--- /dev/null
+This thesis present a method of detecting \emph{singing}-voice segments in non
+standard music genres. The research uses existing techniques on extreme styles
+like \gls{dm} and \gls{dom} and achieves a good performance.
+\emph{Singer}-voice recognition and detection has also been attempted with
+similarl positive results. This is founded basis for attempting lyrics
+synchronization and lyrics recognition.
--- /dev/null
+I would like to thank\ldots
%&asr
-\usepackage[toc,nonumberlist,acronyms]{glossaries}
+\usepackage[nonumberlist]{glossaries}
\makeglossaries%
-\input{acronyms}
\input{glossaries}
\begin{document}
righttext={Louis ten Bosch},
pagenr=1]
+%Abstract
+\addcontentsline{toc}{chapter}{Abstract}
+\chapter*{\centering Abstract}
+\begin{quotation}
+ \centering\noindent
+ \input{abstract}
+\end{quotation}
+
+% Acknowledgements
+\addcontentsline{toc}{chapter}{Acknowledgements}
+\chapter*{\centering Acknowledgements}
+\begin{quotation}
+ \centering\it\noindent
+ \input{acknowledgements}
+\end{quotation}
\tableofcontents
\glsaddall{}
-\printglossaries{}
\mainmatter{}
\appendix
\input{appendices}
+\addcontentsline{toc}{chapter}{Glossaries \& Acronyms}
+\printglossaries%
+
\bibliographystyle{ieeetr}
\bibliography{asr}
\end{document}
also result in a model that is overfitted the three islands in entire space of
grunting voices.
-In this case it seems that the model generalizes well. The alien data similar
-to the trainingsdata offered to the model results in a good performance.
+In this case it seems that the model generalizes well. The alien data --- similar
+to the training data --- offered to the model, results in a good performance.
However, alien data that has a very different style does not perform as good.
While testing \emph{Catacombs} the performance was very poor. Adding
\emph{Catacombs} or a similar style to the training set can probably overcome
The current decorrelation step might be inefficient or unnatural. The \gls{ANN}
train the weights in such a way that performance is maximized. It would be
interesting to see whether this results in a different normalization step. The
-downside of this is that training the model is complexer because there are many
-more weights to train.
+downside of this is that training the model is more complex because there are
+many more weights to train.
\paragraph{Genre detection: }
\emph{Singing}-voice detection and \emph{singer}-voice can be seen as a crude
\newglossaryentry{Viterbi}{name={Viterbi},
description={is a dynamic programming algorithm for finding the most likely
sequence of hidden states in a \gls{HMM}}}
+
+\newcommand{\newglossacr}[2]{\newglossaryentry{#1}{
+ name={#1},
+ first={#2 (#1)},%
+ firstplural={#2\glspluralsuffix{} (#1\glspluralsuffix)},
+ description={#2}}}
+
+\newglossacr{GADT}{Generalized Algebraic Data type}
+\newglossacr{ANN}{Artificial Neural Network}
+\newglossacr{DCT}{Discrete Cosine Transform}
+\newglossacr{DHMM}{Duration-explicit Hidden Markov Model}
+\newglossacr{FA}{Forced Alignment}
+\newglossacr{GMM}{Gaussian Mixture Models}
+\newglossacr{HMM}{Hidden Markov Model}
+\newglossacr{HTK}{Hidden Markov Model Toolkit}
+\newglossacr{IFPI}{International Federation of the Phonographic Industry}
+\newglossacr{LPCC}{Linear Prediction Coefficient Derived Cepstrum}
+\newglossacr{LPC}{Linear Prediction Coefficient}
+\newglossacr{MFCC}{Mel-frequency Cepstral Coefficient}
+\newglossacr{MFC}{Mel-frequency Cepstrum}
+\newglossacr{MLP}{Multi-layer Perceptron}
+\newglossacr{PLP}{Perceptual Linear Prediction}
+\newglossacr{PPF}{Posterior Probability Features}
+\newglossacr{ZCR}{Zero-crossing Rate}
+\newglossacr{RELU}{Rectified Linear Unit}
+\newglossacr{CC}{Cannibal Corpse}
+\newglossacr{DG}{Disgorge}
+\newglossacr{WDISS}{Who Dies in Siberian Slush}
\section{Introduction}
The primary medium for music distribution is rapidly changing from physical
-media to digital media. In 2016 the \gls{IFPI} stated that about $43\%$ of
-music revenue arises from digital distribution. Another $39\%$ arises from the
+media to digital media. In 2016 the \gls{IFPI} stated that about $50\%$ of
+music revenue arises from digital distribution. Another $34\%$ arises from the
physical sale and the remaining $16\%$ is made through performance and
synchronisation revenues. The overtake of digital formats on physical formats
took place somewhere in 2015. Moreover, ever since twenty years the music
available. However, a temporal alignment of the lyrics is not and creating it
involves manual labour.
