. This is particularly evident in the so-called "lumbed" regions of the waveguide, where the waveguide length is small compared to the wavelength.

[0084] The microphone is advantageously positioned outside the mouth. This also allows the microphone to better pick up the signals that reach the user's ear. This is exemplified in Fig. 12 shown. In Fig. 12 is the signal generator, which generates signals itself according to a frequency generator, but can also access previously stored signals. The data processing, visual preparation, and an amplification unit are integrated in a device (5), e.g. in a tablet, PC or a smartphone. The microphone that is usually already integrated in a mobile device can also be used on this device (5). However, an external microphone (4) can also be attached close to the mouth. This is shown in drawing 12 via a cable (not labeled) that is also used to supply power to the loudspeaker (3). The waveguide (6) can transmit the acoustic wave field generated by a loudspeaker (3) into the open or closed mouth (2) of the user (1). The waveguide (6) can also be designed as an exchangeable mouthpiece.The loudspeaker (3), preferably a ferroelectric loudspeaker (piezo loudspeaker) without a waveguide or mouthpiece, can also be positioned near the mouth (2) of the user (1), thus introducing the wave field. Furthermore, the acoustic wave field can be generated directly in the mouth area, preferably by a ferroelectric loudspeaker. When generating the acoustic wave field, it is advantageous to adjust the frequencies used and the phase relationships to the overtones of the vocal tract via the electrical signal controlling the sound generator. The acoustic properties of the waveguide are taken into account in this process.

[0085] The position of the microphone captures as much of the resonant signal amplified by the vocal tract as possible. Therefore, it is advantageous to acoustically isolate the microphone from the sound source. With acoustic isolation from the exciting wave field, the phase position can be more easily detected as resonance (shift of -90°) with a single microphone. This is because both signals (driver and resonant gain) are measured together, and the relative intensity of both signals determines the phase position. For example, if a resonance with a gain of one is present (resonant signal with the same amplitude compared to the driver), a phase shift of the overall signal of only -60° is measured, even in the case of resonance. This relationship between phase and amplitude can be used to determine the amplitude of the incoming wave field.Direct measurement of the resonant signal can be improved by acoustically isolating the driver and microphone. Even a 50% signal attenuation by acoustically isolating the wave field output from the speaker to the microphone can be helpful.

[0086] If the microphone is further away from the mouth, a waveguide to the microphone helps to selectively record the sound field both in front of and away from the mouth area. If the signal is recorded from the mouth area, calculating the measured signal is particularly helpful compared to the measurement of test specimens. The same applies when the excitation signal (including the superimposed adaptation signal with the measurement signal) is used. Comparing the entire setup with one or more test specimens with known acoustic properties provides a reference signal that enables an absolute measurement of the acoustic properties of the vocal tract.

[0087] Miniaturized microphones, such as those found in mobile devices, are particularly well-suited for this purpose, but other microphone types can also be used. An existing microphone, for example, in a setup used to determine the sound field in phoniatrics, can also be used to measure the acoustic properties of the vocal tract. For these applications, a microphone is often attached to the subject's head, allowing a reproducible distance between the mouth and the microphone of, for example, 30 cm. Wireless microphones can also be used, allowing a largely wireless setup.

[0088] The training device can, as in Fig. 13. The acoustic wave field is guided via an enclosed loudspeaker (3) into the mouth (2) of the user (1). The housing and the acoustic insulation (8) of the loudspeaker enable minimal crosstalk to the microphone (4). A partition wall (7) is located on the acoustic insulation (5), which can be straight or curved. In addition to acoustically separating the excitation signal from the microphone, the acoustic partition wall (7) can optionally also be used to adjust the distance between the sound opening of the loudspeaker and microphone on the one side and the mouth on the other side. The already stored signal is made available for control, for example, by a computer / PDA (5). The evaluation and visualization are carried out equally by the computer or PDA.Data processing enables the filtering and processing of the recorded adaptation signal and the measurement signal, which is then fed back to the user as feedback via a monitor (7). Stored acoustic signals or further processed signals can also be retrieved and / or compared with the current values / images. A possible arrangement for introducing the acoustic signals into the mouth is shown in . Fig. 14: The user (1) has introduced a signal coming from a loudspeaker / sound source (3) into the mouth (2) via a waveguide (6). The waveguide is preferably short. The radiation of the sound source outside the transmission through the waveguide can be reduced by a housing (8). In this embodiment, the waveguide is inserted into the mouth (2). This has the advantage that the signal stimulating the vocal tract is picked up less strongly by the microphone (4) mounted outside the mouth and primarily the acoustic interaction of the vocal tract is recorded. The microphone can pick up the acoustic signal from the mouth at a defined distance via a second waveguide (10). The microphone and the opening of the signal generator are expediently mounted at a defined distance from one another. In the embodiment shown, this is achieved by a connection (9) between the two waveguides from the loudspeaker orto the microphone. The speaker and microphone are connected to the mobile device and a computer (5), which evaluates the signals.

