ARCHIVES OF ACOUSTICS

29, 1, 1–21 (2004)

HIGH ACCURACY AND OCTAVE ERROR IMMUNE PITCH DETECTION

ALGORITHMS

M. DZIUBIŃSKI and B. KOSTEK

Multimedia Systems Department
Gdańsk University of Technology
Narutowicza 11/12, 80-952 Gdańsk, Poland
e-mail: kido@sound.eti.pg.gda.pl

The aim of this paper is to present a method improving pitch estimation accuracy, show-
ing high performance for both synthetic harmonic signals and musical instrument sounds.
This method employs an Artificial Neural Network of a feed-forward type. In addition, octave
error optimized pitch detection algorithm, based on spectral analysis is introduced. The pro-
posed algorithm is very effective for signals with strong harmonic, as well as nearly sinusoidal
contents. Experiments were performed on a variety of musical instrument sounds and sample
results exemplifying main issues of both engineered algorithms are shown.

1. Introduction

There are two major difficulties, namely, octave errors and pitch estimation accu-
racy [1–3], that most pitch detection algorithms (PDAs) have to deal with. Octave errors
problems, seems to be present in all pitch tracking algorithms, known so far, however,
these errors are caused by different input signal properties in the estimation process.
In time- domain based algorithms [4–7], i.e., AMDF, modified AMDF [8–10] or nor-
malized cross correlation (NCC) [3, 7, 11], octave errors may be caused by low energy
content of odd harmonics. In some cases AMDF or autocorrelation methods are per-
formed first and in addition some information is gathered from calculated spectrum, in
order to decrease the possibility of estimation errors [12, 13], resulting in more accurate
pitch tracking. Such operations usually require increased computational cost, and larger
block sizes, than PDAs working in the time-domain. In the frequency domain, errors are
caused mostly by low energy content of the lower order harmonics. In cepstral [2], as
well as in autocorrelation of log spectrum (ACOLS) [14] analyses, problems are caused
by high energy content in higher frequency parts of the signal. Some algorithms oper-
ate directly on time-frequency representation, and are based on analysing trajectories of
sinusoidal components in spectrogram (sonogram) of the signal [15, 16]. On the other
2 M. DZIUBIŃSKI and B. KOSTEK

hand, estimation accuracy problem for all mentioned domains is caused by a number of
samples representing analyzed peaks related to fundamental frequency.
There is an additional problem related to pitch detection. For example, in case of
speech signals [1, 17–20], it is very important to determine pitch almost instantaneously,
which means that processed frames of the signal must be small. This is because voiced
fragments of speech may be very short, with rapidly varying pitch. In case of musi-
cal signals, voiced (pitched) fragments are relatively long and pitch fluctuations lower.
This property of musical signals enables the use of larger segments of the signal in the
pitch estimation procedure. But for both application domains, efficient pitch detection
algorithm should estimate pitch periods accurately and smoothly between successive
frames, and produce pitch contour that has high resolution in the time-domain.

2. Spectrum peak analysis algorithm

The proposed pitch detection algorithm, a so-called Spectrum Peak Analysis (SPA),
is based on analyzing peaks in the frequency domain, representing harmonics of a pro-
cessed signal. The general concept is based on such relatively easiness of pitch determi-
nation by observing signal spectrum and especially intervals between partials that are
present in the spectrum. This is independent of the fact that some harmonics may be
absent, or they can be partially obscured by the background noise. It should, however,
be assumed that they are greater than the energy of the background noise. Estimating
pitch contour is performed in block processing, i.e., the signal is divided into blocks
with widths depending on pitch estimated for preceding blocks, whereas overlap can be
time-varying. The width of the first block is initialized to 4096 samples and is decreased
for successive blocks, if the detected pitch is relatively high, and can be represented by
lower spectrum resolution. Similarly, if estimated pitch decreases in consecutive blocks,
the block width is increased, to provide satisfying spectrum resolution. Each block is
weighted by the Hann window.

