asr.tex

   1 %&asr
   2 \usepackage[toc,nonumberlist,acronyms]{glossaries}
   3 \makeglossaries%
   4 \newacronym{ANN}{ANN}{Artificial Neural Network}
   5 \newacronym{HMM}{HMM}{Hidden Markov Model}
   6 \newacronym{GMM}{GMM}{Gaussian Mixture Models}
   7 \newacronym{DHMM}{DHMM}{Duration-explicit \acrlong{HMM}}
   8 \newacronym{HTK}{HTK}{\acrlong{HMM} Toolkit}
   9 \newacronym{FA}{FA}{Forced alignment}
  10 \newacronym{MFC}{MFC}{Mel-frequency cepstrum}
  11 \newacronym{MFCC}{MFCC}{\acrlong{MFC} coefficient}
  12 \newacronym{PPF}{PPF}{Posterior Probability Features}
  13 \newacronym{MLP}{MLP}{Multi-layer Perceptron}
  14 \newacronym{PLP}{PLP}{Perceptual Linear Prediction}
  15 \newacronym{ZCR}{ZCR}{Zero-crossing Rate}
  16 \newacronym{LPC}{LPC}{Linear Prediction Coefficients}
  17 \newacronym{LPCC}{LPCC}{\acrlong{LPC} derivec cepstrum}
  18 \newacronym{IFPI}{IFPI}{International Federation of the Phonographic Industry}
  19 \newglossaryentry{dm}{name={Death Metal},
  20         description={is an extreme heavy metal music style with growling vocals and
  21         pounding drums}}
  22 \newglossaryentry{dom}{name={Doom Metal},
  23         description={is an extreme heavy metal music style with growling vocals and
  24         pounding drums played very slowly}}
  25 \newglossaryentry{FT}{name={Fourier Transform},
  26         description={is a technique of converting a time representation signal to a
  27         frequency representation}}
  28 \newglossaryentry{MS}{name={Mel-Scale},
  29         description={is a human ear inspired scale for spectral signals.}}
  30 \newglossaryentry{Viterbi}{name={Viterbi},
  31         description={is a dynamic programming algorithm for finding the most likely
  32         sequence of hidden states in a \gls{HMM}}}
  33
  34 \begin{document}
  35 \frontmatter{}
  36
  37 \maketitleru[
  38         course={(Automatic) Speech Recognition},
  39         institute={Radboud University Nijmegen},
  40         authorstext={Author:},
  41         pagenr=1]
  42 \listoftodos[Todo]
  43
  44 \tableofcontents
  45
  46 \mainmatter{}
  47 %Berenzweig and Ellis use acoustic classifiers from speech recognition as a
  48 %detector for singing lines.  They achive 80\% accuracy for forty 15 second
  49 %exerpts. They mention people that wrote signal features that discriminate
  50 %between speech and music. Neural net
  51 %\glspl{HMM}~\cite{berenzweig_locating_2001}.
  52 %
  53 %In 2014 Dzhambazov et al.\ applied state of the art segmentation methods to
  54 %polyphonic turkish music, this might be interesting to use for heavy metal.
  55 %They mention Fujihara (2011) to have a similar \gls{FA} system. This method uses
  56 %phone level segmentation, first 12 \gls{MFCC}s. They first do vocal/non-vocal
  57 %detection, then melody extraction, then alignment. They compare results with
  58 %Mesaros \& Virtanen, 2008~\cite{dzhambazov_automatic_2014}. Later they
  59 %specialize in long syllables in a capella. They use \glspl{DHMM} with
  60 %\glspl{GMM} and show that adding knowledge increases alignment (bejing opera
  61 %has long syllables)~\cite{dzhambazov_automatic_2016}.
  62 %
  63
  64
  65 %Introduction, leading to a clearly defined research question
  66 \chapter{Introduction}
  67 \input{intro.tex}
  68
  69 \chapter{Methods}
  70 \input{methods.tex}
  71
  72 \chapter{Conclusion \& Discussion}
  73 \input{conclusion.tex}
  74
  75 %(Appendices)
  76 \appendix
  77 \input{appendices.tex}
  78
  79 \newpage
  80 %Glossaries
  81 \glsaddall{}
  82 \begingroup
  83 \let\clearpage\relax
  84 \let\cleardoublepage\relax
  85 \printglossaries{}
  86 \endgroup
  87
  88 \bibliographystyle{ieeetr}
  89 \bibliography{asr}
  90 \end{document}