WO2008001779A1 - procédé d'estimation de fréquence de référence et système d'estimation de signal acoustique - Google Patents

procédé d'estimation de fréquence de référence et système d'estimation de signal acoustique Download PDF

Info

Publication number
WO2008001779A1
WO2008001779A1 PCT/JP2007/062819 JP2007062819W WO2008001779A1 WO 2008001779 A1 WO2008001779 A1 WO 2008001779A1 JP 2007062819 W JP2007062819 W JP 2007062819W WO 2008001779 A1 WO2008001779 A1 WO 2008001779A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
fundamental frequency
frequency
autocorrelation function
comb filter
Prior art date
Application number
PCT/JP2007/062819
Other languages
English (en)
Japanese (ja)
Inventor
Yoshiaki Tadokoro
Masanori Natsui
Masahiro Ito
Original Assignee
National University Corporation Toyohashi University Of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by National University Corporation Toyohashi University Of Technology filed Critical National University Corporation Toyohashi University Of Technology
Priority to JP2008522593A priority Critical patent/JPWO2008001779A1/ja
Publication of WO2008001779A1 publication Critical patent/WO2008001779A1/fr

Links

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/90—Pitch determination of speech signals

Definitions

  • the present invention relates to a method of estimating the fundamental frequency of each signal when the fundamental frequency component of one signal overlaps the frequency component of another signal in a signal composed of the fundamental frequency and its multiple frequency.
  • the invention relates to a fundamental frequency estimation method capable of obtaining the lowest fundamental frequency from the minimum output value and the autocorrelation function, and an acoustic signal estimation system based thereon.
  • FIG. 1 shows an amplitude spectrum of a certain musical tone.
  • fp fundamental frequency
  • kfp, k 2, 3, ⁇ ⁇ ⁇
  • the pitch the pitch of the sound
  • Patent-related technology relating to the conventional transcription system can not cope with chords for which the above-described pitch estimation is difficult (see, for example, Patent Documents 1 to 3).
  • Method by genetic algorithm hereinafter GA
  • non-patent document 1 method assuming that the sum of amplitude spectra is established (for example, refer to non-patent document 2), music rules etc.
  • An estimation method see, for example, Non-Patent Document 3) and the like have been proposed.
  • the time waveform of each instrument needs to be stored in advance, which limits the scope of use of the technology.
  • Patent Document 1 Japanese Patent Application Laid-Open No. 2002-278544
  • Patent Document 2 Japanese Patent Application Laid-Open No. 5-100660
  • Patent Document 3 Japanese Patent Application Laid-Open No. 5-173557
  • Non-Patent Document 1 Ono, Saito, Ozawa, "Mixed sound estimation method using GA for automatic transcription", Transactions of the Society of Measurement and Automatic Control, Vol. 33, No. 5, pp. 417- 423, 1997
  • Non-patent document 2 Ueda, Hashimoto, "Blind decomposition algorithm for sound source separation", Transactions of Information Processing Society of Japan, Vol. 38, No. 1, pp. 146-157, 1997
  • Non-Patent Document 3 Chino, Kinoshita, Nakasu, Tanaka, “Analysis of roar sound in the processing model of music scene analysis OPTIMA”, Shingaku theory, Vol. J79-D-II, No. 11, pp. 1762 — 1770, 199
  • the present disclosure is formed by a composite signal whose fundamental frequency is difficult to estimate, for example, a composite signal formed by a plurality of signals having the same fundamental frequency, or a plurality of signals whose fundamental frequency is in a fixed ratio relationship.
  • a specific example of the synthetic signal targeted by the present disclosure is a sound (hereinafter referred to as a chord) generated by simultaneously sounding a sound generated by a musical instrument (hereinafter referred to as a musical tone).
  • a chord generated by simultaneously sounding a sound generated by a musical instrument (hereinafter referred to as a musical tone).
  • a musical tone generated by simultaneously sounding a sound generated by a musical instrument (hereinafter referred to as a musical tone).
  • the chord of the musical tone will be described in relation to the present invention, the effect of the present invention on the acoustic signal composed of the fundamental frequency and its multiple frequency is not limited.
  • the musical instrument can be regarded as a different signal source with regard to musical tones. It pays attention to the fact that it is impossible to That is, from the overlapping frequency components of chords whose pitch estimation is difficult, a slight frequency difference is detected as an output to determine differences in the signal source and the occurrence time, and the pitch estimation of the consonance is performed. Let it be possible.