2.1. Harmonic peak frequency estimation

The first step of the estimation process, performed in each block, is finding one peak
that represents any of the signal harmonics. The largest maximum of the spectrum sig-
nal is assumed to be one of harmonics, and it is easy to establish its coordinates in terms
of frequency. The chosen peak is assumed to be at the M -th harmonic of the signal. In
practical experiments M = 20 seemed to satisfy all tested sounds, however, setting M
to any reasonable value is possible. The natural limitation of this approach is the spec-
trum resolution. It is assumed that the minimum distance d between peaks representing
neighboring harmonics must be four samples. Therefore, if detected maximum index is
smaller than M · d, M is automatically decreased by the algorithm to satisfy the for-
mulated condition. In some cases, for low frequency signals, block size in the analysis
must be suitably large to perform pitch tracking. The next step is calculating M possible
 HIGH ACCURACY AND OCTAVE ERROR . . . 3

fundamental frequencies, assuming that a chosen harmonic (the largest maximum of the
spectrum signal) can be 1,2, . . . , or M -th harmonic of the analyzed sound:
M
X FM
Ffund [i] = (1)
i
i=1

where:
Ffund – vector of possible fundamental frequencies,
FM – frequency of the chosen (largest) harmonic.
The main concept of the engineered algorithm is testing the set of K harmonics
related to vector Ffund , that are most likely to be peaks representing pitch. The value of
K is limited by FM as follows:
µ ¶
Fs
K = floor (2)
M
where:
floor (x) – returns the largest integer value not greater than x,
Fs – sampling frequency.
Based on M, Ffund vector and K, the matrix of frequencies used in analysis can be
formed in the following way:
M X
X K
F AM (i, j) = Ffund [i] · j (3)
i=1 j=1

where:
FAM – matrix containing frequencies of M harmonics sets.
If M is significantly larger than K, and most energy carrying harmonics are higher
order harmonics (the energy of first K harmonics is significantly smaller than, for ex-
ample, K, K + 1,. . . , 2 · K, or higher order harmonics), it is better to choose a set
of K consecutive harmonics representing the largest amount of energy. Therefore, fre-
quency of the first harmonic in each set (each row of FAM) does not have to represent
the fundamental frequency. Starting frequencies of chosen sets can be calculated in the
following way:
K
X
Hmaxset [j] = EH(i+j)·Ffund , j = 0, ..., L − 1 (4)
i=1

where:
Hmaxset – vector containing energy of consecutive K harmonics for the chosen set,
where Hmaxset [k] is the sum of K harmonics energies for the following frequencies:
k · Ffund , (k + 1) · Ffund ,. . . , (k + K) · Ffund , EHfund – energy of the harmonic with
Fs
frequency equal to f , L – dimension of Hmaxset vector: L = floor( Ffund − K).
4 M. DZIUBIŃSKI and B. KOSTEK

Starting frequency of each set is based on the index representing the maximum value of
Hmaxset: Fstart [m] = indmax [m] · Ffund [m] for m = 1, . . . , M .
Finally, modified FAM can be formed in the following way:
M X
X K
F AM (i, j) = Fstart [i] + Ffund [i] · (j − 1) (5)
i=1 j=1

2.2. Harmonic peak analysis

Each harmonics set, represented by frequencies contained in each row of FAM is
analyzed in order to evaluate whether it is most likely to be a set of peaks related to
fundamental frequency among the remaining M − 1 sets. This likelihood is represented
by V , while V is calculated for each set in the following way:
K
X
V = Hv [i] (6)
i=1

where:
Hv [i] – value of a spectrum component for i-th frequency for the analyzed set.
If the analyzed spectrum component is not a local maximum – left and right neighboring
samples are not smaller than the one assigned to the local maximum, then it is set at
0. Additionally, if local maxima of neighboring regions of spectrum are found, Hv is
decreased – values of the maxima found are subtracted fromHv .
Neighboring regions of the spectrum surrounding the frequency FHv , representing
Hv , are limited by the following frequencies:
Ffund
FL = FHv − (7a)
2
Ffund
FR = FHv + (7b)
2

where:
FL , FR – frequency boundaries of spectrum regions surrounding FHv ,
Ffund – assumed fundamental frequency of the analyzed set.
The fundamental frequency, related to the largest V , is assumed to be the desired
pitch of the analyzed signal. As observed from Figs. 1–3, three situations are possible.
For example, in Fig. 1, one can see that the analyzed spectrum peak value is not a local
maximum, therefore it is set at 0. In addition, local maxima are detected in surrounding
regions, which subtracted from Hv give negative values. It is clear that in this situation,
it is highly unlikely that Hv is a harmonic. Figure 2 presents a situation in which Hv is
a local maximum, and surrounding maxima have small values, opposite to Fig. 3, where
analyzed regions contain large local maxima. Therefore Fig. 2 represents a peak that is
most likely to be a harmonic.
 HIGH ACCURACY AND OCTAVE ERROR . . . 5