[0089] The formants exhibit a phase shift in their resonance properties. Therefore, it is advantageous to adjust the phase relationship for the resonances in the vocal tract according to the variation in phase positions within the formats to be excited. However, this is also the case in locations where specific boundary conditions of the wave equations cause a pressure antinode or where the phase position is stabilized by the geometry used.

[0090] Acoustic wave fields within a tube, i.e. also in the vocal tract, have defined phase relationships between sound pressure and sound velocity, which are described by the acoustic wave equation. While in free propagation sound velocity and sound pressure are in phase, in a tube both quantities are offset by 90 degrees. When the adaptation signal is delivered to the mouth via a tube, possible resonances must be taken into account when evaluating the signals and designing the training device. The influence of resonances can be reduced by using sound-absorbing materials within the tube, preferably close to the end of the tube towards the mouth. By absorbing the sound waves, reflection at the end of the tube is minimized, which reduces the formation of standing waves, which are necessary for resonance training. At the same time, this means an increase in the impedance of the tube.If this method is used, feedback of the resonances created at the loudspeaker to the user (visual, acoustic or haptic) is possible. Immediate acoustic feedback through a found resonance setting of the vocal tract is better. For this purpose, a sound source with low acoustic impedance is advantageous, as the vocal tract can be loaded without influence on the sound source. In this case too, either an adapted resonance relationship must be set, preferably with a high sound velocity at the end of the tube, i.e. at the mouth, or the waveguide used must be shorter than 1 / 4 of the wavelength used so that the resonance relationships cannot develop in a disruptive manner. In the first case, this means adaptation to specific wavelengths that satisfy this condition.Therefore, two sound sources can be used, delivering a low-impedance and a high-impedance acoustic signal to the vocal tract, either directly or via a waveguide. In the latter case, the sound source should be positioned close to the mouth.

[0091] If an external sound source is used, the phase relationship of the overtones of the acoustic wave field can be adjusted via the length of the waveguide. The phase position of the sound velocity and sound pressure at the transition between the waveguide and the mouth is particularly important. The human vocal tract of adults is 14 to 17 cm long; in children it is correspondingly shorter depending on age and size. If different formants are activated simultaneously, it is advantageous to map the changed conditions of the phase position of the vocal tract in the acoustic wave field. This can be achieved, for example, by adjusting the length of the waveguide between the acoustic source and the mouth. If a straight tube is assumed to be the waveguide, the distance between the loudspeaker and the oral cavity or vestibule should be 0.1 m for frequencies within the range of neutral sounds (so-called schwa sounds: approximately 500 Hz, 1500 Hz, 2500 Hz, etc.).) can be approximately twice the length of the vocal tract or a multiple of twice the length of the vocal tract. The length of the attached waveguide can therefore be varied to enable good phase matching at different frequencies. The length of the waveguide used can be changed, for example, by inserting different tube sections. It is also possible to vary the distance between the mouth and the sound source by inserting two tubes into one another, which can be moved either by sliding against each other or by means of a mechanical guide.

[0092] If the wave fields are generated directly at or inside the mouth, for example, via a ferroelectric loudspeaker or a loudspeaker such as those used in mobile phones, the phase position of the waveforms used can be adjusted directly by the electrical excitation signal. This also applies if a waveguide that is shorter in relation to the wavelength is used.

