Debian Codegraph @

libtoolame-dab/html/vbr.html

raw
1 <html>
2 <head>
3 <title>tooLAME: MPEG Audio Layer II VBR</title>
4 <style>
5 <!-- BODY { BACKGROUND: #FFFFFF; COLOR: #000000; FONT-SIZE: 10pt; FONT-FAMILY: verdana, sans-serif }
6 A { COLOR: #111177; TEXT-DECORATION: none }
7 TD { font-size: medium; font-weight:normal }
8 --!>
9 </STYLE>
10 </head>
11 <body>
12
13 <table border = 0 width="75%" align="center"><tr><td>
14 <h1> tooLAME: MPEG Audio Layer II VBR </h1>
15
16 <h2>Contents</h2>
17 <Ul>
18 <LI>Introduction
19 <LI>Usage
20 <LI>Bitrate Ranges for various Sampling frequencies
21 <LI>Why can't the bitrate vary from 32kbps to 384kbps for every file?
22 <UL>
23 <LI>Short Answer
24 <LI>Long Answer
25 </UL>
26 <LI> Tech Stuff
27 </UL>
28
29 <h2>Introduction</h2>
30 VBR mode works by selecting a different bitrate for each frame. Frames
31 which are harder to encode will be allocated more bits i.e. a higher bitrate.</p>
32
33 LayerII VBR is a complete hack - the ISO standard actually says that decoders are not
34 required to support it. As a hack, its implementation is a pain to try and understand.
35 If you're mega-keen to get full range VBR working, either (a) send me money (b) grab the
36 ISO standard and a C compiler and email me.</p>
37
38 <h2>Usage</h2>
39 <pre>
40 toolame -v [level] inputfile outputfile.
41 </pre>
42 A level of 5 works very well for me.</p>
43
44 The level value can is a measurement of quality - the higher
45 the level the higher the average bitrate of the resultant file.
46 [See TECH STUFF for a better explanation of what the value does]</p>
47
48 The confusing part of my implementation of LayerII VBR is that it's different from MP3 VBR.
49 <UL>
50 <LI>The range of bitrates used is controlled by the input sampling frequency. (See below "Bitrate ranges")
51 <LI>The tendency to use higher bitrates is governed by the <level>.
52 </ul>
53
54 E.g. Say you have a 44.1kHz Stereo file. In VBR mode, the bitrate can range from 192 to 384 kbps.
55 Using "-v -5" will force the encoder to favour the lower bitrate.
56 Using "-v 5" will force the encoder to favour the upper bitrate.
57 The value can actually be *any* int. -27, 233, 47. The larger the number, the greater
58 the bitrate bias.</p>
59
60 <h2>Bitrate Ranges</h2>
61
62 When making a VBR stream, the bitrate is only allowed to vary within
63 set limits</p>
64
65 <pre>
66 48kHz
67 Stereo: 112-384kbps Mono: 56-192kbps
68
69 44.1kHz & 32kHz
70 Stereo: 192-384kbps Mono: 96-192kbps
71
72 24kHz, 22.05kHz & 16kHz
73 Stereo/Mono: 8-160kbps
74 </pre>
75
76 <h2>Why doesn't the VBR mode work the same as MP3VBR? The Short Answer</h2>
77 <b>Why can't the bitrate vary from 32kbps to 384kbps for every file?</b></p>
78 According to the standard (ISO/IEC 11172-3:1993) Section 2.4.2.3
79 <pre>
80 "In order to provide the smallest possible delay and complexity, the
81 decoder is not required to support a continuously variable bitrate when
82 in layer I or II. Layer III supports variable bitrate by switching the
83 bitrate index."
84
85 and
86
87 "For Layer II, not all combinations of total bitrate and mode are allowed."
88 </pre>
89
90 Hence, most LayerII coders would not have been written with VBR in mind, and
91 LayerII VBR is a hack. It works for limited cases. Getting it to work to
92 the same extent as MP3-style VBR will be a major hack.</p>
93
94 (If you *really* want better bitrate ranges, read "The Long Answer" and submit your mega-patch.)</p>
95
96 <h2>Why doesn't the VBR mode work the same as MP3VBR? The Long Answer</h2>
97 <b>Why can't the bitrate vary from 32kbps to 384kbps for every file?</b>
98
99 <h3>Reason 1: The standard limits the range</h3>
100
101 As quoted above from the standard for 48/44.1/32kHz:
102 <pre>
103 "For Layer II, not all combinations of total bitrate and mode are allowed. See
104 the following table."
105
106 Bitrate Allowed Modes
107 (kbps)
108 32 mono only
109 48 mono only
110 56 mono only
111 64 all modes
112 80 mono only
113 96 all modes
114 112 all modes
115 128 all modes
116 160 all modes
117 192 all modes
118 224 stereo only
119 256 stereo only
120 320 stereo only
121 384 stereo only
122 </pre>
123
124 So based upon this table alone, you *could* have VBR stereo encoding which varies
125 smoothly from 96 to 384kbps. Or you could have have VBR mono encoding which varies from
126 32 to 192kbps. But since the top and bottom bitrates don't apply to all modes, it would
127 be impossible to have a stereo file encoded from 32 to 384 kbps.</p>
128
129 But this isn't what is really limiting the allowable bitrate range - the bit allocation
130 tables are the major hurdle.</p>
131
132 <h3>Reason 2: The bit allocation tables don't allow it</h3>
133
134 From the standard, Section 2.4.3.3.1 "Bit allocation decoding"</p>
135 <pre>
136 "For different combinations of bitrate and sampling frequency, different bit
137 allocation tables exist.