Fig. 1. Analysis of a possible harmonic peak and its surrounding region (analyzed fundamental frequency
is not related to peak frequency).

Fig. 2. Analysis of a possible harmonic peak and its surrounding region (analyzed fundamental frequency
is correctly related to peak frequency).
6 M. DZIUBIŃSKI and B. KOSTEK

Fig. 3. Analysis of a possible harmonic peak and its surrounding region (analyzed fundamental frequency
is two times larger than pitch).

3. Pitch estimation algorithm accuracy

Since the spectrum peak representing pitch is sampled with limited resolution, inter-
polation is required to improve the algorithm accuracy. Different linear methods have
been tested in order to find computationally efficient and suitable interpolation tech-
niques, however, estimating pitch based on a discrete spectrum is not a trivial task.
Problems are caused by other frequency components surrounding peak, related to pitch.
In practice, those disturbances are caused by spectral leakage of sinusoidal components
of a signal (higher order harmonics), and depend on frequency distance between those
components and their energy. Therefore, using simple interpolation methods, such as
polynomials or splines, would result in a limited performance. Artificial Neural Net-
works (ANN) seem to be suitable for this task, and are successfully used to improve
estimation accuracy, which is shown in the following sections.

3.1. Artificial Neural Network training

Three samples representing spectrum peak related to fundamental frequency have
been considered as the ANN input. Index values representing a peak have been normal-
ized to –1, 0 and 1, while 0 was treated as the index of peak maximum and indices –1
and 1 were assumed to be indices of the maximum neighboring samples. Synthetic har-
 HIGH ACCURACY AND OCTAVE ERROR . . . 7

monic signals were generated to obtain the training input data and target signal. Each
training signal was synthesized according to the following formula:
K
X 2πniFpitch R[n]
S[n] = sin( )· (8)
Fs i
i=1

where:
R – vector containing uniformly distributed (on the (0, 1) interval) pseudo-random
numbers.
Fpitch – fundamental frequency of the synthesized signal,
Fs – sampling frequency,
K – number of harmonics contained in the signal S. K is defined as follows:
floor(Fs /Fpitch ).
It can be observed that a synthetic signal is most likely to have harmonics with
decreasing energies, similar to musical instrument sounds. Three training processes
were performed, employing various window sizes (different lengths of training signals):
1024, 2048 and 4096 samples, while sampling frequency was equal to 44100. Each sig-
nal was weighted by the Hann window, this was because the Hann window was also
used in the SPA estimation process. A great number of synthetic signals were generated
to obtain training data for each window size, while fundamental frequencies were ran-
domly chosen from Fmin to 4500 Hz. Fmin is the lowest possible frequency in respect of
d, depending on the window size. The neural network used in the training process was
a feed-forward, back-propagation structure with three layers. First layer contained three
neurons, the hidden layer – four neurons and the output layer – one neuron. Hyperbolic
tangent sigmoid transfer function was chosen to activate the first two layers, whilst the
linear identity function was used to activate the last layer. Weights and biases, during the
training process, were updated according to Levenberg–Marquardt optimization [21].
Trained network was used in the estimation process, resulting in performance presented
in the following section.

