JP2023007982A - Player fingering detection system for woodwind instrument - Google Patents

Abstract

To detect the fingering of a player with a device that can be easily attached/detached without degrading the operational feeling of an acoustic woodwind instrument that the user himself/herself always uses.SOLUTION: A mouthpiece having a built-in speaker is attached to an acoustic woodwind instrument. The speaker built in the mouthpiece reproduces a sound having certain frequency characteristics irrespective of time. The sound reflected in the acoustic woodwind instrument is recorded with a microphone. Fingering of a player is estimated based on a signal recorded with the microphone.SELECTED DRAWING: Figure 10

Description

Patent Act Article 30, Paragraph 2 application filed CHI2021 Proceedings, Association for Computing Machinery (publisher name), Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems (05/2020, Publication name) date of issue), https://dl. acm. org/doi/abs/10.1145/3411763.3451719 (publication address) [Publications, etc.] CHI2021 Video Preview, Association for Computing Machinery (publisher name), April 23, 2021 (publication date), https:/ /www. youtube. com/watch? v=spMeqQ4Y20U (published address)

The present invention relates to fingering detection of a player of a woodwind instrument.

Regarding the fingering detection of a woodwind player, Patent Document 1 states that "a plurality of key switches 3a to 3p are arranged on the front side of the main body 1" and that "the key switches 3a to 3p for specifying pitches are arranged." The ROM 56 (FIG. 8) stores in advance a correspondence table showing the correspondence between ON/OFF combinations by pressing and note numbers according to fingering chart rules." Further, Patent Document 2 states that "a small magnet 13 is fixed to the arm portion 11a of each operation element 11, and a flexible substrate 14 is attached to a portion of the peripheral surface of the musical instrument body 10 facing the finger operation portion 12. A Hall sensor 15 is provided on each portion of the flexible substrate 14 that faces the magnet 13. When the manipulator 11 is pressed and the magnet 13 approaches the Hall sensor 15, A Hall sensor 15 detects the magnetism of the magnet 13. This detects which operator 11 has been operated."

Japanese Patent Laid-Open No. 11-85159 JP 2009-258750

Laura A. Stambaugh, "An Examination of a MIDI Wind Controller for Use in Instrumental Research", Proceedings of the International Symposium on Performance Science, 2011 Ministry of the Environment, “Environmental standards for noise”, [online], March 30, 2012, [searched May 4, 2020], Internet <https://www.env.go.jp/ kijun/oto1-1.html＞ Nelson Posse Lago, and one others, "The Quest for Low Latency", PROCEEDINGS OF THE INTERNATIONAL COMPUTER MUSIC CONFERENCE, 2004

Many acoustic woodwind instruments are loud when played, making it difficult to practice playing indoors. One way to practice acoustic woodwind instruments indoors is to use a wind controller, a music controller that mimics an acoustic woodwind instrument. The performer blows into the wind controller and presses a tone hole or key-like switch, and the synthesizer outputs based on the information obtained from the wind controller, including volume, pronunciation timing, note value and scale. By listening to the synthesized sound through headphones, you can practice playing at a low volume. The wind controller is equipped with fingering detection means to determine the scale of the synthesized sound. Here, fingering is a combination of keys pressed by a player or tone holes closed with a finger for an acoustic woodwind instrument, or a combination of keys or switches pressed by a player for a wind synthesizer. Concerns that arise when an existing wind controller is used for performance practice will be described below. The concern relates to existing means of fingering detection in wind controllers.

The means for detecting fingering described in Patent Document 1 detects fingering using a switch imitating a key and having a return spring and a pressure-sensitive sensor attached thereto. However, there is a concern that playing practice using a wind controller that uses such a structure will not sufficiently and efficiently improve a player's ability to play an acoustic woodwind instrument, because the feeling of operation that the player feels is different from that of an acoustic woodwind instrument. be. It has been reported that, in some cases, when a player uses a wind controller with a switch that mimics a key and has a return spring, the player's playing errors are increased compared to acoustic woodwind instruments. (Non-Patent Document 1).

