Journal of Information Security, 2012, 3, 335-340
http://dx.doi.org/10.4236/jis.2012.34041 Published Online October 2012 (http://www.SciRP.org/journal/jis)
Text Independent Automatic Speaker Recognition System
Using Mel-Frequency Cepstrum Coefficient and Gaussian
Alfredo Maesa1, Fabio Garzia1,2, Michele Scarpiniti1, Roberto Cusani1
1Department of Information, Electronics and Telecommunications Engineering, University of Rome, Rome, Italy
2Wessex Institute of Technology, Southampton, UK
Received July 21, 2012; revised August 14, 2012; accepted September 3, 2012
The aim of this paper is to show the accuracy and time results of a text independent automatic speaker recognition
(ASR) system, based on Mel-Frequency Cepstrum Coefficients (MFCC) and Gaussian Mixture Models (GMM), in or-
der to develop a security control access gate. 450 speakers were randomly extracted from the Voxforge.org audio data-
base, their utterances have been improved using spectral subtraction, then MFCC were extracted and these coefficients
were statistically analyzed by GMM in order to build each profile. For each speaker two different speech files were
used: the first one to build the profile database, the second one to test the system performance. The accuracy achieved
by the proposed approach is greater than 96% and the time spent for a single test run, implemented in Matlab language,
is about 2 seconds on a common PC.
Keywords: Automatic Speaker Recognition; Access Control; Voice Recognition; Biometrics
In last decades, an increasing interest in security systems
has arisen. These systems are very useful since they allow
managing security in a very efficient way, reducing the
need of human resources. Most of them implement an
access control system [1-4]. In particular, a huge number
of research efforts were directed to speaker recognition
problem [5-15]. In fact, many strategic places are of vital
importance to the assessment of involved people. A sim-
ple way to verify people identity can consist in analyzing
its voice. In fact, voice based recognition systems repre-
sent biometric systems that allow the access control in a
very fast and low intrusive way, requesting a reduced
collaboration of the people.
The human voice is peculiar to each person and this is
due to the anatomical apparatus of phonation. The vocal
tract consists of three main cavities: the oral cavity, the
nasal cavity and the pharyngeal cavity . The nasal
cavity is essentially bony, hence static in time; further-
more it can be isolated through the soft palate. The oral
cavity is formed by the bony structure of the palate and
soft palate; its conformation can be altered significantly
by the movement of the jaw, lips and tongue. The pharyn-
geal cavity extends to the bottom of the throat and it can
be compressed retracting the base of the tongue towards of
the wall of the pharynx. In the lower part it ends with the
vocal cords: a couple of fleshy membranes traversed by
the air coming from the lungs. During the production of a
sound, the space between the membranes (glottis) can be
completely opened or partially closed.
Due to the peculiarity of the voice formation apparatus,
it can be possible to recognize a particular individual from
its voice. In addition, this operation can be evaluated in an
automatic approach [13-15]. In literature, this problem is
addressed as Automatic Speaker Recognition (ASR) ,
and it is widely discussed by the research community
Speaker recognition is classified as a hybrid biometric
recognition approach, as it has two components: the phy-
sical one related to the anatomy of the vocal apparatus,
and the behavioral component, pertinent to the mood of
the speaker just in the recording moment .
There are several approach to ASR based on features,
vector quantization, score normalization, pattern match-
ing, etc., but the most of them are text dependent [6,7,9-11,
In this paper, we propose text independent ASR system
based on Mel-Frequency Cepstrum Coefficients (MFCC)
[18,19] and Gaussian Mixture Models (GMM) [20-22].
Then the model parameters are estimated with the maxi-
mum similarity making use of the Expectation and
Maximization (EM) algorithm [23,24]. The novel com-
opyright © 2012 SciRes. JIS
A. MAESA ET AL.
bination of these two techniques, allows the system to
reach high recognition rates and high operative velocities,
as shown in the following, allowing to use the proposed
system in real security context. In addition, unlike other
works on ASR presented in literature, because the re-
corded speaker signal could be corrupted by environ-
mental additive noise, a spectral subtraction algorithm
[25,26] is also used. Comparisons with the state of the art
demonstrate the effectiveness of the proposed approach in
terms of accuracy rate.
