Introduction to Audio AnalysisAMATLAB ApproachT. Giannakopoulos
and A. Pikrakis,ElsevierAcademicPress,Waltham,MA(2014),
288pp.,hardbound, 79.96 USD, ISBN 978-0-0-8-099388-1.
This book provides a palatable introduction to the fieldof audio
analysis, presenting concepts and examples us-ing the MATLAB
programming language as a guide.For readers with general audio,
science or math back-grounds who would like to gain insight on the
basic the-ory and practicalities of various audio analysis
topics,this book is a good resource. To aid in the introductionof
these topics, the book contains the MATLAB AudioAnalysis Library
with accompanying MATLAB files.These files contain useful codes for
performing numerousaudio analysis tasks, which are explained in the
text. Thebook also contains exercises at the end of each
chapter,which may be completed by the interested learner to
helpdevelop his or her audio analysis and programming skills.These
exercises would also be useful in an instructionalsetting,
providing good assignments for students in an au-dio analysis class
(as an added bonus, they are even ratedaccording to difficulty).
Practical explanatory examples,which are helpful for interpreting
and contextualizingthe audio analysis theory presented, are
included through-out the text.
The book is divided into three parts, with the firstcovering
fundamental audio analysis topics and techni-ques that are directed
to beginners in the field. The firstchapter succinctly summarizes
the book contents.Chapter 2 provides descriptions and MATLAB
exam-ples on the creation, playback, loading and storage ofaudio
signals, which is necessary before analysis canbegin. Signal
transforms and filtering concepts are dis-cussed in Chapter 3
including the introduction of theFourier, Cosine and Wavelet
Transforms. Aliasing isalso considered. Chapter 4 introduces audio
featuresand feature extraction techniques, which are essentialfor
pattern recognition and machine learning tasks. Dis-cussions of
both time-domain and frequency-domainaudio features are
included.174 Noise Control Engr. J. 62 (3), May-June 2014The second
part of the book delves into more ad-vanced topics related to
characterizing audio content.A reader who is already familiar with
the basics maystart with this part if so inclined. Chapter 5
describeshow the features extracted (per Chapter 4) may beused to
classify audio signals. This chapter concen-trates on
classification algorithms useful for classify-ing audio segments to
predefined classes, includingthe Bayesian classifier,
k-nearest-neighbor classifier,and support vector machines. Chapter
6 focuses onaudio segmentation techniques, which are designedto
separate an audio signal into its homogeneouscontent. Chapter 7
revisits the topic of classificationtechniques, discussing
strategies that utilize the tem-poral evolution of audio signals.
Topics include hid-den Markov modeling, the Viterbi algorithm
andthe BaumWelch algorithm.
The third part of the book provides a short summaryof music
information retrieval tasks, such as musicidentification, which is
used by popular mobile applica-tions to record a music signal and
identify the song. Thediscussion of music information retrieval
provided inPart III is only intended to provide a general insight
intothe topic. The text is not designed for the reader to gainan
in-depth understanding of the music information re-trieval tasks
presented.
In short, this book is excellent starting point forthose
whowould like to learn about audio analysis tech-niques and develop
practical skills for using MATLABto perform audio analysis tasks.
While much of the in-formation presented in Parts II and III would
require ad-ditional reading to master the topics at hand, this
bookserves very well to provide fundamental theory
andcontextualized examples relevant to the audio
analysisfield.Lauren M. RonsseAcoustics Program
Columbia College ChicagoChicago, IL, USA
[email protected]