[0093] The acoustic wave field should emerge from the area surrounding the loudspeaker or waveguide close to the mouth, as the coupling of the wave field with the air column within the vocal tract decreases rapidly with distance. Coupling within the acoustic near field of the sound source is advantageous but not absolutely necessary for the functioning of the training device. The tube can also be inserted into the mouth. In this case, the diameter of the tube is kept small enough so that it can be easily inserted into the mouth. The acoustic wave field can also be generated directly in the mouth. Care must be taken to use materials approved by food regulations. A sound source for generating the adaptation signal should be installed close to the mouth; this has the advantage that high proportions of sound velocity can interact with the sound velocity of the resonator at different frequencies.

[0094] The sound generator can be inserted into the mouth, directly into the mouth or the area between the lips, into the vestibule, or into the oral cavity, where it generates the wave field. The sound generator is ideally designed to radiate as much acoustic power as possible to / into the mouth, while minimizing the amount of power escaping into the environment. Miniaturized loudspeakers are particularly well-suited for this requirement.

[0095] The external acoustic wave field is made available to the user for resonance adjustment. For example, a sine tone of 800 Hz can be presented as a signal and the user adjusts the vocal tract in relation to this signal, i.e. the user amplifies the signal by adjusting the position of the mouth, pharynx, throat, neck, head, chest area and jaw. Alternatively, a targeted reduction in the amplitude of the presented frequency can be achieved by adjusting the vocal tract to the so-called anti-resonances. Sine tones can be helpful for this, as the resonance conditions are better defined. In addition or alternatively, specifying whether breathing also occurs through the mouth or only through the nose can increase the reproducibility of the vocal tract adjustment.

[0096] At the same time, flexible adjustment of the muscles / muscle groups, such as those necessary for voice development, is also important for physiologically healthy posture, particularly in the shoulder girdle, neck, and throat areas. Therefore, in addition to acoustic training, flexibility training can also be used to support and benefit sports medicine and / or rehabilitation or prevention of postural deformities.

[0097] Measuring the sound intensity directly in the mouth is particularly advantageous for the selective measurement of the signal amplified by resonance. This can be achieved using a miniaturized microphone with a suitable housing. However, a microphone located outside the mouth can also be used, which picks up the sound from the oral cavity via a waveguide. This waveguide is advantageously designed so that it can be easily inserted into the mouth. For this purpose, the diameter of the waveguide for picking up the acoustic field should generally not exceed 2 cm, but it is preferably less than 1 cm in diameter. If the acoustic excitation also occurs via a waveguide, the waveguide used to transmit the acoustic field to the microphone can be connected to the other waveguide to save space.

[0098] The use of a microphone in combination with a waveguide is in Fig. 15. The user (1) guides the waveguides (6) and (10) to or into the mouth (2). The waveguide for excitation can also advantageously be placed inside the mouth, and the waveguide to the microphone can also be placed outside the final correction term of the mouth opening. In this case, excitation occurs directly in the mouth, and the distance of the waveguide opening from the mouth should be taken into account when calculating the phase position of the resonance relationship of the vocal tract resonances. Thus, if the waveguide is inserted two cm into the mouth, the phase position of the stimulating signal relative to the mouth opening is still crucial, and the sound velocity should be particularly high two cm from the sound exit opening of the waveguide. In general, the highest possible sound velocity should be used.This can generally be achieved with a volume source, whereby an amplitude of the sound velocity at the location of the sound outlet of at least 30%, better 60% of the maximum amplitude of the sound velocity in the frequency range of the adaptation frequencies is advantageous.

[0099] The waveguides are advantageously designed to be flexible and are advantageously acoustically insulated from one another. Waveguide (6) guides the excitation signal generated by the loudspeaker (3) to the mouth, while waveguide (10) transmits the sound pressure to a microphone (4). The microphone is advantageously designed as a lavalier microphone that can be controlled directly by the microprocessor / computer (9). The microprocessor (computer, mobile device) (9) supplies the excitation signal for the loudspeaker. Additional amplification of the signal is generally not necessary, particularly when using a loudspeaker such as is used in mobile devices. The acoustic signal at the mouth is picked up by the microprocessor, which may also have a lock-in circuit or a phase-selective readout of the acoustic signal upstream or integrated (12). If necessary, the signal is electronically post-processed, filtered, and recorded.The lock-in amplifier or a phase-selective evaluation unit can also be integrated directly as a circuit in the computer / microprocessor. The reference signal can, for example, be obtained from one of the speaker channels (e.g. the left channel) of the computer, while the other channel (e.g. the right) is connected to the microphone. The reference signal can also be generated by a frequency generator integrated into the computer via software, which can be controlled by the software. The feedback can also be provided acoustically via a second speaker or via additional signals fed to the speaker outside the frequencies used for the feedback. Output via a screen (11), which can also be integrated into the tablet, makes it easier to use and adjust the vocal tract. At the same time, tips can be shown on the display as to which further improvements should be made.