In addition, in Patent Document 2, fingering is detected by attaching a Hall effect sensor to each key of an acoustic woodwind instrument to constitute a wind controller. In this case, the player can obtain the original feeling of operating the acoustic woodwind instrument, but there is a concern that practice using this wind controller will not sufficiently improve the performance ability of the player when using the instrument that the player regularly uses. This is because acoustic woodwind instruments require different amounts of force to press each key, even if they are of the same model. This is because there is a difference in operational feeling between In addition, if the instrument used by the player is used as a wind controller by attaching the sensor to the acoustic woodwind instrument that is commonly used by the player for the purpose of obtaining a playing feeling, it is assumed that the time burden imposed on the player will be large. . This is because, after using an acoustic woodwind instrument that is commonly used by a performer as a wind controller, when the performer performs a performance that emits a normal acoustic voice, if the front sensor is removed due to concerns that it will affect the tone and aesthetics, This is because it is necessary to re-install the pre-sensor in order to use the instrument that is used regularly as a wind controller. When installing the sensors, the player needs to install as many sensors and conductors as there are keys so as not to hinder the playing operation.

In view of the above, it is an object of the present invention to detect a player's fingerings using an easily attachable/detachable device without impairing the player's sense of operation of his/her usual acoustic woodwind instrument.

A mouthpiece with a built-in speaker is attached to an acoustic woodwind instrument, the speaker built into the mouthpiece reproduces sound with constant frequency characteristics regardless of time, and the first sound reflected inside the acoustic woodwind instrument is recorded with a microphone. , to estimate the fingering of the performer based on the signals recorded by the microphone;

To detect a player's fingering by using an easily attachable/detachable device without impairing the player's operational feeling of an acoustic woodwind instrument that he or she usually uses.

2 is a block diagram of the fingering detection system described in Example 1. FIG. 4 illustrates the mounting position of the speaker described in Example 1. FIG. 5 is a flowchart showing an example of processing contents of a fingering estimation unit described in Embodiment 1; 4 is a diagram showing the relationship between the signal x acquired by the microphone according to the first embodiment, the sampling time Δt, and the cycle ΔT for fingering estimation. FIG. FIG. 2 is a device diagram relating to the fingering detection system described in Example 1; FIG. 11 is a block diagram of the fingering detection system according to the second embodiment; FIG. 4 illustrates the mounting positions of the air pressure sensor and members described in Example 2. FIG. FIG. 10 is a diagram showing an example of time transition of an air pressure sensor output voltage value and a fingering number acquired by the performance information control unit described in Example 2; FIG. 10 is a device diagram of the fingering detection system described in Example 2; 10 illustrates a configuration when a player uses the fingering detection system described in Example 2. FIG. 3 illustrates the mounting position of the sound insulation partition described in Example 2. FIG.

An embodiment will be described below with reference to the drawings.

In this embodiment, a case where the present invention is applied to fingering detection of a player on an acoustic woodwind instrument will be described.

In this embodiment, acoustic woodwinds are actual woodwinds as distinguished from wind synthesizers. Here, an acoustic woodwind instrument amplifies the air vibration generated by the vibration of the reed by the player's exhalation inside the tube to produce a sound. , is an instrument that changes the scale of the sound produced. The player opens and closes the tone holes by pressing the keys attached to the side of the instrument or by covering the tone holes with the pads of the fingers. At this time, the combination of pressed keys or blocked tone holes is called fingering. Fingering detection is to output fingering information, and the fingering information is, for example, a fingering number assigned to each fingering of the fingering to be detected. Fingering numbers are natural numbers. be.

FIG. 1 is a block diagram showing the configuration of a fingering detection system 100 according to the first embodiment.

The fingering detection system 100 shown in FIG.