3.2. Improved estimation accuracy performance

Pitch estimation accuracy has been tested on synthetic signals, generated accord-
ing to Eq. 8. Since pitch fluctuations of acoustic sounds can be much greater than the
maximum error of the estimation process, using synthetic signals was necessary. The
estimation error was calculated in connection with the formulae:
(fstop − fstart )(n − 1)
f [n] = + fstart , n = 1, ..., N (9)
N −1
|f [n] − P DA(Sf [n] )|
EP DA (f [n]) = 100% (10)
f [n]
where:
N – number of test frequencies,
8 M. DZIUBIŃSKI and B. KOSTEK

f – vector containing test frequencies,

fstart , fstop – starting and stopping frequencies of f ,
Sf [n] – test signal with f [n] pitch.
The proposed SPA algorithm, and in addition, NCC [3] and CA [2] algorithms were
implemented in the Matlab environment to analyze and compare their performance.
Table 1 presents the exemplary average estimation error for the implemented PDAs.
Pitch estimations were performed for block size equal to 2048 samples. In addition,
improvements of the estimation accuracy for SPA (2-nd order polynomial interpolation
and ANN interpolation) are presented, showing the highest performance of the Neural
Network-based approach. The average error is understood to be the arithmetic mean
of estimation errors calculated in respect of Eq. (10), where fstart = 50 Hz, fstop =
3000 Hz and N = 1000, while signals had lengths equal to 2048 samples.

Table 1. Average pitch estimation error.

SPA SPA SPA

PDA: NCC CA
(not optimized) (polynomial) (neural network)
Pitch est. error: 3.3399% 2.8518% 0.2808% 0.00122% 0.00000405%

Figures 4–8 presented estimation errors for all tested signals concerning each al-
gorithm, showing error fluctuations over frequency changes. It can be observed that
time-domain related algorithms show a decrease in accuracy of estimation when the
signal frequency increases, as opposed to frequency-domain related algorithms, where
the situation is the opposite.

Fig. 4. Pitch estimation error of the NCC algorithm.

 HIGH ACCURACY AND OCTAVE ERROR . . . 9

Fig. 5. Pitch estimation error of the CA algorithm.

Fig. 6. Pitch estimation error of the SPA algorithm (not optimized).

10 M. DZIUBIŃSKI and B. KOSTEK

Fig. 7. Pitch estimation error of the SPA algorithm (2nd order polynomial interpolation).

Fig. 8. Pitch estimation error of the SPA algorithm (ANN-based interpolation).

 HIGH ACCURACY AND OCTAVE ERROR . . . 11

Figure 4 presents performance of the NCC algorithm, showing an increase in errors

from 0.2% for the lowest frequencies to 6% for frequencies around 3000 Hz. Figure 5
presents performance of the CA algorithm. It can be observed that in this case also,
error changes in frequencies are similar, however, fluctuations are more significant for
frequencies over 1500 Hz. Figures 6–8 present the behavior of the SPA algorithm. Fig-
ure 6 shows estimation accuracy for the engineered algorithm without interpolating har-
monic peak (i.e. frequency of the maximum value of the peak represents fundamental
frequency), resulting in error equal to 5.8% for the lowest frequencies, and decreasing
to 0.1% for frequencies around 3000 Hz. Figure 7 presents the improved performance
of the algorithm by employing 2nd order polynomial interpolation. This results in er-
rors of 0.027% for the lowest frequencies, decreasing to 0.007% for frequencies around
3000 Hz. Figure 8 shows performance of the ANN-based interpolation of the harmonic
peak. The estimated error is equal to 0.0005% for the lowest frequencies decreasing to
0.00000013% for frequencies around 3000 Hz.

4. Time domain pitch contour correction

In some cases, transients of analyzed instrument sounds, contain only or almost
only odd harmonics, therefore pitch, calculated in short terms for transient parts, can be
perceived as one octave higher than pitch calculated for blocks representing steady state
of the sound. The human brain seems to ignore this fact, and for a listener the perceived
pitch of the whole sound is in accordance with that of the steady-state. However, blocks
containing transient, duplicated in time domain, result in sound with pitch perceived as
one octave higher. This observation calls for post-processing [5], i.e., time domain pitch
contour correction. Optimizing pitch tracks is relatively easy, since such problems are
only encountered for transient parts of musical sounds and in the majority of cases pitch
contour represents the expected (perceived) fundamental frequency. In Fig. 9 one can
observe that for an oboe, for one block in the transient phase, the estimated pitch is one
octave higher than that estimated for the steady-state, however, the overall pitch was
recognized correctly.