The data acquisition can be performed through simple
microphones which are well spread and their cost is neg-
ligible. However cheap instrumentation may be more
affected by disturbances such as background noise and the
spectral subtraction algorithm could be no more sufficient
for efficient noise suppression.
The paper is organized as follows: Section 2 describes
the ASR problem. Section 3 introduces the MFCC tech-
nique, while Section 4 introduces the GMM models. Sec-
tion 5 describes the proposed ASR system and Section 6
shows some interesting experimental results. Finally Sec-
tion 7 concludes the work.
2. System Description
A biometric recognition system generally consists of:
A sensor which makes acquisition of data and its
subsequent sampling: in the specific case the sensor is
a microphone, possibly with a high Signal to Ratio
(SNR) value. Since the input signal is essentially
speech, the sampling rate is usually set to 8 kHz;
A step of preprocessing that in the voice context is
constituted by the signal cleaning: simply denoising
algorithm can be applied to recorded data after a
normalization procedure. In order to clean recorded
speech signal from environmental additive noise, a
spectral subtraction algorithm is used [23,24] in this
The extraction of the peculiar characteristics (feature
extraction): in this stage Mel frequency cepstral coef-
ficients are evaluated using a Mel filter bank after a
transformation of the frequency axis in a logarithmic
The generation of a specific template for each speaker:
in this work we have decided to use the Gaussian
Mixture Models (GMM) where model parameters are
estimated with the maximum similarity making use of
the Expectation and Maximization (EM) algorithm;
In case of the user is registering (enrollment) for the
first time to the system, this template will be added to
the database, using some database programming tech-
Otherwise, in case of test among users already present
in the database, a comparison (matcher) determines
which profile matches the generated template of the
test speech. The matcher utilizes a similarity test, ob-
taining by a ratio value that can be accepted if it is
higher than a decision threshold.
The typical ASR system is shown in Figure 1.
The technologies used for the development of the
biometric system are the MMFCC for the extraction of
the characteristics and the GMM for the statistical analy-
sis of the data obtained, for the templates generation and
for the comparison.
3. Mel Frequency Cepstral Coefficient
The term “cepstrum” is a pun where the first letters of the
term “spectrum” are reversed. It was described in 1963
by Bogert et al. . Cepstrum is defined as the inverse
Fourier transform of the logarithm of the spectrum of a
The cepstrum transform the signal from the frequency
domain into the quefrency domain.
When cepstrum is applied to the voice, its strength is
to be able to divide excitation and transfer function. In a
signal y(n) based on the source-filter model, in this spe-
cific context, respectively the vocal cords and the vocal
tract, cepstrum allows separation in
where the source x(n) passes through a filter described by
the impulse response h(n). The spectrum of y(n) obtained
by the Fourier transform is
YkXk Hk where
k index of discrete frequencies, i.e. the product of two
spectra, respectively the source and the filter one. Sepa-
rating these two spectra is complicated. On the contrary,
it is possible to separate the real envelope of the filter
from the remaining spectrum by formulating all the
phase at the beginning. The cepstrum is based on the
properties of the logarithm that can transform the product
of the argument in sums of logarithms.
Starting from the logarithm of the modulus of the
Figure 1. A typical ASR system.
Copyright © 2012 SciRes. JIS
A. MAESA ET AL. 337
it is possible to separate the fast oscillating component
from the slow one, respectively by means of a high and
low pass filter, obtaining:
log H k
that is the signal cepstrum in the quefrency domain. In
the low quefrencies are described the transfer function
information, in the high quefrencies there is data about
Hence the initial wave of percussion created by the
vocal cords and shaped by the throat, nose and mouth can
be analyzed as a sum of a source function (given by the
excitation of the vocal cords) and a filter (throat, nose,
mouth). The separation between high and low quefrency,
can be obtained by a high pass lifter (filter) for the fast
oscillation and a low pass lifter for the slow one.