The utterance control unit 101 determines a signal to be uttered by the speaker 102 , amplifies the signal, and outputs the amplified signal to the speaker 102 . The first signal is, for example, an audio file read in advance from a ROM (not shown in FIG. 1) and analog-converted by an A/D converter (not shown in FIG. 1). A sound file is characterized in that the frequency characteristics of the sound emitted from the speaker 102 are constant with respect to time. The audio file is, for example, series data in which values sampled from a Gaussian distribution are arranged, and the series data is expected to be white noise whose intensity is the same at all frequencies and whose characteristics do not change with time. A voice file is stored in advance in a ROM (not shown in FIG. 1).

Speaker 102 generates sound based on the pre-amplified signal. The speaker 102 is characterized by being installed inside the acoustic woodwind instrument 103 . This is because the frequency characteristics of the signal recorded by the microphone 104 must well reflect the frequency characteristics of the acoustic woodwind instrument 103 determined by fingering. Furthermore, in order to detect all fingerings to be detected, the speaker 102 is placed in the shortest effective pipe of the acoustic woodwind instrument 103 corresponding to all fingerings to be detected. must be attached. The mounting position of the speaker 102 is, for example, inside the mouthpiece. FIG. 2 shows an example of the internal structure of a saxophone mouthpiece in which the speaker 102 is attached. In FIG. 2, the speaker 102 is mounted inside the mouthpiece so as to effectively vibrate the air inside the acoustic woodwind instrument 103 . For example, when the speaker is a circular speaker unit composed of a coil, a magnet, and a diaphragm, it is fitted into the mouthpiece with the direction in which the diaphragm is exposed facing the tubular body. Also, the speaker 102 is characterized in that it can be easily attached to and detached from the acoustic woodwind instrument 103 . For this reason, for example, the speaker 102 is mounted inside a mouthpiece that is different from the mouthpiece that the player normally uses during playing. In this case, the player can easily attach and detach the speaker 102 by exchanging the mouthpiece.

The acoustic woodwind instrument 103 amplifies or attenuates the intensity corresponding to some frequencies of the sound emitted by the speaker in the instrument, and changes the frequency characteristics of air vibration inside and outside the tube. Since the acoustic woodwind instrument 103 changes the initial frequency characteristic according to the fingering of the player, the changed frequency characteristic is expected to differ for each fingering. The acoustic woodwind instrument 103 is, for example, a saxophone.

The microphone 104 converts air vibrations around the microphone 104 into electrical signals and outputs the electrical signals. The microphone 104 may be attached at a position where it can observe air vibrations with changed frequency characteristics, for example, attached to the bell at the end of the acoustic woodwind instrument 103 using a spring clip.

A fingering estimation unit 105 acquires an electrical signal from the microphone 104 and outputs fingering information based on the electrical signal. The fingering estimation unit 105 executes the processing shown in FIG. 3, for example, in order to output the player's fingering information at time t.