5. Experiments and results

In order to determine the efficiency of presented SPA, 412 musical instrument sounds
were tested. Analyses of six instruments in their full scale, representing diverse groups,
and one instrument with all articulation types, were carried out. Recordings of tested
sounds were made in the Multimedia Systems Department of the Faculty of Electron-
ics, Telecommunications and Informatics, of Gdańsk University of Technology, Poland
[10]. Tables (Tabs. 2–4) and figures (Figs. 10–18) present estimated average pitch, note
played by the instrument according to ASA standard, and the nominal frequency of the
note, as specified by the ASA. Results for oboe for three types of articulations: non
legato, staccato and portato are presented in Tables. 2–4. Results for other instruments,
dynamics and articulations are presented in Figs. 10–18.
12 M. DZIUBIŃSKI and B. KOSTEK

Table 2. Pitch estimation results for oboe (articulation: non legato, dynamics: mezzo forte).

Tone Estimated Nominal Octave

(ASA) pitch [Hz] freq. [Hz] error
A3# 234.24 233.08 NO
B3 245.46 246.94 NO
C4 263.22 261.63 NO
C4# 279.8 277.18 NO
D4 295.94 293.66 NO
D4# 314.52 311.13 NO
E4 332.35 329.63 NO
F4 351.04 349.23 NO
F4# 371.95 369.99 NO
G4 394.19 392 NO
G4# 417.42 415.3 NO
A4 442.4 440 NO
A4# 471.37 466.16 NO
B4 498.13 493.88 NO
C5 528.85 523.25 NO
C5# 563.3 554.37 NO
D5 597.98 587.33 NO
D5# 632.25 622.25 NO
E5 669.99 659.26 NO
F5 708.24 698.46 NO
F5# 755.94 739.99 NO
G5 799.07 783.99 NO
G5# 842.1 830.61 NO
A5 888.01 880 NO
A5# 936.42 932.33 NO
B5 997.3 987.77 NO
C6 1052.2 1046.5 NO
C6# 1124.5 1108.7 NO
D6 1185.5 1174.7 NO
D6# 1272.8 1244.5 NO
E6 1326.3 1318.5 NO
F6 1407.1 1396.9 NO
F6# 1502.1 1480 NO
 HIGH ACCURACY AND OCTAVE ERROR . . . 13

Table 3. Pitch estimation results for oboe (articulation: portato, dynamics:mezzo forte).

Tone Estimated Nominal Octave

(ASA) pitch [Hz] freq. [Hz] error
A3# 234.98 233.08 NO
B3 246.48 246.94 NO
C4 263.76 261.63 NO
C4# 279.92 277.18 NO
D4 296.12 293.66 NO
D4# 313.06 311.13 NO
E4 332.96 329.63 NO
F4 352.04 349.23 NO
F4# 373.6 369.99 NO
G4 396.97 392 NO
G4# 422.38 415.3 NO
A4 447.6 440 NO
A4# 472.71 466.16 NO
B4 500.22 493.88 NO
C5 530.36 523.25 NO
C5# 564.36 554.37 NO
D5 594.88 587.33 NO
D5# 631.44 622.25 NO
E5 668.94 659.26 NO
F5 706.49 698.46 NO
F5# 753.12 739.99 NO
G5 798.01 783.99 NO
G5# 846.62 830.61 NO
A5 896.12 880 NO
A5# 947.18 932.33 NO
B5 1005.1 987.77 NO
C6 1058.4 1046.5 NO
C6# 1131.3 1108.7 NO
D6 1202.9 1174.7 NO
D6# 1296.1 1244.5 NO
E6 1435.5 1318.5 NO
14 M. DZIUBIŃSKI and B. KOSTEK

Table 4. Pitch estimation results for oboe (articulation: double staccato, dynamics:mezzo forte).