  • the determination as to whether or not frequency components overlap is focused on the phase of each frequency component obtained from the discrete Fourier transform (hereinafter referred to as DFT) of the input signal, and It is also possible to distinguish the two clearly from the time change of the phase. That is, in a general single tone musical tone, the phase difference between the phase of the fundamental frequency and the phase of its overtone frequency is almost constant. However, in chords, the phase difference changes with time in components whose frequency components overlap. This occurs because there is a slight difference in frequency between overlapping frequency components. This makes it possible to distinguish overlapping frequency components. This feature is particularly useful in distinguishing between tones and chords.
  • DFT discrete Fourier transform
  • FIG. 1 is a diagram showing an amplitude spectrum of a musical tone.
  • Fig. 2 is a schematic view showing an overlap of frequency components of a chord that is difficult to estimate.
  • FIG. 4 is a conceptual diagram of a parallel configuration comb filter system for pitch estimation. (Hq, p (z).
  • FIG. 5 is a block diagram of an acoustic signal estimation system.
  • FIG. 6 is a flowchart showing the basic operation of the acoustic signal estimation system.
  • FIG. 7 It is a time waveform diagram of a target musical tone (clarinet C4).
  • FIG. 8 It is a time waveform diagram of the target tone (clarinet C4 + alt-sax C4).
  • FIG. 9 It is a time waveform diagram of the target tone (clarinet C4 + alt-sax C5).
  • FIG. 10 It is a time waveform diagram of a target musical tone (clarinet C4 + alt-sax G5).
  • FIG. 15 is a diagram showing the autocorrelation function value of the output of the comb filter system of FIG.
  • FIG. 16 is a diagram showing autocorrelation function values of the comb filter system output of FIG. 11;
  • FIG. 17 shows the autocorrelation function values of the output of the comb filter system of FIG.
  • FIG. 18 is a diagram showing autocorrelation function values of the comb filter system output of FIG. 13;
  • FIG. 21 is a block diagram of an acoustic signal estimation system using phase information of DFT.
  • FIG. 22 is a flow chart showing the basic operation of an acoustic signal estimation system using phase information of DFT.
  • FIG. 23 is a waveform diagram showing the time change of the phase of the target musical tone (Horn A3).
  • FIG. 24 is a waveform diagram showing the time change of the phase of the target musical tone (Horn A3 + Horn A3).
  • FIG. 25 A waveform diagram showing the time change of the phase of the target musical tone (Horn A3 + Horn A4).
  • FIG. 26 A waveform diagram showing the time change of the phase of the target musical tone (Horn A3 + Horn E5). BEST MODE FOR CARRYING OUT THE INVENTION
  • the acoustic signal estimation system includes an input unit 1 for inputting an acoustic signal, a comb filter system unit 2 having a comb filter having the amplitude characteristic shown in FIG. 3, an autocorrelation function for calculating an autocorrelation function.
  • the processing unit 3 comprises an acoustic signal determination unit 4 that determines the type of autocorrelation function, etc., and an output unit 5 that outputs an estimation result of the acoustic signal.
  • a fundamental frequency estimation method using a comb filter will be described. The fundamental frequency estimation method is performed by the fundamental frequency estimation system shown in FIG.
  • the symbol "[]" indicates that the number is converted to an integer by rounding to the nearest decimal point (the same applies hereinafter). Then, by inputting a chord into this system and detecting its minimum output value, it is possible to estimate the fundamental frequency (pitch) and name of the chord.
  • the octave of the note name can be determined from the output change, and the pitch of the chord is estimated. Have confirmed that.
  • the target chord is input to the corresponding comb filter.
  • the output should be zero in principle, but there is a slight frequency difference between the overlapping frequency components of the chord and a slight frequency difference because the tone is not a perfect stationary signal.
  • the autocorrelation function of the output signal is obtained. Then, from the period of the autocorrelation function, it can be determined whether the chord is a unison power, octave chord or k-harmonic chord.
  • unison means that the fundamental frequencies of the signals included in the chord are equal.
  • an octave chord means a chord in which the fundamental frequency ratio of the signal contained in the chord has a relationship of 1: (2 n ), where n is a positive integer.
  • the k-th harmonic means a chord having a fundamental frequency ratio of a signal included in a chord having a relationship of l: k (where k is a positive integer excluding 2 n ).
  • the delay number of the comb filter is changed and the self correlation function of the output is obtained. If the autocorrelation waveform does not change much, it will be judged as a single note, and if a change is seen, it will be judged as a chord.
  • the period by determining the autocorrelation function of the output waveform of the comb filter if the period is not clear, similarly change the number of delays of the comb filter to obtain the autocorrelation function of the output.