Psychoacoustic studies [30-32] have shown that the
mind perception of the frequency content of the sound
follows a nearly logarithmic scale, the Mel scale, which
is linear up to 1 kHz and logarithmic thereafter:
mel 2595log 17000
if 1 kHzf
The Mel scale is shown in Figure 2, where it is clear
the compression of the Mel scale (reported in y-axis)
with respect the Hertz scale (in x-axis) for frequencies
greater than 1 kHz. In this scale pitches are judged by
listeners to be equal in distance from one another.
Mel-cepstrum estimates the spectral envelope of the
output of the filter bank. Let Yn represent the logarithm of
the output energy from channel n, applying the discrete
cosine transform (DCT) we obtain the cepstral coeffi-
cients MFCC through the equation:
0,, k K
The simplified spectral envelope is rebuilt with the
Figure 2. Mel filter bank.
first Km coefficients, with Km < K:
where Bm is the bandwidth analyzed in Mel domain and
Km = 20 is a typical value assumed by Km. c0 is the mean
value in dB of the energy of the filter bank channels,
hence it is in direct relation with the energy of the sound
and it can be used for the estimation of the energy.
Schematically, the coefficients are derived in the fol-
lowing way: the spectrum of the original signal is com-
puted with the Fourier transform; the obtained spectrum
is mapped in Mel making use of appropriate overlapping
windows; for each obtained function the logarithm is
calculated; the discrete cosine transform is calculated
(DCT); the coefficients are the amplitudes of the result-
ing spectrum. In order to emphasize the low quefrencies
DCT is chosen.
4. Gaussian Mixture Model
Each arbitrary probability density function (pdf) can be
approximated by a linear combination of unimodal
Gaussian density . Under this assumption, Gaussian
mixture models have been applied to model the distribu-
tion of a sequence of vectors 12 tT
each one of dimension D, containing data on the charac-
teristics extracted from the voice of a subject, according
where wi are the weights of the corresponding mixtures to
the unimodal Gaussian densities pi with
The weights of the mixtures satisfy the constraint:
Each speaker is identified by a
mdel obtained from
GMM analysis. In particular lambda is defined as:
is the mean vector and is the covariance
Given a characteristic vector sequence of the speaker to
be identified, the model parameters are estimated with the
Copyright © 2012 SciRes. JIS
A. MAESA ET AL.
making use of the Expectation
and Maximization algorithm [23,24]. The
compared with a characteristic vector X by calculating the
log-likelihood similarity :
In order to decide, it is utilized a similarity test obtained
by the following ratio:
is the dec on the contrary, a collection of
models of different speakers. The final score of a certain
subject c over an X vector containing the voice features
of the test is given by:
where represents the similarity value of X vector
with respect to c compared with the characteristics of
other individuals in the database (pop), excluding the one
taken into account.
5. System Implementation
In the pre-processing phase, the signal has been im-
proved using spectral subtraction [33,34] and segmented
into frames partially overlying (50%) and relatively small.
Frames not containing voice were skipped. The size of
each frame is less than 20 ms in order to make the con-
tained wave stationary. Each frame has been subjected to
the Hamming window to minimize the discontinuities at
the edges of the frame. For each frame 20 MFCC were
calculated. The obtained data represents the characteris-
tics of a speaker. This information, organized in a matrix
containing a vector of Mel-Cepstral coefficients for each
frame, is analyzed by the GMM using 32 mixtures. The
result is a set of statistical data characterized by a mean
vector, a covariance matrix and a weight vector which
constitute the template itself. The template is employed
when a speaker is added into the system or for the test
step among the users already registered.
The public voice database Voxforge.org  was used
in order to validate the system. Voxforge is an internet
community including researchers and “donors” of human
voice. The preset aims are to support who intends to re-
alize and test an automatic speaker recognition system, a
speech recognition engine, or any application related to
analysis, to the recognition and more generally to the
study of the human voice. Anyone can register on the
website and send his own voice recordings to be made
available to the whole community. For this study 450
speaker utterances were randomly extracted from Vox-
forge website. For each speaker two speeches were em-
ployed: the first one in order to perform the training
phase and the second one to test the system. Since the
recognition system is text independent, each speech con-
tains different words (typically reads paragraphs of popu-
In the training step each template generated from the
analysis of the speakers’ utterances is stored into the
system. This set of information represents the knowledge
base of the system obtained in the training phase. The
test stage was made utilizing the test templates of each
speaker compared to the whole knowledge base of the
system, i.e. all the templates stored in the training phase.
