1 <HTML><HEAD><TITLE>xiph.org: Ogg Vorbis documentation</TITLE>
2 <BODY bgcolor="#ffffff" text="#202020" link="#006666" vlink="#000000">
3 <nobr><a href="http://www.xiph.org/ogg/index.html"><img src="white-ogg.png" border=0><img
4 src="vorbisword2.png" border=0></a></nobr><p>
6 <h1><font color=#000070>
7 Ogg logical bitstream framing
10 <em>Last update to this document: July 14, 2002</em><br>
12 <h2>Ogg bitstreams</h2>
14 The Ogg transport bitstream is designed to provide framing, error
15 protection and seeking structure for higher-level codec streams that
16 consist of raw, unencapsulated data packets, such as the Vorbis audio
17 codec or Tarkin video codec.
19 <h2>Application example: Vorbis</h2>
20 Vorbis encodes short-time blocks of PCM data into raw packets of
21 bit-packed data. These raw packets may be used directly by transport
22 mechanisms that provide their own framing and packet-separation
23 mechanisms (such as UDP datagrams). For stream based storage (such as
24 files) and transport (such as TCP streams or pipes), Vorbis uses the
25 Ogg bitstream format to provide framing/sync, sync recapture
26 after error, landmarks during seeking, and enough information to
27 properly separate data back into packets at the original packet
28 boundaries without relying on decoding to find packet boundaries.<p>
30 <h2>Design constraints for Ogg bitstreams</h2>
32 <ol><li>True streaming; we must not need to seek to build a 100%
35 <li> Use no more than approximately 1-2% of bitstream bandwidth for
36 packet boundary marking, high-level framing, sync and seeking.
38 <li> Specification of absolute position within the original sample
41 <li> Simple mechanism to ease limited editing, such as a simplified
42 concatenation mechanism.
44 <li> Detection of corruption, recapture after error and direct, random
45 access to data at arbitrary positions in the bitstream.
48 <h2>Logical and Physical Bitstreams</h2>
50 A <em>logical</em> Ogg bitstream is a contiguous stream of
51 sequential pages belonging only to the logical bitstream. A
52 <em>physical</em> Ogg bitstream is constructed from one or more
53 than one logical Ogg bitstream (the simplest physical bitstream
54 is simply a single logical bitstream). We describe below the exact
55 formatting of an Ogg logical bitstream. Combining logical
56 bitstreams into more complex physical bitstreams is described in the
57 <a href="oggstream.html">Ogg bitstream overview</a>. The exact
58 mapping of raw Vorbis packets into a valid Ogg Vorbis physical
59 bitstream is described in <a href="vorbis-stream.html">Vorbis
60 bitstream mapping</a>.
62 <h2>Bitstream structure</h2>
64 An Ogg stream is structured by dividing incoming packets into
65 segments of up to 255 bytes and then wrapping a group of contiguous
66 packet segments into a variable length page preceded by a page
67 header. Both the header size and page size are variable; the page
68 header contains sizing information and checksum data to determine
69 header/page size and data integrity.<p>
71 The bitstream is captured (or recaptured) by looking for the beginning
72 of a page, specifically the capture pattern. Once the capture pattern
73 is found, the decoder verifies page sync and integrity by computing
74 and comparing the checksum. At that point, the decoder can extract the
75 packets themselves.<p>
77 <h3>Packet segmentation</h3>
79 Packets are logically divided into multiple segments before encoding
80 into a page. Note that the segmentation and fragmentation process is a
81 logical one; it's used to compute page header values and the original
82 page data need not be disturbed, even when a packet spans page
85 The raw packet is logically divided into [n] 255 byte segments and a
86 last fractional segment of < 255 bytes. A packet size may well
87 consist only of the trailing fractional segment, and a fractional
88 segment may be zero length. These values, called "lacing values" are
89 then saved and placed into the header segment table.<p>
91 An example should make the basic concept clear:<p>
96 ___________________________________________
97 |______________packet data__________________| 753 bytes
99 lacing values for page header segment table: 255,255,243
103 We simply add the lacing values for the total size; the last lacing
104 value for a packet is always the value that is less than 255. Note
105 that this encoding both avoids imposing a maximum packet size as well
106 as imposing minimum overhead on small packets (as opposed to, eg,
107 simply using two bytes at the head of every packet and having a max
108 packet size of 32k. Small packets (<255, the typical case) are
109 penalized with twice the segmentation overhead). Using the lacing
110 values as suggested, small packets see the minimum possible
111 byte-aligned overheade (1 byte) and large packets, over 512 bytes or
112 so, see a fairly constant ~.5% overhead on encoding space.<p>
114 Note that a lacing value of 255 implies that a second lacing value
115 follows in the packet, and a value of < 255 marks the end of the
116 packet after that many additional bytes. A packet of 255 bytes (or a
117 multiple of 255 bytes) is terminated by a lacing value of 0:<p>
121 _______________________________
122 |________packet data____________| 255 bytes
124 lacing values: 255, 0
127 Note also that a 'nil' (zero length) packet is not an error; it
128 consists of nothing more than a lacing value of zero in the header.<p>
130 <h3>Packets spanning pages</h3>
132 Packets are not restricted to beginning and ending within a page,
133 although individual segments are, by definition, required to do so.