Step 301: Acquire the parameter vector of the fingering estimation model used for fingering estimation from the ROM (not shown in FIG. 1), and proceed to step 302 . The fingering estimation model is a function that receives a signal obtained from the microphone 104 and outputs fingering information. For example, the fingering estimation model receives a power spectrum density vector calculated by applying a short-time Fourier transform to a digital signal obtained by digitally converting a signal for Δt seconds observed by the microphone 104 as an input, and calculates the fingering number. The output is a multinomial logistic regression model. The fingering estimation model should estimate as accurately as possible the fingering numbers that the output fingering numbers correspond to the player's fingerings from time t−Δt seconds to time t seconds. For this reason, for example, model parameters are learned in advance by maximum likelihood estimation using a set of pairs of fingering numbers corresponding to the power spectrum density vector of the digital signal and the fingering of the performer at the time the digital signal was acquired. and stored in a ROM (not shown in FIG. 1). Δt is the sampling time required for one fingering estimation. Δt is a value sufficiently smaller than the note value used in general music in order to follow changes in the fingering of the performer with a sufficiently small delay, and is sufficient to obtain sufficient frequency resolution for fingering estimation. need to be big. Δt is, for example, 30 milliseconds.
Step 302: Set the current time to τ, set t to τ+Δt, and proceed to step 303 .
Step 303: Wait until τ=t, and proceed to step 304; During this time, the analog signal acquired from the microphone 104 is converted into a digital signal by an A/D converter (not shown in FIG. 1) at a frame rate of f times/second and written to a RAM (not shown in FIG. 1).
Step 304: Acquire the digital signal x from time t−Δt seconds to time t seconds from the RAM, and proceed to step 305 . x is a vector of length fΔt where f is the sampling rate.
Step 305: Convert x into a complex frequency vector using short-time Fourier transform, and obtain a power spectral density vector X obtained by squaring the magnitude of each element of the complex frequency vector, and proceed to step 306;
Step 306: Input X into the fingering estimation model to obtain and output the fingering number, and proceed to step 308;
Step 308 : Substitute t+ΔT for t and proceed to step 303 . Here, ΔT is the cycle of fingering estimation. ΔT needs to be set to a value sufficiently smaller than the note value used in general music in order to follow changes in the fingering of the performer with a sufficiently small delay. ΔT is, for example, 5 milliseconds. Step 3 is executed multiple times every ΔT seconds. The relationship between ΔT and Δt versus a given number of iterations N and different x in step 304 for the next iteration N+1 is illustrated in FIG. If ΔT<Δt, as shown in FIG. 4, two adjacent x's contain the same signal in their portion.

The fingering detection system 100 has the hardware configuration shown in FIG. 5, for example. The fingering detection system 100 in FIG. The ROM 505 is a non-volatile memory that stores an audio file used by the utterance control unit 101 and a parameter vector of a fingering estimation model used by the fingering estimation unit 105 . A CPU 506 is used for calculations of the utterance control unit 101 and the fingering estimation unit 105, and a RAM 504 is a volatile memory used during the previous calculation. A bus 503 connects the A/D converter 502 , RAM 504 , ROM 505 and CPU 506 . The A/D converter 502 converts the voltage obtained from the microphone 104 into a digital value, and the digital signal based on the audio file obtained from the bus 503 into an analog signal. The amplifier 501 amplifies the analog signal and outputs it to the speaker 102 .

According to the present invention, the fingering detection system uses an easily attachable/detachable device to detect the player's fingering without impairing the player's sense of operation of the acoustic woodwind instrument that he or she usually uses. .

Although the amplifier 501 is connected to the bus 503 to which the microphone 104 is connected in FIG. 5 of this embodiment, the amplifier 501 and the speaker 102 are electrically and physically independent of the bus 503 to which the microphone 104 is connected. It may be configured within the device. In this case, the independent device comprises an amplifier 501, a speaker 102, an A/D converter, a bus, a ROM, a RAM and a CPU, and the A/D converter, ROM, RAM and CPU are connected to the bus to which the microphone 104 is connected. They are different from the A/D converter 502, ROM 505, RAM 504, and CPU 506 connected to 503, respectively.

In addition, in this embodiment, the speaker 102 is attached inside the mouthpiece, but it should be attached in the shortest effective duct of the acoustic woodwind instrument 103 corresponding to all the fingerings to be detected. Just do it. For example, speaker 102 may be mounted inside the neck of a saxophone.
Also, although the acoustic woodwind instrument 103 is a saxophone in this embodiment, it may be any actual woodwind instrument that is distinguished from a wind synthesizer, such as a clarinet, an oboe, or a flute.

