1 % -*- mode: latex; TeX-master: "Vorbis_I_spec"; -*-
2 %!TEX root = Vorbis_I_spec.tex
4 \section{Floor type 1 setup and decode} \label{vorbis:spec:floor1}
8 Vorbis floor type one uses a piecewise straight-line representation to
9 encode a spectral envelope curve. The representation plots this curve
10 mechanically on a linear frequency axis and a logarithmic (dB)
11 amplitude axis. The integer plotting algorithm used is similar to
12 Bresenham's algorithm.
16 \subsection{Floor 1 format}
20 Floor type one represents a spectral curve as a series of
21 line segments. Synthesis constructs a floor curve using iterative
22 prediction in a process roughly equivalent to the following simplified
26 \item the first line segment (base case) is a logical line spanning
27 from x_0,y_0 to x_1,y_1 where in the base case x_0=0 and x_1=[n], the
28 full range of the spectral floor to be computed.
30 \item the induction step chooses a point x_new within an existing
31 logical line segment and produces a y_new value at that point computed
32 from the existing line's y value at x_new (as plotted by the line) and
33 a difference value decoded from the bitstream packet.
35 \item floor computation produces two new line segments, one running from
36 x_0,y_0 to x_new,y_new and from x_new,y_new to x_1,y_1. This step is
37 performed logically even if y_new represents no change to the
38 amplitude value at x_new so that later refinement is additionally
41 \item the induction step repeats, using a list of x values specified in
42 the codec setup header at floor 1 initialization time. Computation
43 is completed at the end of the x value list.
48 Consider the following example, with values chosen for ease of
49 understanding rather than representing typical configuration:
51 For the below example, we assume a floor setup with an [n] of 128.
52 The list of selected X values in increasing order is
53 0,16,32,48,64,80,96,112 and 128. In list order, the values interleave
54 as 0, 128, 64, 32, 96, 16, 48, 80 and 112. The corresponding
55 list-order Y values as decoded from an example packet are 110, 20, -5,
56 -45, 0, -25, -10, 30 and -10. We compute the floor in the following
57 way, beginning with the first line:
60 \includegraphics[width=8cm]{floor1-1}
61 \captionof{figure}{graph of example floor}
64 We now draw new logical lines to reflect the correction to new_Y, and
65 iterate for X positions 32 and 96:
68 \includegraphics[width=8cm]{floor1-2}
69 \captionof{figure}{graph of example floor}
72 Although the new Y value at X position 96 is unchanged, it is still
73 used later as an endpoint for further refinement. From here on, the
74 pattern should be clear; we complete the floor computation as follows:
77 \includegraphics[width=8cm]{floor1-3}
78 \captionof{figure}{graph of example floor}
82 \includegraphics[width=8cm]{floor1-4}
83 \captionof{figure}{graph of example floor}
86 A more efficient algorithm with carefully defined integer rounding
87 behavior is used for actual decode, as described later. The actual
88 algorithm splits Y value computation and line plotting into two steps
89 with modifications to the above algorithm to eliminate noise
90 accumulation through integer roundoff/truncation.
94 \subsubsection{header decode}
96 A list of floor X values is stored in the packet header in interleaved
97 format (used in list order during packet decode and synthesis). This
98 list is split into partitions, and each partition is assigned to a
99 partition class. X positions 0 and [n] are implicit and do not belong
100 to an explicit partition or partition class.