  • an acoustic signal whose fundamental frequency is difficult to estimate is input to the input unit 1 (Sl). Then, the input acoustic signal is passed to the comb filter system unit 2, and filtering by the comb filter is performed (S2).
  • the comb filter circuit is properly designed for signal frequency components. For example, for musical tones, it corresponds to the lowest fundamental frequency of the target pitch. Then, the comb filter cancels out the target frequency component and outputs a signal component that generates a difference between the fundamental frequency of the frequency component.
  • the autocorrelation function (R1) of the output of the comb filter system unit 2 is calculated, and the period of the output signal is determined (S3).
  • the type of chord of the acoustic signal is discriminated from the period of the autocorrelation function obtained by the autocorrelation function processing unit 3 (S4). Furthermore, it is determined whether or not the period of the autocorrelation function matches the period targeted by the comb filter (S5). If it is determined that they match (S5: YES), it is necessary to further distinguish whether the tone is a single note or a chord. Therefore, filtering is performed while changing the number of delays of the comb filter (S6), and the autocorrelation function (R2) of the output of the filtering is obtained (S7).
  • the waveform of the autocorrelation function R1 and the waveform of the autocorrelation function R2 do not change so much, it is determined as a single tone, and if a change is observed, it is determined as a chord (S8). Then, the judgment result is processed as a system output by the output unit 5 (S9).
  • the chord is a chord of an octave, or a k-th harmonic Since the chord is a chord, the discrimination result is processed as a system output by the output unit 5 (S9).
  • the output unit 5 When measuring the period by determining the autocorrelation function of the output waveform of the comb filter, if the period is not clear, change the number of delays of the comb filter in the same way, and calculate the autocorrelation function of the output.
  • FIGS. 7 to 10 respectively show time waveforms of the four target tones.
  • the horizontal axis represents time
  • the vertical axis represents amplitude.
  • the lowest pitches of these tones (that is, C4) are detected as the minimum output of the parallel configuration comb filter shown in FIG.
  • the approximate value affects the accuracy of the estimation process, but in the present invention, the approximate value here does not affect the estimation accuracy.
  • Np 169
  • the horizontal and vertical axes are time and amplitude, respectively.
  • Fig. 15 Force et al. 20 show the results of the autocorrelation function of the following equation (Equation 1), using the data of 2000 (about 45 ms) number of samples shown in Figs. It shows each. All five figures are expressed as autocorrelation function values (vertical axis) with respect to the number of delays (horizontal axis).
  • x (n) represents sample data of a musical tone.
  • Np is the number of samples substantially corresponding to the basic period of pitch C4, but a more accurate period can be determined by setting the maximum value of k to Np + a.
  • Figs. 15 and 16 the periods 168 and 169 are measured in number of samples, respectively. That Therefore, it can be seen that Fig. 7 and Fig. 8 are single tone (C4) or unison (C4 + C4).
  • Fig. 17 and Fig. 18 periods 85 (rounded value after rounding to 169/2) and 56 (rounded values after rounding to 169/3) are detected, and the input of Fig. 9 is an octave (C4 + C5) It can be seen that the input in Fig. 10 is the third harmonic (C4 + G5).
  • each musical tone is passed through the comb filter corresponding to the period measured in FIG. 15 and FIG. Find the autocorrelation function of.
  • the results are shown in Figure 19 and Figure 20.
  • FIG. 19 is almost the same as the self-correlation function of FIG. 15, and FIG. 7 is judged as a single tone.
  • the output of the parallel configuration comb filter is used to target the synthesized signal whose fundamental frequency component can not be separated, and to use the harmonic component and the autocorrelation function that are not removed by the comb filter. It was described in "The pitch estimation method for the transcription that paid attention to the value” (Morita Morita, Mitsuru Yamaguchi, Yoshiaki Tadokoro, No. J87-D- ⁇ , .. 12, pp. 2271-2279, 2004) There is an inventive step to the means.
  • each pitch of the consonant which has been considered to be difficult conventionally.
  • the frequency resolution can be increased even with a small number of samples, high quality acoustic signal processing can be performed.
  • the present invention can also be applied to acoustic signals including voices and the like that are sounded only with musical tones.
  • acoustic signals can be freely combined and used.
  • a method based on tone phase information will be described as an effective method of discriminating between single tones and chords.
  • This method can support the correctness of the result of the method of discriminating between single tones and chords described in [0017] and [0018]. That is, the correctness of the process in the sound signal discrimination unit 4 can be reconfirmed.