-A lot of the current day musical distribution goes via non-official channels
+A lot of the current day music distribution goes via non-official channels
such as YouTube\footnote{\url{https://youtube.com}} in which fans of the
performers often accompany the music with synchronized lyrics. This means that
there is an enormous treasure of lyrics-annotated music available. However, the
that music has different properties than speech. Music uses a wider spectral
bandwidth in which events happen. Music contains more tonality and rhythm.
Multivariate Gaussian classifiers were used to discriminate the categories with
-an average performance of $90\%$~\cite{saunders_real-time_1996}.
+an average accuracy of $90\%$~\cite{saunders_real-time_1996}.
Williams and Ellis were inspired by the aforementioned research and tried to
separate the singing segments from the instrumental segments~%
to detect a singing voice~\cite{berenzweig_using_2002}. Nwe et al.\ showed that
there is not much difference in accuracy when using different features founded
in speech processing. They tested several features and found accuracies differ
-less that a few percent. Moreover, they found that others have tried to tackle
+less than a few percent. Moreover, they found that others have tried to tackle
the problem using myriads of different approaches such as using \gls{ZCR},
\gls{MFCC} and \gls{LPCC} as features and \glspl{HMM} or \glspl{GMM} as
classifiers~\cite{nwe_singing_2004}.
\begin{figure}[ht]
\centering
\includegraphics[width=.7\linewidth]{cement}
- \caption{A vocal segment of the \acrlong{CC} song
+ \caption{A vocal segment of the Cannibal Corpse song
\emph{Bloodstained Cement}}\label{fig:bloodstained}
\end{figure}
\begin{figure}[ht]
\centering
\includegraphics[width=.7\linewidth]{abominations}
- \caption{A vocal segment of the \acrlong{DG} song
+ \caption{A vocal segment of the Disgorge song
\emph{Enthroned Abominations}}\label{fig:abominations}
\end{figure}
piano's and synthesizers. The droning synthesizers often operate in the same
frequency as the vocals.
-Additional detailss about the dataset are listed in Appendix~\ref{app:data}.
+Additional details about the dataset are listed in Appendix~\ref{app:data}.
The data is labeled as singing and instrumental and labeled per band. The
distribution for this is shown in Table~\ref{tbl:distribution}.
\begin{table}[H]
\caption{Data distribution}\label{tbl:distribution}
\end{table}
-\section{\acrlong{MFCC} Features}
+\section{Mel-frequencey Cepstral Features}
The waveforms in itself are not very suitable to be used as features due to the
high dimensionality and correlation in the temporal domain. Therefore we use
the often used \glspl{MFCC} feature vectors which have shown to be suitable in
magnitudes\footnote{Fechner, Gustav Theodor (1860). Elemente der
Psychophysik}. They found that energy is perceived in logarithmic
increments. This means that twice the amount of energy does not mean
- twice the amount of perceived loudness. Therefore we take the log of
- the energy or amplitude of the \gls{MS} spectrum to closer match the
+ twice the amount of perceived loudness. Therefore we take the logarithm
+ of the energy or amplitude of the \gls{MS} spectrum to closer match the
human hearing.
\item The amplitudes of the spectrum are highly correlated and therefore
the last step is a decorrelation step. \Gls{DCT} is applied on the
represents the overall energy in the \gls{MS}. Another option would be
$\log{(E)}$ which is the logarithm of the raw energy of the sample.
-\section{\acrlong{ANN}}
+\section{Artificial Neural Network}
The data is classified using standard \gls{ANN} techniques, namely \glspl{MLP}.
-The classification problems are only binary and four-class so it is
+The classification problems are only binary or four-class problems so it is
interesting to see where the bottleneck lies; how abstract can the abstraction
be made. The \gls{ANN} is built with the Keras\footnote{\url{https://keras.io}}
using the TensorFlow\footnote{\url{https://github.com/tensorflow/tensorflow}}
which is fully connected too the output layer. The activation function used is
a \gls{RELU}. The \gls{RELU} function is a monotonic symmetric one-sided
function that is also known as the ramp function. The definition is given in
-Equation~\ref{eq:relu}. \gls{RELU} was chosen because of its symmetry and
-efficient computation. The activation function between the hidden layer and the
-output layer is the sigmoid function in the case of binary classification, of
-which the definition is shown in Equation~\ref{eq:sigmoid}. The sigmoid is a
-monotonic function that is differentiable on all values of $x$ and always
-yields a non-negative derivative. For the multiclass classification the softmax
-function is used between the hidden layer and the output layer. Softmax is an
-activation function suitable for multiple output nodes. The definition is given
-in Equation~\ref{eq:softmax}.
