[BoW] r6 - in trunk: audio_coding/audio_coding.tex basic_glossary.tex introduction/introduction.tex
kostya
subversion at mplayerhq.hu
Sun Sep 28 14:21:08 CEST 2008
Author: kostya
Date: Sun Sep 28 14:21:07 2008
New Revision: 6
Log:
Fill BoW with some info.
I hope that after some formatting it can serve as an example
for future entries.
Added:
trunk/basic_glossary.tex
Modified:
trunk/audio_coding/audio_coding.tex
trunk/introduction/introduction.tex
Modified: trunk/audio_coding/audio_coding.tex
==============================================================================
--- trunk/audio_coding/audio_coding.tex (original)
+++ trunk/audio_coding/audio_coding.tex Sun Sep 28 14:21:07 2008
@@ -1 +1,98 @@
\part{Audio Coding}
+
+\section{Psychouacoustics}
+
+\section{Quality Metrics}
+
+\section{Predictive Coding}
+
+Prediction is wide-employed method for effective data compression. One can think about prediction as a FIR filter $y_n = x_n - f(x_{n-1}, x_{n-2}, \ddots)$ where $x_n$ is $n$-th input sample and $y_n$ is $n$-th output sample and $f(x)$ is a prediction function. Depending on prediction function type there is linear prediction (i.e. $y_n = x_n - \sum _{i=0} ^N a_i x_{n-i-1}$) and nonlinear prediction. Prediction is called fixed when prediction function does not employ any internal state and adaptive prediction when
+output depends on some internal state (e.g. linear prediction coefficients are adjusted depending on previous outputs).
+
+Linear prediction is commonly used in audio compression techniques --- (A)DPCM codecs, speech coding, lossless coding.
+
+\subsection{Temporal Correlation}
+
+Temporal correlation means correlating samples in time domain, i.e. predicting next waveform sample and coding the difference.
+
+\subsubsection{DPCM}
+Differential PCM is without a doubt one of the oldest audio compression techniques employed due its very low computation cost. Typical compression ratio is 2.
+
+\begin{verbatim}
+while there are samples left to decode
+ diff <-- get difference
+ predictor <-- predictor + diff
+ output predictor as a new sample
+\end{verbatim}
+
+\subsubsection{ADPCM}
+
+Most popular is IMA ADPCM compression scheme, which is used as a basis for many proprietary variations that are usually used in games and several widespread ADPCM codecs such as Apple IMA ADPCM and Microsoft IMA ADPCM.
+
+\begin{verbatim}
+while there are samples left to decode
+ diff <-- get difference
+ predictor <-- predictor + diff*step
+ output predictor as a new sample
+ adjust step based on diff
+\end{verbatim}
+
+\begin{table}
+\caption{Known (A)DPCM codecs comparison}
+\label{dpcmcmp}
+\begin{tabular}{|l|l|l|r|r|}
+\hline
+{\bf Name} &
+{\bf Employed at} &
+{\bf Difference type} &
+{\bf Source} &
+{\bf Compression rate}
+\\ \hline
+
+Id RoQ DPCM &
+Id RoQ &
+quadratic &
+16-bit &
+2:1
+\\ \hline
+
+IMA ADPCM &
+WAV, AVI, game formats &
+ &
+8-bit, 16-bit &
+4:1
+\\ \hline
+
+Interplay DPCM &
+Interplay MVE &
+custom &
+16-bit &
+2:1
+\\ \hline
+
+MS ADPCM &
+AVI, WAV &
+ &
+16-bit &
+4:1
+\\ \hline
+
+Sierra DPCM &
+Sierra .SOL, .VMD &
+logarithmic &
+8-bit, 16-bit &
+2:1
+\\ \hline
+
+Xan DPCM &
+AVI &
+logarithmic &
+16-bit &
+2:1
+\\ \hline
+\end{tabular}
+\end{table}
+
+\subsection{Spatial Correlation}
+
+\section{Error Concealment}
Added: trunk/basic_glossary.tex
==============================================================================
--- (empty file)
+++ trunk/basic_glossary.tex Sun Sep 28 14:21:07 2008
@@ -0,0 +1,11 @@
+\section*{Glossary}
+
+\begin{itemize}
+
+\item[ADPCM] Adaptive Differential PCM
+
+\item[DPCM] Differential PCM
+
+\item[PCM] Pulse Coded Modulation
+
+\end{itemize}
Modified: trunk/introduction/introduction.tex
==============================================================================
--- trunk/introduction/introduction.tex (original)
+++ trunk/introduction/introduction.tex Sun Sep 28 14:21:07 2008
@@ -1,3 +1,4 @@
-%\input{introduction.tex}
+\part*{Introduction}
+\addcontentsline{toc}{part}{Introduction}
-%\input{basic_glossary.tex}
+\input{basic_glossary.tex}
More information about the BoW
mailing list