[0100] In Fig. Figure 16 shows an arrangement using a microphone, such as is also used in devices for recording the voice field. Typically, a microphone (4) is placed on the user's head (1) at a defined distance from the mouth (2). The microphone can optionally be designed as a wireless microphone that transmits the signals to the computer (5) via a cable or transmitter (13). In the example shown here, the loudspeaker (3) is acoustically insulated and housed (8). The acoustic signal is guided to the mouth (2) via a short and flexible waveguide (6), which can also be acoustically insulated, or via a mouthpiece into the mouth.

[0101] Flexible adjustment of the vocal tract can also be supported by changing adaptation signals that drift or vary between different frequencies or amplitudes. For example, adjustment of the vocal tract at a frequency around 900 Hz as well as an adjustment around 700 Hz is supported. In this example, a sweep from 900 Hz to 700 Hz can be performed over a period of, for example, 20 seconds. This gives the user the opportunity to adapt the vocal tract to change the adaptation frequency and thus keep it flexible. The sweep should not be too fast to ensure that the user can still adapt. For training purposes, a rapid change between two or more frequencies can also be carried out, whereby the user adapts the vocal tract to the new frequency as quickly as possible.In this case, the frequencies used should remain constant for at least 0.1 s to allow adjustment of the vocal tract.

[0102] Since adapting the shape of the vocal tract to an adaptation signal is easier the higher the intensity of the applied wave field, varying the amplitude of the signals used to generate the wave field can also be helpful. If varying the amplitude is used, it is advantageous to vary the amplitude in an increasing or decreasing direction and / or by oscillating around an amplitude value. Varying the amplitude in the form of a step function can also be helpful.

[0103] Predefined electrical signals are provided for standardized use. They can be generated by a function generator or accessed by the user as electronically stored waveforms. The signals can be pre-prepared, for example, according to the voice categories soprano, alto, tenor, bass, or intermediate categories such as baritone or mezzo-soprano, or they can be tuned for use primarily by speakers.

[0104] In addition, the visualization of the signal recorded by the microphone on a monitor or mobile device can be enhanced through the use of colors. If a monitor is used, further information can be displayed for the user - such as the achieved bandwidth. In addition, instructions can be given for specific movements that need to be changed to optimize the vocal tract, such as opening the jaw, raising the soft palate or a nodding movement. Likewise, the influence of specific movements can be correlated with the change in the signal. This allows the user to learn which changes have the greatest influence on the signal or the feedback signal. These movements can then be used in further exercises. If these movements are recorded using 3D measuring technology, for example, they can be saved for further optimization and offered to the user again if necessary.

[0105] In particular, the method not only allows for real-time display of the frequency position, damping factor, bandwidth, and quality factor of resonances and / or antiresonances, but also for their evaluation as a time-varying quantity or as a function of frequency. Advantageously, the display can also be performed as maximum and / or minimum and / or time-averaged values ​​and / or as a function of time and / or frequency. One or more of these displays can be performed as a quasi-immediate evaluation during the measurement or subsequently.

[0106] Monitoring success during or after use of the device can be achieved by comparing the distribution of fundamental tone and overtones of the spoken or sung voice before and after use of the training device. An external microphone or that of a mobile device is used for this purpose. The overtones of the recorded signal are preferably filtered or displayed using a Fourier analysis or in a narrower acoustic spectral range. It is advantageous to record the analyses and microphone recordings for later review and to document vocal progress. The analyzed vocal signal can be accessed for direct comparison and to motivate the user, either alone or in comparison with spectra of other users. An example of a display of overtones before and after use of the device is shown in Fig. 6 can be seen.