This comparison was performed using the criterion of
log-likelihood previously described. The output of the
test phase is a matrix containing the similarity estimation
of each test with respect to each profile stored in the sys-
tem. This matrix is structured in this way: the rows rep-
resent the ith test and the column the j-training. Hence in
position (i,j) is contained the value representing the
similarity likelihood of test speaker i with respect to
training speaker j. Since the comparison is made by
log-likelihood, for each row (test) the system nominates
the column (speaker in the system) containing the maxi-
mum value as the owner of the speech.
6. Experimental Results
As shown in Table 1 there were 433 identifications on
450 subjects, this means that accuracy rate is 96.22%.
Since the system creates a hierarchy of candidates owners
of each test, if the top five were accepted as good results, it
would be achieved a recognition rate of 97.78%.
With regard to temporal performances, it should be
taken into account that the complete computation test
involves the training data processing, the test data elabo-
ration and the comparison from training and test data.
Obviously it is also possible perform a single test and
compare it to profiles in the system. These performance
results in terms of time required, are specific to the data-
base used, since the system developed can run with audio
files containing variable size, speech length and sampling.
The temporal performances are exposed in Table 2.
7. Comparison with the State of Art
This section discusses about the main speaker recogni-
tion systems found in scientific literature. In 1995 Rey-
nolds  implemented an identification system based on
spectral variability obtaining a 96.80% accuracy rate
with 49 speakers. In 2009 Revathi, Ganapathy and
Venkataramani  through an iterative clustering ap-
Table 1. Accuracy performances.
Speakers Hit Accuracy
1st 450 433 96.22%
In top 5 450 440 97.78%
Copyright © 2012 SciRes. JIS
A. MAESA ET AL. 339
proach, PLP (Perceptual Linear Predictive cepstrum) and
MF-PLP (Mel Frequency PLP) achieved 91% accuracy
rate with 50 speakers randomly chosen from TIMIT da-
tabase . In 2009 Chakroborty and Saha  combin-
ing MFCC and IMFCC (Inverted MFCC) based on gaus-
sian filter, reached 97.42% accuracy rate with 131 sub-
jects of YOHO database . In 2010 Saeidi, Mowlaee,
Kinnunen and Zheng-Hua  through Kullback-Leibler
divergence achieved 97% accuracy rate with 34 speakers.
In 2011 Gomez  implemented an identification sys-
tem based on novel parametric neural network, reaching
94% accuracy with 40 speakers. In 2011 Rao, Prasada
and Nagesh  made a study comparing GMM, HMM
(Hidden Markov Models) and MFCC. The accuracy rate
obtained in best test condition was 99% on 200 subjects
taken from TIMIT database.
Table 3 summarizes the accuracy rates reached by the
In this paper we have introduced an ASR system based on
MFCC and GMM. The accuracy of the proposed system is
greater than 96% and with 450 speakers.
Ith, as shown as a high recognition rate on a wide
number of subjects, together with a high operative velo-
city, make it useful for real security access control appli-
Table 2. Time performances.
Whole computation 17:03
Test & comparison 16:15
Single test 00:02
Table 3. Comparison with the state of the art.
Approach Accuracy rate
Reynolds  96.80%
Revathi et al.  91%
Chakroborty and Saha  97.42%
Saeidi et al.  97%
Gomez  94%
Rao et al.  99%
 F. Garzia, E. Sammarco and R. Cusani, “The Integrated
Security System of the Vatican City State,” International
Journal of Safety & Security Engineering, Vol. 1, No. 1,
2011, pp. 1-17. doi:10.2495/SAFE-V1-N1-1-17
 G. Contardi, F. Garzia and R. Cusani, “The Integrated
Security System of the Senate of the Italian Republic,”
International Journal of Safety & Security Engineering,
Vol. 1, No. 3, 2011, pp. 219-247.