+Equation~\ref{eq:relu}. \gls{RELU} has the downside that it can create
+unreachable nodes in a deep network. This is not a problem in this network
+since it only has one hidden layer. \gls{RELU} was also chosen because of its
+efficient computation and nature inspiredness.
+
+The activation function between the hidden layer and the output layer is the
+sigmoid function in the case of binary classification, of which the definition
+is shown in Equation~\ref{eq:sigmoid}. The sigmoid is a monotonic function that
+is differentiable on all values of $x$ and always yields a non-negative
+derivative. For the multiclass classification the softmax function is used
+between the hidden layer and the output layer. Softmax is an activation
+function suitable for multiple output nodes. The definition is given in
+Equation~\ref{eq:softmax}.
The data is shuffled before fed to the network to mitigate the risk of
-overfitting on one album. Every model was trained using $10$ epochs and a
-batch size of $32$. The training set and test set are separated by taking a
-$90\%$ slice of all the data.
+overfitting on one album. Every model was trained using $10$ epochs which means
+that all training data is offered to the model $10$ times. The training set and
+test set are separated by taking a $90\%$ slice of all the data.
\begin{equation}\label{eq:relu}
f(x) = \left\{\begin{array}{rcl}
\includegraphics[width=.8\linewidth]{mcann}
\caption{Multiclass classifier network architecture}\label{fig:mcann}
\end{subfigure}
- \caption{\acrlong{ANN} architectures.}
+ \caption{Artificial Neural Network architectures.}
\end{figure}
\section{Experimental setup}
\subsection{Features}
-The thirteen \gls{MFCC} features are used as the input. Th parameters of the
+The thirteen \gls{MFCC} features are used as the input. The parameters of the
\gls{MFCC} features are varied in window step and window length. The default
speech processing parameters are tested but also bigger window sizes since
arguably the minimal size of a singing voice segment is a lot bigger than the
\begin{table}[H]
\centering
- \begin{tabular}{rccc}
+ \begin{tabular}{rrccc}
\toprule
- & \multicolumn{3}{c}{Parameters (step/length)}\\
- & 10/25 & 40/100 & 80/200\\
+ & & \multicolumn{3}{c}{Parameters (step/length)}\\
+ & & 10/25 & 40/100 & 80/200\\
\midrule
\multirow{4}{*}{Hidden Nodes}
- & 0.86 (0.34) & 0.87 (0.32) & 0.85 (0.35)\\
- & 0.87 (0.31) & 0.88 (0.30) & 0.87 (0.32)\\
- & 0.88 (0.30) & 0.88 (0.31) & 0.88 (0.29)\\
- & 0.89 (0.28) & 0.89 (0.29) & 0.88 (0.30)\\
+ & 3 & 0.86 (0.34) & 0.87 (0.32) & 0.85 (0.35)\\
+ & 5 & 0.87 (0.31) & 0.88 (0.30) & 0.87 (0.32)\\
+ & 8 & 0.88 (0.30) & 0.88 (0.31) & 0.88 (0.29)\\
+ & 13 & 0.89 (0.28) & 0.89 (0.29) & 0.88 (0.30)\\
\bottomrule
\end{tabular}
\caption{Binary classification results (accuracy (loss))}%
\begin{table}[H]
\centering
- \begin{tabular}{rccc}
+ \begin{tabular}{rrccc}
\toprule
- & \multicolumn{3}{c}{Parameters (step/length)}\\
- & 10/25 & 40/100 & 80/200\\
+ & & \multicolumn{3}{c}{Parameters (step/length)}\\
+ & & 10/25 & 40/100 & 80/200\\
\midrule
\multirow{4}{*}{Hidden Nodes}
- & 0.83 (0.48) & 0.82 (0.48) & 0.82 (0.48)\\
- & 0.85 (0.43) & 0.84 (0.44) & 0.84 (0.44)\\
- & 0.86 (0.41) & 0.86 (0.39) & 0.86 (0.40)\\
- & 0.87 (0.37) & 0.87 (0.38) & 0.86 (0.39)\\
+ & 3 & 0.83 (0.48) & 0.82 (0.48) & 0.82 (0.48)\\
+ & 5 & 0.85 (0.43) & 0.84 (0.44) & 0.84 (0.44)\\
+ & 8 & 0.86 (0.41) & 0.86 (0.39) & 0.86 (0.40)\\
+ & 13 & 0.87 (0.37) & 0.87 (0.38) & 0.86 (0.39)\\
\bottomrule
\end{tabular}
\caption{Multiclass classification results (accuracy
repositories / asr1617.git / commitdiff