102 A partition class consists of a representation vector width (the
103 number of Y values which the partition class encodes at once), a
104 'subclass' value representing the number of alternate entropy books
105 the partition class may use in representing Y values, the list of
106 [subclass] books and a master book used to encode which alternate
107 books were chosen for representation in a given packet. The
108 master/subclass mechanism is meant to be used as a flexible
109 representation cascade while still using codebooks only in a scalar
112 \begin{Verbatim}[commandchars=\\\{\}]
114 1) [floor1\_partitions] = read 5 bits as unsigned integer
115 2) [maximum\_class] = -1
116 3) iterate [i] over the range 0 ... [floor1\_partitions]-1 \{
118 4) vector [floor1\_partition\_class\_list] element [i] = read 4 bits as unsigned integer
122 5) [maximum\_class] = largest integer scalar value in vector [floor1\_partition\_class\_list]
123 6) iterate [i] over the range 0 ... [maximum\_class] \{
125 7) vector [floor1\_class\_dimensions] element [i] = read 3 bits as unsigned integer and add 1
126 8) vector [floor1\_class\_subclasses] element [i] = read 2 bits as unsigned integer
127 9) if ( vector [floor1\_class\_subclasses] element [i] is nonzero ) \{
129 10) vector [floor1\_class\_masterbooks] element [i] = read 8 bits as unsigned integer
133 11) iterate [j] over the range 0 ... (2 exponent [floor1\_class\_subclasses] element [i]) - 1 \{
135 12) array [floor1\_subclass\_books] element [i],[j] =
136 read 8 bits as unsigned integer and subtract one
140 13) [floor1\_multiplier] = read 2 bits as unsigned integer and add one
141 14) [rangebits] = read 4 bits as unsigned integer
142 15) vector [floor1\_X\_list] element [0] = 0
143 16) vector [floor1\_X\_list] element [1] = 2 exponent [rangebits];
144 17) [floor1\_values] = 2
145 18) iterate [i] over the range 0 ... [floor1\_partitions]-1 \{
147 19) [current\_class\_number] = vector [floor1\_partition\_class\_list] element [i]
148 20) iterate [j] over the range 0 ... ([floor1\_class\_dimensions] element [current\_class\_number])-1 \{
149 21) vector [floor1\_X\_list] element ([floor1\_values]) =
150 read [rangebits] bits as unsigned integer
151 22) increment [floor1\_values] by one
158 An end-of-packet condition while reading any aspect of a floor 1
159 configuration during setup renders a stream undecodable. In addition,
160 a \varname{[floor1\_class\_masterbooks]} or
161 \varname{[floor1\_subclass\_books]} scalar element greater than the
162 highest numbered codebook configured in this stream is an error
163 condition that renders the stream undecodable. Vector
164 [floor1\_x\_list] is limited to a maximum length of 65 elements; a
165 setup indicating more than 65 total elements (including elements 0 and
166 1 set prior to the read loop) renders the stream undecodable. All
167 vector [floor1\_x\_list] element values must be unique within the
168 vector; a non-unique value renders the stream undecodable.
170 \paragraph{packet decode} \label{vorbis:spec:floor1-decode}
172 Packet decode begins by checking the \varname{[nonzero]} flag:
174 \begin{Verbatim}[commandchars=\\\{\}]
175 1) [nonzero] = read 1 bit as boolean
178 If \varname{[nonzero]} is unset, that indicates this channel contained
179 no audio energy in this frame. Decode immediately returns a status
180 indicating this floor curve (and thus this channel) is unused this
181 frame. (A return status of 'unused' is different from decoding a
182 floor that has all points set to minimum representation amplitude,
183 which happens to be approximately -140dB).