 F. Garzia and R. Cusani, “The Safety/Security/Communi-
cation System of the Gran Sasso Mountain in Italy,” In-
ternational Journal of Safety & Security Engineering, Vol.
2, No. 1, 2012, pp. 13-39.
 F. Garzia, E. Sammarco and R. Cusani, “Vehicle/People
Access Control System for Security Management in Ports,”
International Journal of Safety & Security Engineering.
 H. Beigi, “Fundamentals of Speaker Recognition,” VDM
Verlag, Saarbrücken, 2011.
 R. J. Mammone, X. Y. Zhang and R. P. Ramachandran,
“Robust Speaker Recognition: A Feature-Based Approach,”
IEEE Signal Processing Magazine, Vol. 13, No. 5, 1996,
pp. 1290-1312. doi:10.1109/79.536825
 F. Soong, A. Rosenberg, L. Rabiner and B. Juang, “A
Vector Quantization Approach to Speaker Recognition,”
Acoustics, Speech and Signal Processing (ICASSP), 1985,
 R. Auckenthaler, M. Carey and H. Lloyd-Thomas, “Score
Normalization for Text-Independent Speaker Verification
Systems,” Digital Signal Processing, Vol. 10, No. 1-3,
2000, pp. 42-54. doi:10.1006/dspr.1999.0360
 S. Furui, “Recent Advances in Speaker Recognition,”
Pattern Recognition Letters, Vol. 18, No. 9, 1997, pp.
 S. Pruzansky, “Pattern-Matching Procedure for Automa-
tic Talker Recognition,” Journal of the Acoustical Society
of America, Vol. 26, No. 1, 1963, pp. 403-406.
 P. D. Bricker and S. Pruzansky, “Effects of Stimulus Con-
tent and Duration on Talker Identification,” Journal of the
Acoustical Society of America, Vol. 44, No. 3, 1968, pp.
 D. Jurafsky and J. H. Martin, “Speech and Language
Processing,” Prentice Hall, Boston, 2008.
 M. Farrùs, “Prosody in Automatic Speaker Recognition:
Applications in Biometrics and Voice Imitation,” VDM
Verlag, Saarbrücken, 2010.
 D. A. Reynolds, “An Overview of Automatic Speaker
Recognition Technology,” Acoustics, Speech and Signal
Processing (ICASSP), 2002, pp. 4072-4075.
 J. Mariani and F. Bimbot, “Language and Speech Pro-
cessing,” John Wiley & Sons, Chichester, 2010.
 I. R. Titze, “Principles of Voice Production,” Prentice
Hall, Boston, 1994.
 N. Morgan, H. Bourlard and H. Hermansky, “Speech
Processing in the Auditory System, Chapter Automatic
Speech Recognition: An Auditory Perspective,” Springer,
Copyright © 2012 SciRes. JIS
A. MAESA ET AL.
Copyright © 2012 SciRes. JIS
 F. Zheng, G. Zhang and Z. Song, “Comparison of Dif-
ferent Implementations of MFCC,” Journal of Computer
Science & Technology, Vol. 16, No. 6, 2001, pp. 582-589.
 M. Sahidullah and G. Saha, “Design, Analysis and Ex-
perimental Evaluation of Block Based Transformation in
MFCC Computation for Speaker Recognition,” Speech
Communication, Vol. 54, No. 4, 2012, pp. 543-565.
 C. Bishop, “Pattern Recognition and Machine Learning,”
Springer, Berlin, 2006.
 D. A. Reynolds, “Gaussian Mixture Models,” Technical
Report, MIT Lincoln Laboratory, Cincinnati, 2001.
 M. A. T. Figueiredo and A. K. Jain, “Unsupervised Learn-
ing of Finite Mixture Models,” IEEE Transactions on
Pattern Analysis and Machine Intelligence, Vol. 24, No.