Tone Estimated Nominal Octave

(ASA) pitch [Hz] freq. [Hz] error
A3# 234.39 233.08 NO
B3 245.85 246.94 NO
C4 264.03 261.63 NO
C4# 279.62 277.18 NO
D4 294.79 293.66 NO
D4# 313.57 311.13 NO
E4 331.49 329.63 NO
F4 351.44 349.23 NO
F4# 374.11 369.99 NO
G4 396.23 392 NO
G4# 422.11 415.3 NO
A4 442.7 440 NO
A4# 471.08 466.16 NO
B4 498.67 493.88 NO
C5 532.04 523.25 NO
C5# 568.96 554.37 NO
D5 598.54 587.33 NO
D5# 635.64 622.25 NO
E5 669.15 659.26 NO
F5 708.18 698.46 NO
F5# 754.25 739.99 NO
G5 798.97 783.99 NO
G5# 850.02 830.61 NO
A5 898.31 880 NO
A5# 941 932.33 NO
B5 1007.5 987.77 NO
C6 1068.4 1046.5 NO
C6# 1152.3 1108.7 NO
D6 1217.7 1174.7 NO
D6# 1298.3 1244.5 NO
E6 1425.6 1318.5 NO
F6 1494.5 1396.9 NO
 HIGH ACCURACY AND OCTAVE ERROR . . . 15

Fig. 9. Octave fluctuations of pitch in transient of oboe (non legato).

Fig. 10. Pitch estimation results for baritone saxophone (articulation: non legato, dynamics: forte,
range: C2# – A4).
16 M. DZIUBIŃSKI and B. KOSTEK

Fig. 11. Pitch estimation results for bassoon (articulation: non legato, dynamics: forte, range:
A1# – C5).

Fig. 12. Pitch estimation results for trumpet (articulation: non legato, dynamics: forte, range:
E3 - G5#).
 HIGH ACCURACY AND OCTAVE ERROR . . . 17

Fig. 13. Pitch estimation results for tuba F (articulation: non legato, dynamics: forte, range: F1 - C4#).

Fig. 14. Pitch estimation results for viola (articulation: non legato, dynamics: forte, range: C3 - A6).
18 M. DZIUBIŃSKI and B. KOSTEK

Fig. 15. Pitch estimation results for oboe (articulation: non legato, dynamics: forte, range: A3# - F6).

Fig. 16. Pitch estimation results for oboe (articulation: non legato, dynamics: piano, range: A3# - F6# ).
 HIGH ACCURACY AND OCTAVE ERROR . . . 19

Fig. 17. Pitch estimation results for oboe (articulation: vibrato, dynamics: mezzo forte, range: A3# – F6).

Fig. 18. Pitch estimation results for oboe (articulation: single staccato, dynamics: mezzo forte, range:
A3# – G6).
20 M. DZIUBIŃSKI and B. KOSTEK

As seen from tables and figures presented, no octave related errors were detected
by the engineered algorithm. Different articulations and dynamics of sounds seemed
not to affect the octave error estimation accuracy of the SPA. Differences, sometimes
significant, between estimated pitch and tone frequency arise as the result of musicians
playing solo. Moreover, instruments were not tuned to exactly the same pitch before the
recordings.

6. Conclusion

The proposed algorithms have been tested on a variety of sounds with differentiated
articulations and dynamics, showing high resistance to octave errors (octave error was
not detected among all tested sounds). In addition, there is no limitation to harmonic
sounds in the analysis (while periodicity has to be maintained), which is the case with
other algorithms, such as, for example, CA and ACOLS algorithms. Moreover, energy
of harmonics does not have to be concentrated around a fundamental frequency, which is
an important issue for both: NCC and AMDF algorithms. The main disadvantage of the
SPA presented is its limited frequency range for small window sizes (lower boundary).
On the other hand, the NCC algorithm has an extended lower frequency limit. However,
in case of fast pitch fluctuations of low pitched sounds, the overlap can be decreased
significantly, while keeping large window sizes and resolution of calculated pitch track
may be preserved.
In addition, presented algorithm accuracy optimization seems to be very effective,
resulting in very precise pitch estimation. An optimized SPA algorithm gives far more
precise results than classic PDAs, these characteristics may be useful in sound separa-
tion and parameterization processes.

Acknowledgment

The research is sponsored by the Committee for Scientific Research, Warsaw, Grant
No. 4T11D 014 22, and by the Foundation for Polish Science, Poland.