138 </pre>
139 These bit allocation tables are pre-determined tables (in Annex B of the standard) which
140 indicate
141 <UL>
142 <LI>how many bits to read for the initial data (2,3 or 4)
143 <LI>these bits are then used as an index back into the table to
144 find the number of quantize levels for the samples in this subband
145 </ul>
146 But the table used (and hence the number of bits and the calculated index) are different
147 for different combinations of bitrate and sampling frequency.</p>
148
149 I will use TableB.2a as an example.</p>
150
151 Table B.2a Applies for the following combinations.
152 <pre>
153 Sampling Freq Bitrates in (kbps/channel) [emphasis: this is a PER CHANNEL bitrate]
154 48 56, 64, 80, 96, 112, 128, 160, 192
155 44.1 56, 64, 80
156 32 56, 64, 80
157 </pre>
158 If we have a STEREO 48kHz input file, and we use this table, then the bitrates
159 we could calculate from this would be 112, 128, 160, 192, 224, 256, 320 and 384 kbps.</p>
160
161 This table contains no information on how to encode stuff at bitrates less than 112kbps
162 (for a stereo file). You would have to load allocation table B.2c to encode stereo at
163 64kbps and 128kbps.</p>
164
165 Since it would be a MAJOR piece of hacking to get the different tables shifted in and out
166 during the encoding process, once an allocation table is loaded *IT IS NOT CHANGED*.</p>
167
168 Hence, the best table is picked at the start of the encoding process, and the encoder
169 is stuck with it for the rest of the encode. </p>
170
171 For toolame-02j, I have picked the table it loads for different
172 sampling frequencies in order to optimize the range of bitrates possible.
173 <pre>
174 48 kHz - Table B.2a
175 Stereo Bitrate Range: 112 - 384
176 Mono Bitrate Range : 56 - 192
177
178 44.1/32 kHz - Table B.2b
179 Stereo Bitrate Range: 192 - 384
180 Mono Bitrate Range: 96 - 192
181
182 24/22.05/16 kHz - LSF Table (Standard ISO/IEC 13818.3:1995 Annex B, Table B.1)
183 There is only 1 table for the Lower Sampling Frequencies
184 All modes (mono and stereo) are allowable at all bitrates
185 So at the Lower Sampling Frequencies you *can* have a completely variable
186 bitrate over the entire range.
187 </pre>
188 <h2>Tech Stuff</h2>
189
190 The VBR mode is mainly centered around the main_bit_allocation() and
191 a_bit_allocation() routines in encode.c.</p>
192
193 The limited range of VBR is due to my particular implementation which restricts
194 ranges to within one alloc table (see tables B.2a, B.2b, B.2c and B.2d in ISO 11172).
195 The VBR range for 32/44.1khz lies within B.2b, and the 48khz VBR lies within table B.2a.</p>
196
197 I'm not sure whether it is worth extending these ranges down to lower bitrates.
198 The work required to switch alloc tables *during* the encoding is major.</p>
199
200 In the case of silence, it might be worth doing a quick check for very low signals
201 and writing a pre-calculated *blank* 32kpbs frame. [probably also a lot of work].</p>
202
203 <h3>How CBR works</h3>
204 <UL>
205 <LI> Use the psycho model to determine the MNRs for each subband
206 [MNR = the ratio of "masking" to "noise"]
207 (From an encoding perspective, a bigger MNR in a subband means that
208 it sounds better since the noise is more masked))
209 <LI> calculate the available data bits (adb) for this bitrate.
210 <LI> Based upon the MNR (Masking:Noise Ratio) values, allocate bits to each
211 subband
212 <LI> Keep increasing the bits to whichever subband currently has the min MNR
213 value until we have no bits left.
214 <LI> This mode does not guarentee that all the subbands are without noise
215 ie there may still be subbands with MNR less than 0.0 (noisy!)
216 </ul>
217
218 <h3>How VBR works</h3>
219 <UL>
220 <LI> pretend we have lots of bits to spare, and work out the bits which would
221 raise the MNR in each subband to the level given by the argument on the
222 command line "-v [int]"
223 <LI> Pick the bitrate which has more bits than the required_bits we just calculated
224 <LI> calculate a_bit_allocation()
225 <LI> VBR "guarantees" that all subbands have MNR > VBRLEVEL or that we have
226 reached the maximum bitrate.
227 </ul>
228
229 <h2>FUTURE</h2>
230 <UL>
231 <LI> with this VBR mode, we know the bits aren't going to run out, so we can
232 just assign them "greedily".
233 <LI> VBR_a_bit_allocation() is yet to be written :)
234 </ul>
235
236
237 </tr></td>
238 </body>
239 </html>