[0107] The frequency or frequencies to which the user adapts the vocal tract have an impact on the speech or vocal sound or the sound of the wind instruments after the training phase. For example, in the case of vocal changes in an adult male, optimising the vocal tract to a frequency of 700 Hz can result in a more bass-like speech or vocal sound, while using a frequency of around 900 Hz gives the voice more of a tenor sound, i.e. setting the vocal tract to higher frequencies above 900 Hz disproportionately increases the overtone content of the voice. The specifications can also result from a calculation based on the response recorded via the microphone from the introduced acoustic wave field. However, they can also relate to the recording of a speech signal or a vocal signal or a so-calledghost voice or breathing noises before or after using the device. For example, a high resting voice indicates that you should start with frequencies that are not too low, but that the frequency should be reduced over the course of training. Based on the analysis of the vocal signal, the algorithm can also be designed to suggest new or additional exercises. For example, a user with a particularly bright voice will tend to be offered exercises with a frequency of around 1000 Hz or below, whereas a user with a voice that is poor in overtones will tend to be offered frequencies above 900 Hz. Exercises with a sweep or with step functions of the adaptation frequency or the adaptation frequency bands are also suggested. If the formants for vowel recognition are not particularly pronounced, two or more frequencies or bands can be used.Frequency bands are offered, recommended, or selected according to the vowels being addressed. For example, an adaptation signal composed of tones around 300 Hz and above 1500 Hz can improve the bright vowels e and i. These adaptation signals can also be offered one after the other, e.g., first the frequency around 300 Hz and, after sufficient amplification of this signal, also frequencies above 1500 Hz. If background noises, particularly noise components, are detected during voice analysis, which suggests little support for voice production by the vocal tract, then adaptation frequencies between 800 and 900 Hz, for example, are advantageous. Depending on the measured resonance behavior at these frequencies, the frequency range can be expanded upwards and / or downwards in further exercises.

[0108] Recording during phonation before and after adjusting the vocal tract allows the user to directly audibly and / or visually perceive the influence of the training device. The vocal tract can remain in the adjusted position, or spoken or sung sections can be saved and / or the various frequency ranges or overtone spectra can be displayed and / or visualized using Fourier analysis. This allows evaluation not only of the volume of the sung or spoken tone, but also of the energy distribution in different frequency ranges (e.g., by displaying the Hamarberg Index or other parameters characterizing a powerful voice), the ratio of fundamental tones to overtones of a sung tone, and / or fluctuations in pitch or intensity (jitter or shimmer).

[0109] The signal output can be very specifically achieved by comparing a reference signal with the measurement signal using the lock-in amplifier principle, whereby the latter can be operated with or without additional filters. The lock-in amplifier analyzes the signal for intensity and phase. When adjusting the vocal tract, both the intensity of the introduced acoustic signal and the phase are changed. Therefore, a two-phase lock-in amplifier is particularly suitable, as it can calculate the amplitude independently of the phase. This lock-in amplifier can also be integrated into a computer or mobile device as a data processing module or separate program. A phase-selective evaluation of the acoustic signal compared to the excitation signal is also possible by digitally comparing the signals from the microphone and the loudspeaker excitation using standard electrical engineering techniques.An output signal can be output via its phase-independent amplitude. For example, a frequency of 2 kHz is specified as the reference signal, while the sound introduced via a loudspeaker only contains a small portion of this frequency range. In this case, the feedback signal is output via the filter at exactly 2 kHz, acoustically and / or visually, and / or haptically via a vibration. The amplification of the signal's resonance properties by the introduced acoustic field can be made audible to the user, but also practically inaudible. The same applies to the use of filters instead of a lock-in amplifier.

[0110] Since the invention adjusts the vocal tract without phonation, an acoustic signal is clearly audible and generally particularly helpful. Therefore, the frequencies selected above can also be output acoustically. This allows the user to concentrate on amplifying the signal without using their own vocal cords for vocal production.

[0111] An example process is shown below. Pre-programmed processes are already stored on the computer. All information can be requested, for example, by a voice specialist or musician, or automatically via a device using a menu. This allows the user to be prompted to record a specific vocal activity, such as singing or speaking. This recording is captured by the microphone and can be saved and / or analyzed, for example, visualized in the form of a spectrogram.

[0112] An algorithm can be used to determine key parameters from the signal. These can include, in addition to the exercises described here, established vocal exercises. The subsequent training phase can last for varying lengths. Between exercises, further feedback can be provided on the acoustic properties of the vocal tract or on vocal development.