  • the DFT of the input tone is calculated, and the phase of each frequency component is determined. This DFT is calculated based on one cycle of the lowest fundamental frequency. Then, the DFT calculation is performed while shifting the time, and the phase difference between the phase of each frequency component and the phase of the fundamental frequency is plotted with respect to time. If the phase difference is almost constant with respect to time, it can be judged as a single tone, and if it changes, it can be judged as a chord. The time change of this phase difference occurs because there is a slight difference in frequency between overlapping frequency components.
  • phase difference changes even with a single sound.
  • the phase change becomes larger as the higher frequency components. Therefore, it is possible to distinguish between single notes and chords by judging whether or not the regularity of phase change is maintained in these instruments.
  • the acoustic signal discrimination system includes an input unit 11 for inputting an acoustic signal, a DFT calculation unit 12 for calculating a DFT of the acoustic signal, and a phase calculation unit for calculating a phase from the calculation result of DFT.
  • the phase difference between the fundamental frequency and its overtone frequency component is calculated and plotted against time.
  • the phase calculation unit 14 determines the superimposed frequency component from the time change of the phase difference, and determines the type of acoustic signal.
  • a signal discrimination unit 15 comprises an output unit 16 for outputting the discrimination result of the acoustic signal.
  • an acoustic signal whose fundamental frequency is difficult to estimate is input to the input unit 11 (Sl 1). Then, the input acoustic signal is delivered to the DFT calculation unit 12. The DFT calculation unit 12 calculates a DFT having the lowest fundamental frequency of the input acoustic signal as a basic component (S12).
  • phase calculation unit 13 the phase of each frequency component is calculated from the ratio of the real part and the imaginary part of the DFT coefficient (S13). And, in time 'phase calculation section 14, the fundamental frequency of DFT The phase difference between the number of phases and the phase of its overtone frequency component is determined (S14). Furthermore, the phase change with respect to time is plotted (S15). Next, in the acoustic signal discrimination unit 15, it is determined whether or not the discrimination of the superimposed frequency component is possible from the change of the phase difference with respect to time (S16). Here, if the determination can not be made (S16: NO), the range of the DFT calculation is shifted (S17), and the process returns to S12 to calculate the DFT of the acoustic signal again.
  • the superimposed frequency component S16: YES
  • the type of the acoustic signal is determined (S18).
  • the determination result is processed as a system output by the output unit 16 (S19).
  • FIGS. 23 to 26 are plots of the phase differences between the fundamental frequency components of the four musical tones of interest and their overtone frequency components with respect to time.
  • the phase of each frequency component is substantially constant, and this musical tone is determined to be a single tone.
  • the phase of one of the frequency components changes with time, and these musical tones are determined to be chords. These phase changes occur due to the frequency difference between the overlapping frequency components.
  • Fig. 24 it can be seen that all the frequency components are changing, and this chord is unison.
  • FIG. 25 the phases of even-numbered frequency components are changed, and it can be seen that this musical tone is an octave tone.
  • the phase of the frequency component of the third harmonic is changed, and it can be seen that this musical tone is the third harmonic.
  • the present invention is not limited to the above embodiment, and various modifications are possible.
  • the acoustic signal estimation system has been described as an apparatus including each processing unit.
  • the processing executed by the comb filter system unit 2, the autocorrelation function processing unit 3 and the acoustic signal determination unit 4 is a computer Let's run it in software using the CPU.
  • the processing executed by the DFT calculation unit 12, the phase calculation unit 13, the time / phase calculation unit 14, and the sound signal determination unit 15 is It may be executed as software using the computer's CPU.
  • the present invention can be used to enhance efficiency and efficiency in a wide range of applications such as ethnic music and improvisation recording, composition work and music education.
  • the performance concerning secrecy is determined by the accuracy of pitch estimation.
  • a full-fledged acoustic signal estimation system can be applied as a highly accurate estimation technology even to devices that have large technical specifications.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Auxiliary Devices For Music (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)