186 Assuming \varname{[nonzero]} is set, decode proceeds as follows:
188 \begin{Verbatim}[commandchars=\\\{\}]
189 1) [range] = vector \{ 256, 128, 86, 64 \} element ([floor1\_multiplier]-1)
190 2) vector [floor1\_Y] element [0] = read \link{vorbis:spec:ilog}{ilog}([range]-1) bits as unsigned integer
191 3) vector [floor1\_Y] element [1] = read \link{vorbis:spec:ilog}{ilog}([range]-1) bits as unsigned integer
193 5) iterate [i] over the range 0 ... [floor1\_partitions]-1 \{
195 6) [class] = vector [floor1\_partition\_class] element [i]
196 7) [cdim] = vector [floor1\_class\_dimensions] element [class]
197 8) [cbits] = vector [floor1\_class\_subclasses] element [class]
198 9) [csub] = (2 exponent [cbits])-1
200 11) if ( [cbits] is greater than zero ) \{
202 12) [cval] = read from packet using codebook number
203 (vector [floor1\_class\_masterbooks] element [class]) in scalar context
206 13) iterate [j] over the range 0 ... [cdim]-1 \{
208 14) [book] = array [floor1\_subclass\_books] element [class],([cval] bitwise AND [csub])
209 15) [cval] = [cval] right shifted [cbits] bits
210 16) if ( [book] is not less than zero ) \{
212 17) vector [floor1\_Y] element ([j]+[offset]) = read from packet using codebook
213 [book] in scalar context
215 \} else [book] is less than zero \{
217 18) vector [floor1\_Y] element ([j]+[offset]) = 0
222 19) [offset] = [offset] + [cdim]
229 An end-of-packet condition during curve decode should be considered a
230 nominal occurrence; if end-of-packet is reached during any read
231 operation above, floor decode is to return 'unused' status as if the
232 \varname{[nonzero]} flag had been unset at the beginning of decode.
235 Vector \varname{[floor1\_Y]} contains the values from packet decode
236 needed for floor 1 synthesis.
240 \paragraph{curve computation} \label{vorbis:spec:floor1-synth}
242 Curve computation is split into two logical steps; the first step
243 derives final Y amplitude values from the encoded, wrapped difference
244 values taken from the bitstream. The second step plots the curve
245 lines. Also, although zero-difference values are used in the
246 iterative prediction to find final Y values, these points are
247 conditionally skipped during final line computation in step two.
248 Skipping zero-difference values allows a smoother line fit.
250 Although some aspects of the below algorithm look like inconsequential
251 optimizations, implementors are warned to follow the details closely.
252 Deviation from implementing a strictly equivalent algorithm can result
253 in serious decoding errors.
255 {\em Additional note:} Although predicted values in the prediction
256 loop and at the end of step 1 (that is, the values in vector
257 \varname{[floor1\_final\_Y]} are inherently limited by the prediction
258 algorithm to \[0, \varname{[range]}\), it is possible to abuse the
259 setup and codebook machinery to produce negative or over-range
260 results. We suggest that decoder implementations guard the values in
261 vector \varname{[floor1\_final\_Y]} by clamping each element to \[0,
262 \varname{[range]}\) after step 1. Variants of this suggestion are
263 acceptable as valid floor1 setups cannot produce out of range values.
266 \item[step 1: amplitude value synthesis]
268 Unwrap the always-positive-or-zero values read from the packet into
269 +/- difference values, then apply to line prediction.
271 \begin{Verbatim}[commandchars=\\\{\}]
272 1) [range] = vector \{ 256, 128, 86, 64 \} element ([floor1\_multiplier]-1)
273 2) vector [floor1\_step2\_flag] element [0] = set
274 3) vector [floor1\_step2\_flag] element [1] = set
275 4) vector [floor1\_final\_Y] element [0] = vector [floor1\_Y] element [0]
276 5) vector [floor1\_final\_Y] element [1] = vector [floor1\_Y] element [1]
277 6) iterate [i] over the range 2 ... [floor1\_values]-1 \{