In the present embodiment, the fingering estimation unit 105 uses a multinomial logistic regression model that outputs a fingering number to a power spectrum density vector as an input. You can use a function that For example, the fingering estimator 105 may use an ordinal logistic regression to estimate the fingering numbers with the power spectrum density vector as an input, or may use a linear regression model to estimate the scale corresponding to each fingering with a digital signal as an input. , or a non-learning rule-based model, such as a function that outputs the maximum value of the power spectral density vector, may be used to estimate the frequencies that are particularly amplified. When the estimated value is the frequency using a linear regression model or the like, the fingering number corresponding to the scale having the frequency closest to the estimated frequency is output.

In this embodiment, the fingerings to be detected are all fingerings corresponding to the sound range of the acoustic woodwind instrument, but the fingerings to be detected may be a part of these fingerings or similar fingerings. Fingerings may be treated as identical fingerings. An example of handling similar fingerings as identical fingerings is a case where fingering pairs that cannot be identified with high accuracy by the fingering estimation unit 105 are handled as identical fingerings. The first fingering pair is, for example, two saxophone fingerings that differ only in the presence or absence of octave keys. However, in this case, information on whether or not the octave key is pressed is separately required for determining the fingering information. For this reason, for example, the fingering detection system 100 has a switch that is turned on only while the octave key is pressed, and the fingering estimation unit 105 detects the fingering that is estimated based on the state of the switch. can be determined whether it contains The early switch is, for example, a button switch attached between the octave key and the instrument tube. Such a single switch attachment does not compromise the ease of attaching and detaching the present invention to an acoustic woodwind instrument.

In this embodiment, the fingering detection system 100 of the first embodiment further includes a vibration suppression unit, an expiration detection unit, and a performance information control unit to determine and output performance information, and further includes a synthesizer and headphones. This is the case where voice is synthesized based on performance information and fed back to the performer. Here, the performance information is information used by the synthesizer to determine the synthesized sound signal. For example, the performance information is data in MIDI format, such as "C5, sound on, volume 50".

In this embodiment, the fingering detection system 100 includes a vibration suppression unit 106, an expiration detection unit 107, a performance information control unit 108, a synthesizer 109, and headphones 110 shown in FIG. 6 in addition to the configuration shown in the first embodiment. The configuration added in this embodiment will be described below.

The breath detector 107 and the vibration suppressor 106 will be described with reference to FIG. The mouthpiece shown in FIG. 7 has an air pressure sensor 703, an exhalation hole 701 and a member 704 therein. Also, the tip of the member 704 is called a member tip 705 .

The breath detector 107 acquires and outputs breath information for the purpose of determining the pronunciation timing, note value, and volume included in the performance information. For example, the breath detector 107 and breath information are the air pressure sensor 703 and the air pressure sensor output voltage value, respectively. For example, the air pressure sensor 703 is mounted inside the mouthpiece as shown in FIG. The atmospheric pressure sensor 703 changes the magnitude of the output voltage value according to the atmospheric pressure inside the mouthpiece. Also, the player's breath is discharged through the breath discharge hole 701 . The exhalation hole 701 is designed to be sufficiently small so that the air pressure inside the mouthpiece becomes higher than the atmospheric pressure when the player blows into the mouthpiece.

The vibration suppression unit 106 prevents the air inside the acoustic woodwind instrument 103 from vibrating due to blowing in by the player, thereby generating performance sounds, affecting signals observed by the microphone 104, and generating loud vocalizations. It is a structure mounted on the mouthpiece for the purpose of preventing The vibration suppressing portion 106 imitates, for example, the shape of a natural reed, and has a sufficiently large curvature in the cross-sectional contour of the member distal end 705 with respect to the plane defined by the first and second directions shown in FIG. A member 704 is a member that is attached to the mouthpiece. In this case, it is possible to make the feel of the player holding the mouthpiece closer to the feel of a normal mouthpiece while preventing the player from blowing into the mouthpiece to produce a performance sound.