Abstract

L'invention concerne un procédé d'estimation de fréquence de référence qui fait attention à un signal de synthèse formé par des composants de fréquence de référence inséparables et au fait qu'un cycle de signal puisse être mesuré à partir d'une micro-sortie d'un filtre en forme de peigne qui a été traité comme une erreur. Le procédé peut estimer chaque fréquence de référence des composants à régler. D'abord, un signal de synthèse d'objet est traité en utilisant un filtre en forme de peigne correspondant à la fréquence de référence la plus faible du signal de synthèse et une fonction d'autocorrélation de la sortie est utilisée pour mesurer le cycle du signal, obtenant de ce fait chaque référence du signal de synthèse. En outre, afin d'assurer une résolution temporelle, en utilisant le filtre en forme de peigne et la fonction d'autocorrélation, le procédé peut également être appliqué à un échantillon de temps court. Ainsi, il est possible de réaliser un traitement avec une résolution de fréquence très précise sans entraîner un nombre insuffisant d'échantillons.
PCT/JP2007/062819 2006-06-27 2007-06-26 procédé d'estimation de fréquence de référence et système d'estimation de signal acoustique WO2008001779A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2008522593A JPWO2008001779A1 (ja) 2006-06-27 2007-06-26 基本周波数推定法および音響信号推定システム

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
JP2006-176615 2006-06-27
JP2006176615 2006-06-27
JP2006-356148 2006-12-28
JP2006356148 2006-12-28

Publications (1)

Publication Number Publication Date
WO2008001779A1 true WO2008001779A1 (fr) 2008-01-03

Family

ID=38845543

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2007/062819 WO2008001779A1 (fr) 2006-06-27 2007-06-26 procédé d'estimation de fréquence de référence et système d'estimation de signal acoustique

Country Status (2)

Country Link
JP (1) JPWO2008001779A1 (fr)
WO (1) WO2008001779A1 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2015001586A (ja) * 2013-06-14 2015-01-05 ブラザー工業株式会社 弦楽器演奏評価装置及び弦楽器演奏評価プログラム
JP2015001587A (ja) * 2013-06-14 2015-01-05 ブラザー工業株式会社 弦楽器演奏評価装置及び弦楽器演奏評価プログラム
CN114088973A (zh) * 2021-11-20 2022-02-25 吉林大学 一种基于双psd数字锁相放大器的超声波测风系统及方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS62203199A (ja) * 1986-03-03 1987-09-07 富士通株式会社 ピツチ周期抽出方式
JPH10124094A (ja) * 1996-10-18 1998-05-15 Sony Corp 音声分析方法、音声符号化方法および装置
JP2001249676A (ja) * 2000-03-06 2001-09-14 Japan Science & Technology Corp 雑音が付加された周期波形の基本周期あるいは基本周波数の抽出方法
JP2003108200A (ja) * 2001-09-27 2003-04-11 Kenwood Corp 音声信号雑音除去装置、音声信号雑音除去方法及びプログラム
JP2003280696A (ja) * 2002-03-19 2003-10-02 Matsushita Electric Ind Co Ltd 音声強調装置及び音声強調方法
JP2005195975A (ja) * 2004-01-08 2005-07-21 Nippon Telegr & Teleph Corp <Ntt> 音声信号分析方法およびこの分析方法を実施する装置、この音声信号分析装置を用いた音声認識装置、この分析方法を実行するプログラムおよびその記憶媒体