References

[1] W. H ESS, Pitch determination of speech signal processing, Springer-Verlag, New York 1983.
[2] A. M. N OLL, Cepstrum pitch determination, J. Acoust. Soc. Am., 14, 293–309 (1967).
[3] L. R. R ABINER, On the use of autocorrelation analysis for pitch detection, IEEE Trans. on ASSP,
25, 24–33 (1977).
[4] X. Q UIAN , R. K IMARESAN, A variable frame pitch estimator and test results, IEEE Int. Conf. On
Acoustics, Speech, and Signal Processing, 1, Atlanta GA, 228–231, May (1996).
[5] D. TALKIN, A robust algorithm for pitch tracking (RAPT), Speech Coding And Synthesis, pp. 495-
518, Elsevier, 1995.
 HIGH ACCURACY AND OCTAVE ERROR . . . 21

[6] G. S. Y ING , L. H. JAMIESON , C. D. M ICHELL, A probabilistic approach to AMDF pitch detection,

http://purcell.ecn.purdue.edu/∼speechg
[7] Y. M EDAN , E. YAIR , D. C HAZAN, An accurate pitch detection algorithm, 9-th Int. Conference on
Pattern Recognition, Rome, Italy, 1, 476–80, November (1988).
[8] W. Z HANG , G. X U , Y. WANG, Pitch estimation based on circular AMDF, ICASSP 1, 341–344
(2002).
[9] X. M EI , J. PAN , S. S UN, Efficient algorithms for speech pitch estimation, Proceedings of 2001
International Symposium on Intelligent Multimedia, Video and Speech Proc. Hong Kong, pp. 421-
424, (2001).
[10] B. KOSTEK , A. C ZY ŻEWSKI, Representing musical instrument sounds for their automatic classifi-
cation, J. Audio Eng. Soc., 49, 9, 768–785 (2001).
[11] J. D. W IZE , J. R. C APRIO , T. W. PARKS, Maximum-likelihood pitch estimation, IEEE Trans. of
ASSP, 24, 418–423, October (1976).
[12] J. H U , S. X U , J. C HEN, A modified pitch detection algorithm, IEEE Communications Letters, 5, 2
(2001).
[13] K. K ASI , S. A. Z AHORIAN, Yet another algorithm for pitch tracking, ICASSP, 1, 361–364 (2002).
[14] N. K UNIEDA , T. S HIMAMURA , J. S UZUKI, Robust method of measurement of fundamental fre-
quency by ACOLS-autocorrelation of log spectrum, IEEE Int. Conf. On Acoustics, Speech, and
Signal Processing, 1, Atlanta, GA, 232–235, May (1996).
[15] L. JANER, Modulated gaussian wavelet transform based speech analyser pitch detection algorithm,
Proc. EUROSPEECH, 1, 401–404 (1995).
[16] R. J. M C AULAY, T. F. Q UATIERI, Pitch estimation and voicing detection based on a sinusoidal
speech model, ICASSP, 1, 249–252 (1990).
[17] L. R. R ABINER , M. J. C HENG , A. E. ROSENBERG , C. A. M CGOGENAL, A comparative perfor-
mance study of several pitch detection algorithms, IEEE Trans. on Acoustics, Speech and Signal
Proc., ASSP-24, 5, October (1976).
[18] C. A. M C G OGENAL , L. R. R ABINER , A. E. ROSENBERG, A subjective evaluation of pitch detec-
tion methods using LPC synthesized speech, IEEE Trans. on Acoustics, Speech and Signal Proc.,
ASSP-25, 3, June (1977).
[19] R. A HN , W. H. H OLMES, An improved harmonic-plus-noise decomposition method and its appli-
cation in pitch determination, Proc. IEEE Workshop on Speech Coding for Telecommunications,
Pocono Manor, Pennsylvania, pp. 41-42, (1997).
[20] C. D’A LESSANDRO , B. Y EGNANARAYANA , V. DARSINOS, Decomposition of speech signals into
deterministic and stochastic components, ICASSP, 1, 760–763 (1995).
[21] S. O SOWSKI, Artificial neural networks in algorithmic approach [in Polish], WNT, Warsaw 1996.