[0113] In general, a duration of 10 minutes at a time should not be exceeded; shorter intervals are more beneficial, after which the user can relax, do other vocal exercises and / or readjust the resonance formation. The resonance between the acoustic wave field and the vocal tract is audible, but even at lower intensities the signal can be used instantaneously as feedback after measurement, for example via a microphone. The rapid feedback provided by the amplification of the signal in relation to changes in the vocal tract is a great advantage, as even short-term positive adjustments - possibly discovered by chance - can be immediately picked up by the user. This is therefore a real-time signal that enables an improved adjustment of the vocal tract within just a few training cycles compared to the provided acoustic field.

[0114] All data and evaluations can be saved and documented so that they are available for long-term analysis.

[0115] A comparative display of recordings of vocal signals taken at different times can, for example, display and / or visualize the amplification of the vocal function at different frequency ranges, changes in speech intelligibility, and / or changes in volume. The device used for this purpose is preferably designed to allow the measured and / or derived data to be stored and further processed.

[0116] Different settings of the vocal tract are also possible for amplification according to a given adaptation signal. For example, a signal of 800 Hz can be emitted during breathing through the mouth and nose, or during breathing only through the nose, or only through the mouth with the mouth open. In addition or alternatively, the indication of whether breathing also occurs through the mouth or only through the nose can increase the reproducibility of the vocal tract setting. Composite frequencies consisting of two or more sine tones of different frequencies specify the desired resonance conditions of the vocal tract even more precisely as an adaptation signal. By simultaneously amplifying both adaptation signals, a defined setting of the geometry of the vocal tract is achieved or made reproducible. For example, an adaptation frequency in the range of 800 Hz can be combined with another at 1200 Hz.By actively amplifying both frequencies in a defined relationship to each other through the active adjustment of the vocal tract, the position is achieved even more precisely than when using a single frequency.

[0117] In Fig. Figure 18 shows a possible comparison for calibrating the measurements using a sample body (a tube closed at one end), in which the attenuation was adjusted using different amounts of absorbing material. The frequency was tuned using a sweep, and the amplitude and phase (not shown) of the resulting signal were measured. The half-width or quality factor can be determined from the change in the sample body's resonance. Fig.Figure 17 also shows the amplitude range in which a user achieved amplification by adjusting the vocal tract resonance while breathing freely through the mouth using the same device (indicated by the arrow at 460 Hz). From the comparison of the amplitudes, the quality factor or bandwidth of the vocal tract formant adjusted according to the adaptation signal can be determined. The same applies to the use of phase shift. 1.5 Advantages of the invention

[0118] The invention represents a way to generate an acoustic wave field through a control system that is capable of changing the geometry of hollow bodies or the vocal tract according to specified acoustic properties and simultaneously determining the acoustic properties traversed. The specification of specific frequencies or frequency bands, whose intensity is amplified or attenuated, enables a shift in the geometry of a hollow body or the vocal tract to achieve a cavity optimized for the acoustic specifications. Thus, the invention allows the reverse engineering approach to be applied to the adjustment of acoustic parameters.

[0119] Frequencies and / or frequency bands and / or overtones with defined phase positions or phase relationships to each other are presented to the vocal tract in the phase position corresponding to the resonance condition at the mouth. For this purpose, just one frequency or a few frequencies are sufficient. The phase position of the overtones when using multiple frequencies to each other is viewed according to the phase relationship of the formants or resonances of the vocal tract. Establishing specific phase relationships of the measurement signal simultaneously facilitates phase-related evaluations for determining the bandwidth or quality factor.

[0120] If resonances in particular are amplified, the effectiveness of the device and method can be directly correlated with the improvement in the quality factor of the hollow body at the corresponding frequencies. In the case of the vocal tract, this can address the formants and thus speech intelligibility. With a specific adjustment for anti-resonances, the method can achieve increased attenuation.

[0121] Another advantage is the placement of a microphone away from the sound output port of an acoustic waveguide. This reduces contamination or damage that can occur due to proximity to the openings of hollow bodies. Likewise, the evaluation of the temporal shift in the amplitude / phase relationship is a significant advantage, as it allows for the identification of interfering signals and their separation from the desired signal.