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS62203199A (ja) * 1986-03-03 1987-09-07 富士通株式会社 ピツチ周期抽出方式
JPH10124094A (ja) * 1996-10-18 1998-05-15 Sony Corp 音声分析方法、音声符号化方法および装置
JP2001249676A (ja) * 2000-03-06 2001-09-14 Japan Science & Technology Corp 雑音が付加された周期波形の基本周期あるいは基本周波数の抽出方法
JP2003108200A (ja) * 2001-09-27 2003-04-11 Kenwood Corp 音声信号雑音除去装置、音声信号雑音除去方法及びプログラム
JP2003280696A (ja) * 2002-03-19 2003-10-02 Matsushita Electric Ind Co Ltd 音声強調装置及び音声強調方法
JP2005195975A (ja) * 2004-01-08 2005-07-21 Nippon Telegr & Teleph Corp <Ntt> 音声信号分析方法およびこの分析方法を実施する装置、この音声信号分析装置を用いた音声認識装置、この分析方法を実行するプログラムおよびその記憶媒体

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
NISHI K. ET AL.: "Shinpuku Joho to Pitch Hendo o Doji ni Koryo shita Giji Shuki Shingo Chushutsu no Tame no IIR Kushigata Digital Filter", IEICE TECHNICAL REPORT, September 1999 (1999-09-01), pages 29 - 34, XP003019674 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2015001586A (ja) * 2013-06-14 2015-01-05 ブラザー工業株式会社 弦楽器演奏評価装置及び弦楽器演奏評価プログラム
JP2015001587A (ja) * 2013-06-14 2015-01-05 ブラザー工業株式会社 弦楽器演奏評価装置及び弦楽器演奏評価プログラム
CN114088973A (zh) * 2021-11-20 2022-02-25 吉林大学 一种基于双psd数字锁相放大器的超声波测风系统及方法
CN114088973B (zh) * 2021-11-20 2024-04-02 吉林大学 一种基于双psd数字锁相放大器的超声波测风系统及方法

Also Published As

Publication number Publication date
JPWO2008001779A1 (ja) 2009-11-26

Similar Documents

Publication Publication Date Title
Bittner et al. Deep Salience Representations for F0 Estimation in Polyphonic Music.
US6930236B2 (en) Apparatus for analyzing music using sounds of instruments
JP4767691B2 (ja) テンポ検出装置、コード名検出装置及びプログラム
Klapuri et al. Robust multipitch estimation for the analysis and manipulation of polyphonic musical signals
WO2007010637A1 (fr) Détecteur de rythme, détecteur de nom de corde et programme
Traube et al. Estimating the plucking point on a guitar string
Argenti et al. Automatic transcription of polyphonic music based on the constant-Q bispectral analysis
Lenssen et al. An introduction to fourier analysis with applications to music
McLeod Fast, accurate pitch detection tools for music analysis
Virtanen Audio signal modeling with sinusoids plus noise
CN108292499A (zh) 技巧确定装置和记录介质
WO2008001779A1 (fr) procédé d&#39;estimation de fréquence de référence et système d&#39;estimation de signal acoustique
Goto A predominant-f0 estimation method for real-world musical audio signals: MAP estimation for incorporating prior knowledge about f0s and tone models
Azarov et al. Guslar: a framework for automated singing voice correction
Woodruff et al. Resolving overlapping harmonics for monaural musical sound separation using pitch and common amplitude modulation
Coyle et al. Onset detection using comb filters
Stöter et al. Unison Source Separation.
Shalom et al. Optimal filtering of an instrument sound in a mixed recording using harmonic model and score alignment
Beauchamp Comparison of vocal and violin vibrato with relationship to the source/filter model
Bhagwat et al. Enhanced audio source separation and musical component analysis
JPH1011066A (ja) 和音抽出装置
JP3501199B2 (ja) 音響信号分離方法
Rychlicki-Kicior et al. Multipitch estimation using judge-based model
JP2011164497A (ja) テンポ値検出装置およびテンポ値検出方法
Yamaguchi et al. Pitch Estimation for Mixed Tones With Overlapped Harmonic Components Using Comb Filters and Auto-correlation Functions

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 07767625

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2008522593

Country of ref document: JP

NENP Non-entry into the national phase

Ref country code: DE

NENP Non-entry into the national phase

Ref country code: RU

122 Ep: pct application non-entry in european phase

Ref document number: 07767625

Country of ref document: EP

Kind code of ref document: A1