279 7) [low\_neighbor\_offset] = \link{vorbis:spec:low:neighbor}{low\_neighbor}([floor1\_X\_list],[i])
280 8) [high\_neighbor\_offset] = \link{vorbis:spec:high:neighbor}{high\_neighbor}([floor1\_X\_list],[i])
282 9) [predicted] = \link{vorbis:spec:render:point}{render\_point}( vector [floor1\_X\_list] element [low\_neighbor\_offset],
283 vector [floor1\_final\_Y] element [low\_neighbor\_offset],
284 vector [floor1\_X\_list] element [high\_neighbor\_offset],
285 vector [floor1\_final\_Y] element [high\_neighbor\_offset],
286 vector [floor1\_X\_list] element [i] )
288 10) [val] = vector [floor1\_Y] element [i]
289 11) [highroom] = [range] - [predicted]
290 12) [lowroom] = [predicted]
291 13) if ( [highroom] is less than [lowroom] ) \{
293 14) [room] = [highroom] * 2
295 \} else [highroom] is not less than [lowroom] \{
297 15) [room] = [lowroom] * 2
301 16) if ( [val] is nonzero ) \{
303 17) vector [floor1\_step2\_flag] element [low\_neighbor\_offset] = set
304 18) vector [floor1\_step2\_flag] element [high\_neighbor\_offset] = set
305 19) vector [floor1\_step2\_flag] element [i] = set
306 20) if ( [val] is greater than or equal to [room] ) \{
308 21) if ( [highroom] is greater than [lowroom] ) \{
310 22) vector [floor1\_final\_Y] element [i] = [val] - [lowroom] + [predicted]
312 \} else [highroom] is not greater than [lowroom] \{
314 23) vector [floor1\_final\_Y] element [i] = [predicted] - [val] + [highroom] - 1
318 \} else [val] is less than [room] \{
320 24) if ([val] is odd) \{
322 25) vector [floor1\_final\_Y] element [i] =
323 [predicted] - (([val] + 1) divided by 2 using integer division)
325 \} else [val] is even \{
327 26) vector [floor1\_final\_Y] element [i] =
328 [predicted] + ([val] / 2 using integer division)
334 \} else [val] is zero \{
336 27) vector [floor1\_step2\_flag] element [i] = unset
337 28) vector [floor1\_final\_Y] element [i] = [predicted]
349 \item[step 2: curve synthesis]
351 Curve synthesis generates a return vector \varname{[floor]} of length
352 \varname{[n]} (where \varname{[n]} is provided by the decode process
353 calling to floor decode). Floor 1 curve synthesis makes use of the
354 \varname{[floor1\_X\_list]}, \varname{[floor1\_final\_Y]} and
355 \varname{[floor1\_step2\_flag]} vectors, as well as [floor1\_multiplier]
356 and [floor1\_values] values.
358 Decode begins by sorting the scalars from vectors
359 \varname{[floor1\_X\_list]}, \varname{[floor1\_final\_Y]} and
360 \varname{[floor1\_step2\_flag]} together into new vectors
361 \varname{[floor1\_X\_list]'}, \varname{[floor1\_final\_Y]'} and
362 \varname{[floor1\_step2\_flag]'} according to ascending sort order of the
363 values in \varname{[floor1\_X\_list]}. That is, sort the values of
364 \varname{[floor1\_X\_list]} and then apply the same permutation to
365 elements of the other two vectors so that the X, Y and step2\_flag
368 Then compute the final curve in one pass:
370 \begin{Verbatim}[commandchars=\\\{\}]
373 3) [ly] = vector [floor1\_final\_Y]' element [0] * [floor1\_multiplier]
374 4) iterate [i] over the range 1 ... [floor1\_values]-1 \{
376 5) if ( [floor1\_step2\_flag]' element [i] is set ) \{
378 6) [hy] = [floor1\_final\_Y]' element [i] * [floor1\_multiplier]
379 7) [hx] = [floor1\_X\_list]' element [i]
380 8) \link{vorbis:spec:render:line}{render\_line}( [lx], [ly], [hx], [hy], [floor] )
386 11) if ( [hx] is less than [n] ) \{
388 12) \link{vorbis:spec:render:line}{render\_line}( [hx], [hy], [n], [hy], [floor] )
392 13) if ( [hx] is greater than [n] ) \{
394 14) truncate vector [floor] to [n] elements
398 15) for each scalar in vector [floor], perform a lookup substitution using
399 the scalar value from [floor] as an offset into the vector \link{vorbis:spec:floor1:inverse:dB:table}{[floor1\_inverse\_dB\_static\_table]}