1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
|
% -*- mode: latex; TeX-master: "Vorbis_I_spec"; -*-
%!TEX root = Vorbis_I_spec.tex
\section{Floor type 0 setup and decode} \label{vorbis:spec:floor0}
\subsection{Overview}
Vorbis floor type zero uses Line Spectral Pair (LSP, also alternately
known as Line Spectral Frequency or LSF) representation to encode a
smooth spectral envelope curve as the frequency response of the LSP
filter. This representation is equivalent to a traditional all-pole
infinite impulse response filter as would be used in linear predictive
coding; LSP representation may be converted to LPC representation and
vice-versa.
\subsection{Floor 0 format}
Floor zero configuration consists of six integer fields and a list of
VQ codebooks for use in coding/decoding the LSP filter coefficient
values used by each frame.
\subsubsection{header decode}
Configuration information for instances of floor zero decodes from the
codec setup header (third packet). configuration decode proceeds as
follows:
\begin{Verbatim}[commandchars=\\\{\}]
1) [floor0\_order] = read an unsigned integer of 8 bits
2) [floor0\_rate] = read an unsigned integer of 16 bits
3) [floor0\_bark\_map\_size] = read an unsigned integer of 16 bits
4) [floor0\_amplitude\_bits] = read an unsigned integer of six bits
5) [floor0\_amplitude\_offset] = read an unsigned integer of eight bits
6) [floor0\_number\_of\_books] = read an unsigned integer of four bits and add 1
7) array [floor0\_book\_list] = read a list of [floor0\_number\_of\_books] unsigned integers of eight bits each;
\end{Verbatim}
An end-of-packet condition during any of these bitstream reads renders
this stream undecodable. In addition, any element of the array
\varname{[floor0\_book\_list]} that is greater than the maximum codebook
number for this bitstream is an error condition that also renders the
stream undecodable.
\subsubsection{packet decode} \label{vorbis:spec:floor0-decode}
Extracting a floor0 curve from an audio packet consists of first
decoding the curve amplitude and \varname{[floor0\_order]} LSP
coefficient values from the bitstream, and then computing the floor
curve, which is defined as the frequency response of the decoded LSP
filter.
Packet decode proceeds as follows:
\begin{Verbatim}[commandchars=\\\{\}]
1) [amplitude] = read an unsigned integer of [floor0\_amplitude\_bits] bits
2) if ( [amplitude] is greater than zero ) \{
3) [coefficients] is an empty, zero length vector
4) [booknumber] = read an unsigned integer of \link{vorbis:spec:ilog}{ilog}( [floor0\_number\_of\_books] ) bits
5) if ( [booknumber] is greater than the highest number decode codebook ) then packet is undecodable
6) [last] = zero;
7) vector [temp\_vector] = read vector from bitstream using codebook number [floor0\_book\_list] element [booknumber] in VQ context.
8) add the scalar value [last] to each scalar in vector [temp\_vector]
9) [last] = the value of the last scalar in vector [temp\_vector]
10) concatenate [temp\_vector] onto the end of the [coefficients] vector
11) if (length of vector [coefficients] is less than [floor0\_order], continue at step 6
\}
12) done.
\end{Verbatim}
Take note of the following properties of decode:
\begin{itemize}
\item An \varname{[amplitude]} value of zero must result in a return code that indicates this channel is unused in this frame (the output of the channel will be all-zeroes in synthesis). Several later stages of decode don't occur for an unused channel.
\item An end-of-packet condition during decode should be considered a
nominal occruence; if end-of-packet is reached during any read
operation above, floor decode is to return 'unused' status as if the
\varname{[amplitude]} value had read zero at the beginning of decode.
\item The book number used for decode
can, in fact, be stored in the bitstream in \link{vorbis:spec:ilog}{ilog}( \varname{[floor0\_number\_of\_books]} -
1 ) bits. Nevertheless, the above specification is correct and values
greater than the maximum possible book value are reserved.
\item The number of scalars read into the vector \varname{[coefficients]}
may be greater than \varname{[floor0\_order]}, the number actually
required for curve computation. For example, if the VQ codebook used
for the floor currently being decoded has a
\varname{[codebook\_dimensions]} value of three and
\varname{[floor0\_order]} is ten, the only way to fill all the needed
scalars in \varname{[coefficients]} is to to read a total of twelve
scalars as four vectors of three scalars each. This is not an error
condition, and care must be taken not to allow a buffer overflow in
decode. The extra values are not used and may be ignored or discarded.
\end{itemize}
\subsubsection{curve computation} \label{vorbis:spec:floor0-synth}
Given an \varname{[amplitude]} integer and \varname{[coefficients]}
vector from packet decode as well as the [floor0\_order],
[floor0\_rate], [floor0\_bark\_map\_size], [floor0\_amplitude\_bits] and
[floor0\_amplitude\_offset] values from floor setup, and an output
vector size \varname{[n]} specified by the decode process, we compute a
floor output vector.