The performance information control section 108 uses the breath information obtained from the breath detection section 107 and the fingering information obtained from the fingering estimation section 105 to output performance information. FIG. 8 shows an example of temporal transition of the atmospheric pressure sensor output voltage value and the fingering number when the breath information is the atmospheric pressure sensor output voltage value and the fingering information is the fingering number. An example of a method for determining performance information by the performance information control section 108 will be described below with reference to FIG. In FIG. 8, it is assumed that the performance information control unit 108 simultaneously acquires the atmospheric pressure sensor output voltage value and the fingering information every ΔT seconds, and outputs the performance information in synchronization therewith. Performance information control section 108 determines and outputs performance information at time 1001, time 1002, time 1003, time 1004 and time 1005 as follows. It is assumed that the performance information before time 1001 is "voice off, volume 0".

Time 1001: Since the performance information before time 1001 includes "voice off" and the pressure sensor output voltage value at time 1001 is smaller than the sound generation threshold value 1006, it is determined that the sound is muted, and "voice off, volume 0". to output
Time 1002: Since the atmospheric pressure sensor output voltage value at time 1002 is greater than the sound generation threshold value 1006, it is determined that sound is being generated. Sound on, volume 50". The sound volume is given by, for example, the product of the air pressure sensor output voltage value at the output time of the performance information and a scale factor stored in advance in the RAM.
Time 1003: The immediately preceding performance information includes sound ON, the air pressure sensor output voltage value is greater than the value at time 1002 and the mute threshold 1007, and the fingering number at time 1003 is "12" corresponding to A4#. Therefore, "A4#, sound on, volume 60" is output. This corresponds to the operation of increasing the volume after pronunciation.
Time 1004: The immediately preceding performance information includes sound on, the air pressure sensor output voltage value is greater than the value at time 1003 and the mute threshold 1007, and the fingering number at time 1004 is "11" corresponding to A4. Therefore, output like "A4, sound on, volume 70". This corresponds to the operation of changing the scale while increasing the volume.
Time 1005: Since the immediately preceding performance information includes sound on, the air pressure sensor output voltage value is smaller than the mute threshold value 1007, and the fingering number at time 1005 is 11 corresponding to A4, "A4, sound off , volume 0”. This corresponds to an operation to stop sounding.

A synthesizer 109 generates and outputs synthesized sound based on a series of performance information acquired from the performance information control unit 108 . Synthetic sounds are analog signals, for example, software sound sources generated from pre-recorded acoustic woodwind sounds.

Headphones 110 are worn near the ears of the performer, and convert synthesized sounds obtained from the synthesizer 109 and amplified into air vibrations.

In this embodiment, the fingering detection system 100 includes the hardware configuration of FIG. 5 shown in Embodiment 1, and an atmospheric pressure sensor 703, output I/F 507, synthesizer 109, and headphones 110 shown in FIG. An analog signal acquired by the atmospheric pressure sensor 703 is converted into a digital signal by the A/D converter 502 in the same manner as the microphone 104 . Also, the output I/F 507 is an interface for outputting performance information to the synthesizer 109, and is, for example, MIDI-OUT. Also, the synthesizer 109 is connected to the output I/F 507 and the headphone 110 is connected to the synthesizer 109 . When the A/D converter 502, the bus 503, the RAM 504, the ROM 505, the CPU 506 and the output I/F 507 are stored in the computing device 900 as shown in FIG. Become.

According to this configuration, the fingering detection system 100 can output a series of performance information including the scale, pronunciation timing, volume, and note value, and feed back synthesized sounds to the performer.

In the present embodiment, the environment in which the present invention is used is not limited, but when the present invention is used for practice playing a woodwind instrument in a house, the volume of the sound produced by the speaker 102 is equal to the volume of the surroundings of the instrument. Do not exceed acceptable levels for your environment. The previous permissible level is 60 decibels described in Non-Patent Document 2, for example. As a result, the sound volume generated when using the present invention is lower than that of an actual performance, and it is expected that the occurrence of noise problems when performing an actual performance at home can be suppressed.