[0122] At the same time, a barrier between the outlet of the sound source and the opening of a sound sensor can improve the signal-to-noise ratio and measure the amplified resonating signal without any significant portion of the input signal. Since in this case the direct signal from a sound generator is reduced, amplification by a vocal tract resonance of one (resonance frequency of the vocal tract far from the frequency of an incoming wave field) is easier to measure. The evaluation becomes even more precise if a second microphone outside the immediate vicinity of the sound source records the joint signal, i.e. the amplified or attenuated signal of the vocal tract together with the incoming signal. In this case the signal from the second microphone can be used as a reference for the precise determination of the resonating signal of the vocal tract.For this purpose, the second microphone should not be placed further than 50 cm from the sound outlet due to additional interference signals, for example due to the existing room acoustics.

[0123] The adjustment of the phase position in the resonance condition for the opening of the hollow body is a further advantage of the invention. The individual phase positions can be specifically adapted to the resonance conditions using overtones.

[0124] This provides a method and an apparatus with which acoustic data from hollow bodies or the human vocal tract can be measured in a simple manner.

Claims

[1] A method for supporting the adjustment of acoustic properties of hollow bodies, in particular of a vocal tract and thus of the vocal sound, wherein an acoustic wave field consisting of several frequencies or frequency bands is generated by at least one loudspeaker and / or with at least one waveguide connected to a loudspeaker on or in the mouth, wherein the distance of the sound outlet opening to the mouth is less than 0.8 * radius of the respective mouth opening, wherein frequencies in the range of 200 to 10,000 Hz are used, wherein the wave field is amplified or attenuated by adjusting the geometry of the vocal tract of a user, wherein to adjust the geometry of the vocal tract of a user, at least three adaptation frequencies which are immediately fed back to the user during use and which correspond to at least three addressed vocal tract resonances are amplified or attenuated,wherein the resulting wave field is measured via at least one external microphone mounted outside the mouth or the waveguide and acoustically isolated from the sound source, the distance from the microphone to the mouth being greater than 0.8 * radius of the mouth opening. [2] Method according to claim 1 characterized by that the distance between the microphone and the sound outlet is kept constant during a measurement or when influencing the acoustic properties of the vocal tract or hollow body. [3] Method according to one of the preceding claims, characterized by that the temporal variation of phase and / or amplitude of one or more frequencies used is used to determine one or more resonance frequencies and / or one or more bandwidths of resonances and / or antiresonances of the vocal tract. [4] Method according to one of the preceding claims, characterized bythat the evaluation of the bandwidth of the resonance of a vocal tract is carried out on the basis of the amplitude and / or phase of the measured waveforms and / or from the change in amplitude and / or phase of the waveforms resulting from changes in the vocal tract geometry. [5] Method according to one of the preceding claims, characterized by that the wave field having at least one adaptation signal consists of frequencies and / or frequency bands which are temporally changed in their amplitude and / or frequency via an oscillation and / or a sweep and / or via a step function. [6] Method according to one of the preceding claims, characterized by that the adaptation signals used for amplification or attenuation are combined with a broadband measurement signal in the frequency range from 200 to 10,000 Hz. [7] Method according to one of the preceding claims, characterized bythat an evaluation of acoustic quantities such as frequency position, damping factor, bandwidth, quality factor of resonances and / or anti-resonances is carried out as maximum and / or minimum and / or time-averaged value or values ​​and / or as a function of time and / or frequency, whereby the representation is carried out as a quasi-immediate evaluation during the measurement and / or afterwards via acoustic and / or haptic and / or visual feedback. [8] Method according to one of the preceding claims, characterized by that the electrical signals used to control a loudspeaker are composed over a sequence of time segments, whereby the signals of the time segments are kept constant in their frequency and / or amplitude or change in their frequency and / or amplitude over time. [9] Method according to one of the preceding claims, characterized bythat before, during or after an amplification or attenuation of frequencies and / or frequency bands of a wave field, a vocal and / or noise-like signal generated during a breathing process is recorded by one or more microphones and / or further processed and / or compared with each other and / or presented individually. [10] Device for supporting the adjustment of acoustic properties of hollow bodies, in particular of a vocal tract characterized by that the method features of claims 1 to 8 are implemented in a technical device. [11] A computer program comprising instructions causing the apparatus of claim 9 to carry out the method steps of claims 1 to 8.

Images