If the value \varname{[amplitude]} is zero, the return value is a
length \varname{[n]} vector with all-zero scalars. Otherwise, begin by
assuming the following definitions for the given vector to be
synthesized:
\begin{displaymath}
\mathrm{map}_i = \left\{
\begin{array}{ll}
\min (
\mathtt{floor0\texttt{\_}bark\texttt{\_}map\texttt{\_}size} - 1,
foobar
) & \textrm{for } i \in [0,n-1] \\
-1 & \textrm{for } i = n
\end{array}
\right.
\end{displaymath}
where
\begin{displaymath}
foobar =
\left\lfloor
\mathrm{bark}\left(\frac{\mathtt{floor0\texttt{\_}rate} \cdot i}{2n}\right) \cdot \frac{\mathtt{floor0\texttt{\_}bark\texttt{\_}map\texttt{\_}size}} {\mathrm{bark}(.5 \cdot \mathtt{floor0\texttt{\_}rate})}
\right\rfloor
\end{displaymath}
and
\begin{displaymath}
\mathrm{bark}(x) = 13.1 \arctan (.00074x) + 2.24 \arctan (.0000000185x^2) + .0001x
\end{displaymath}
The above is used to synthesize the LSP curve on a Bark-scale frequency
axis, then map the result to a linear-scale frequency axis.
Similarly, the below calculation synthesizes the output LSP curve \varname{[output]} on a log
(dB) amplitude scale, mapping it to linear amplitude in the last step:
\begin{enumerate}
\item \varname{[i]} = 0
\item \varname{[$\omega$]} = $\pi$ * map element \varname{[i]} / \varname{[floor0\_bark\_map\_size]}
\item if ( \varname{[floor0\_order]} is odd ) {
\begin{enumerate}
\item calculate \varname{[p]} and \varname{[q]} according to:
\begin{eqnarray*}
p & = & (1 - \cos^2\omega)\prod_{j=0}^{\frac{\mathtt{floor0\texttt{\_}order}-3}{2}} 4 (\cos([\mathtt{coefficients}]_{2j+1}) - \cos \omega)^2 \\
q & = & \frac{1}{4} \prod_{j=0}^{\frac{\mathtt{floor0\texttt{\_}order}-1}{2}} 4 (\cos([\mathtt{coefficients}]_{2j}) - \cos \omega)^2
\end{eqnarray*}
\end{enumerate}
} else \varname{[floor0\_order]} is even {
\begin{enumerate}[resume]
\item calculate \varname{[p]} and \varname{[q]} according to:
\begin{eqnarray*}
p & = & \frac{(1 - \cos\omega)}{2} \prod_{j=0}^{\frac{\mathtt{floor0\texttt{\_}order}-2}{2}} 4 (\cos([\mathtt{coefficients}]_{2j+1}) - \cos \omega)^2 \\
q & = & \frac{(1 + \cos\omega)}{2} \prod_{j=0}^{\frac{\mathtt{floor0\texttt{\_}order}-2}{2}} 4 (\cos([\mathtt{coefficients}]_{2j}) - \cos \omega)^2
\end{eqnarray*}
\end{enumerate}
}
\item calculate \varname{[linear\_floor\_value]} according to:
\begin{displaymath}
\exp \left( .11512925 \left(\frac{\mathtt{amplitude} \cdot \mathtt{floor0\texttt{\_}amplitute\texttt{\_}offset}}{(2^{\mathtt{floor0\texttt{\_}amplitude\texttt{\_}bits}}-1)\sqrt{p+q}}
- \mathtt{floor0\texttt{\_}amplitude\texttt{\_}offset} \right) \right)
\end{displaymath}
\item \varname{[iteration\_condition]} = map element \varname{[i]}
\item \varname{[output]} element \varname{[i]} = \varname{[linear\_floor\_value]}
\item increment \varname{[i]}
\item if ( map element \varname{[i]} is equal to \varname{[iteration\_condition]} ) continue at step 5
\item if ( \varname{[i]} is less than \varname{[n]} ) continue at step 2
\item done
\end{enumerate}
\paragraph{Errata 20150227: Bark scale computation}
Due to a typo when typesetting this version of the specification from the original HTML document, the Bark scale computation previously erroneously read:
\begin{displaymath}
\hbox{\sout{$
\mathrm{bark}(x) = 13.1 \arctan (.00074x) + 2.24 \arctan (.0000000185x^2 + .0001x)
$}}
\end{displaymath}
Note that the last parenthesis is misplaced. This document now uses the correct equation as it appeared in the original HTML spec document:
\begin{displaymath}
\mathrm{bark}(x) = 13.1 \arctan (.00074x) + 2.24 \arctan (.0000000185x^2) + .0001x
\end{displaymath}
|