In this embodiment, the vibration suppressing portion 106 is a member 704 that imitates the shape of a natural reed, has a sufficiently large curvature in the cross section of the member tip 705, and is attached to the mouthpiece instead of the natural reed. Any structure may be used as long as it prevents the air inside the acoustic woodwind instrument 103 from vibrating due to the breath of the player blowing into it. For example, the vibration suppression part 106 may be a member having the same shape as the natural reed and having a sufficiently high elastic modulus, which is attached instead of the natural reed. It may be a structure that physically separates it from the space inside the instrument tube.

Also, in this embodiment, the acoustic woodwind instrument is the saxophone, but the acoustic woodwind instrument may be a flute, piccolo, or other instrument with air reeds. In this case, the vibration suppression unit 106 physically separates the space through which the mouthpiece communicates with the space inside the instrument tube so that air vibrations generated when the player blows on the mouthpiece do not excite air vibrations in the tube. Structure.

Also, in this embodiment, when the present invention is used for practicing the acoustic woodwind instrument 103, in order to obtain a performance feeling closer to that of the acoustic woodwind instrument 103, it is necessary to adjust the timing from the player's performance operation until the player approaches the synthesized sound. delay should be sufficiently short. When considering this, Δt should generally be set smaller than the allowed delay. Assuming that the generally allowable delay is 30 milliseconds described in Non-Patent Document 3, Δt is 30 when considering the processing time of the fingering estimation unit 105 and the processing time of the synthesizer 109 after fingering detection. Must be set to less than milliseconds. For example, Δt is set to 25 milliseconds.

The mouthpiece may also include a sound insulation partition 702 between the air pressure sensor 703 and the speaker 102 as shown in FIG. As a result, even when the sound of the performance is suppressed by the vibration suppression unit 106, noise generated by the player's exhalation can be prevented from being transmitted through the pipe of the acoustic woodwind instrument 103 and affecting the sound collected by the microphone 104. can. The sound insulating partition 702 is, for example, a circular sound absorbing sponge, and is adhered to the inside of the mouthpiece so as not to form a gap with the inner wall of the mouthpiece.

DESCRIPTION OF SYMBOLS 100... Fingering detection system, 103... Acoustic woodwind instrument, 104... Microphone, 101... Vocalization control part, 102... Speaker, 105... Fingering estimation part, 501... Amplifier, 502... A/D conversion part, 503... Bus, 504...RAM, 505...ROM, 506...CPU, 106...Vibration suppression unit, 107...Expiration detection unit, 108...Performance information control unit, 109...Synthesizer, 110...Headphones, 703...Air pressure sensor, 701...Expiration hole, 702 Sound insulation partition wall 704 Member 705 Tip of member 507 Output I/F 900 Computing device 1006 Sound generation threshold 1007 Silence threshold

Claims (4)

an acoustic woodwind instrument, a speaker mounted inside the acoustic woodwind instrument and continuously emitting sound characterized by constant frequency characteristics regardless of time, a microphone positioned near the acoustic woodwind instrument, and A fingering detection system for estimating a player's fingering on said acoustic woodwind instrument from moment to moment based on a signal acquired from a microphone. 2. The fingering detection according to claim 1, wherein said fingering is estimated by outputting a fingering number corresponding to each fingering with a power spectrum density vector calculated from a signal obtained from said microphone as an input. system. The sound volume emitted by the speaker is 60 decibels or less in the vicinity of the instrument, and the acoustic woodwind instrument is provided with a vibration suppressing section that prevents excitation of air vibration inside the acoustic woodwind instrument due to blowing of a player's breath. The fingering detection system according to any one of claims 1 and 2. Fingering detection according to any one of claims 1 to 3, wherein the length of the signal obtained from the microphone and used to estimate the fingering once is less than 30 milliseconds. system.

Images