3, 2002, pp. 381-396. doi:10.1109/34.990138
 D. A. Reynolds, T. F. Quatieri and R. B. Dunn, “Speaker
Verification Using Adapted Gaussian Mixture Models,”
Digital Signal Processing, Vol. 10, No. 2, 2000, pp. 19-
 L. Xu and I. Jordan, “On Convergence Properties of the
EM Algorithm for Gaussian Mixtures,” Neural Computa-
tion, Vol. 8, No. 1, 1996, pp. 129-151.
 S. V. Vaseghi, “Advanced Digital Signal Processing and
Noise Reduction,” John Wiley & Sons, Chichester, 2006.
 S. Boll, “Suppression of Acoustic Noise in Speech Using
Spectral Subtraction,” IEEE Transactions on Acoustics,
Speech and Signal Processing, Vol. 27, No. 2, 1979, pp.
 B. P. Bogert, J. R. Healy and J. W. Tukey, “The Que-
frency Analysis of Time Series for Echoes: Cepstrum,
Pseudo-Autocovariance, Cross-Cepstrum, and Saphe Crack-
ing,” Proceedings of the Symposium on Time Series Ana-
lysis, 1963, pp. 209-243.
 C. Roads, “The Computer Music Tutorial,” MIT Press,
 J. G. Proakis and D. G. Manolakis, “Digital Signal Pro-
cessing,” Prentice Hall, Boston, 2007.
 D. O’Shaughnessy, “Speech Communication: Human and
Machine,” Addison-Wesley, Boston, 1987.
 S. Stevens, J. Stanley, J. Volkman and E. B. Newman, “A
Scale for the Measurement of the Psychological Magni-
tude Pitch,” Journal of the Acoustical Society of America,
Vol. 8, No. 3, 1937, pp. 185-190. doi:10.1121/1.1915893
 M. Gold, “Speech and Audio Signal Processing,” John
Wiley & Sons, Chichester, 2002.
 M. Berouti, R. Schwartz and J. Makhoul, “Enhancement
of Speech Corrupted by Acoustic Noise, Proceedings of
IEEE International Conference on Acoustics, Speech, and
signal Processing (ICASSP1979), 1979, pp. 208-211.
 R. Martin, “Noise Power Spectral Density Estimation Based
on Optimal Smoothing and Minimum Statistics,” Speech
and Audio Processing, Vol. 9, No. 5, 2001, pp. 504-512.
 “Voxforge Database.” www.voxforge.org
 R. Reynolds, “Robust Text-Independent Speaker Identi-
fication Using Gaussian Mixture Speaker Models,” IEEE
Transactions on Speech and Audio Processing, 1995, pp.
 A. Revathi, R. Ganapathy and Y. Venkataramani, “Text
Independent Speaker Recognition and Speaker Inde-
pendent Speech Recognition Using Iterative Clustering
Approach,” International Journal of Computer Science &
Information Technology, Vol. 1, No. 2, 2009, pp. 30-42.
 “TIMIT Speech Database.” http://www.ldc.upenn.edu
 S. Chakroborty and G. Saha, “Improved Text-Indepen-
dent Speaker Identification Using Fused MFCC and
IMFCC feature Sets Based on Gaussian Filter,” Interna-
tional Journal of Signal Processing, Vol. 5, No. 1, 2009,
 “Yoho Speech Database.” http://www.ldc.upenn.edu
 R. Saeidi, P. Mowlaee, T. Kinnunen and Z. H. Tan, “Sig-
nal-to-Signal Ratio Independent Speaker Identification
for Co-Channel Speech Signals,” Proceedings of Interna-
tional Conference on Pattern Recognition (ICPR2009),
2009, pp. 4565-4568.
 P. Gomez, “A Text Independent Speaker Recognition
System Using a Novel Parametric Neural Network,” Pro-
ceedings of International Journal of Signal Processing,
Image Processing and Pattern Recognition, December
2011, pp. 1-16.
 R. R. Rao, V. K. Prasad and A. Nagesh, “Performance
Evaluation of Statistical Approaches for Text-Indepen-
dent Speaker Recognition Using Source Feature,” Inter-
JRI Computer Science and Networking, Vol. 2, No. 1,
2010, pp. 8-13.