134 Packets are not restricted to a maximum size, although excessively
135 large packets in the data stream are discouraged; the Ogg
136 bitstream specification strongly recommends nominal page size of
137 approximately 4-8kB (large packets are foreseen as being useful for
138 initialization data at the beginning of a logical bitstream).<p>
140 After segmenting a packet, the encoder may decide not to place all the
141 resulting segments into the current page; to do so, the encoder places
142 the lacing values of the segments it wishes to belong to the current
143 page into the current segment table, then finishes the page. The next
144 page is begun with the first value in the segment table belonging to
145 the next packet segment, thus continuing the packet (data in the
146 packet body must also correspond properly to the lacing values in the
147 spanned pages. The segment data in the first packet corresponding to
148 the lacing values of the first page belong in that page; packet
149 segments listed in the segment table of the following page must begin
150 the page body of the subsequent page).<p>
152 The last mechanic to spanning a page boundary is to set the header
153 flag in the new page to indicate that the first lacing value in the
154 segment table continues rather than begins a packet; a header flag of
155 0x01 is set to indicate a continued packet. Although mandatory, it
156 is not actually algorithmically necessary; one could inspect the
157 preceding segment table to determine if the packet is new or
158 continued. Adding the information to the packet_header flag allows a
159 simpler design (with no overhead) that needs only inspect the current
160 page header after frame capture. This also allows faster error
161 recovery in the event that the packet originates in a corrupt
162 preceding page, implying that the previous page's segment table
163 cannot be trusted.<p>
165 Note that a packet can span an arbitrary number of pages; the above
166 spanning process is repeated for each spanned page boundary. Also a
167 'zero termination' on a packet size that is an even multiple of 255
168 must appear even if the lacing value appears in the next page as a
169 zero-length continuation of the current packet. The header flag
170 should be set to 0x01 to indicate that the packet spanned, even though
171 the span is a nil case as far as data is concerned.<p>
173 The encoding looks odd, but is properly optimized for speed and the
174 expected case of the majority of packets being between 50 and 200
175 bytes (note that it is designed such that packets of wildly different
176 sizes can be handled within the model; placing packet size
177 restrictions on the encoder would have only slightly simplified design
178 in page generation and increased overall encoder complexity).<p>
180 The main point behind tracking individual packets (and packet
181 segments) is to allow more flexible encoding tricks that requiring
182 explicit knowledge of packet size. An example is simple bandwidth
183 limiting, implemented by simply truncating packets in the nominal case
184 if the packet is arranged so that the least sensitive portion of the
189 The headering mechanism is designed to avoid copying and re-assembly
190 of the packet data (ie, making the packet segmentation process a
191 logical one); the header can be generated directly from incoming
192 packet data. The encoder buffers packet data until it finishes a
193 complete page at which point it writes the header followed by the
194 buffered packet segments.<p>
196 <h4>capture_pattern</h4>
198 A header begins with a capture pattern that simplifies identifying
199 pages; once the decoder has found the capture pattern it can do a more
200 intensive job of verifying that it has in fact found a page boundary
201 (as opposed to an inadvertent coincidence in the byte stream).<p>
212 <h4>stream_structure_version</h4>
214 The capture pattern is followed by the stream structure revision:
222 <h4>header_type_flag</h4>
224 The header type flag identifies this page's context in the bitstream:
229 5 bitflags: 0x01: unset = fresh packet
230 set = continued packet
231 0x02: unset = not first page of logical bitstream
232 set = first page of logical bitstream (bos)
233 0x04: unset = not last page of logical bitstream
234 set = last page of logical bitstream (eos)
237 <h4>absolute granule position</h4>
239 (This is packed in the same way the rest of Ogg data is packed; LSb
240 of LSB first. Note that the 'position' data specifies a 'sample'
241 number (eg, in a CD quality sample is four octets, 16 bits for left
242 and 16 bits for right; in video it would likely be the frame number.
243 It is up to the specific codec in use to define the semantic meaning
244 of the granule position value). The position specified is the total
245 samples encoded after including all packets finished on this page
246 (packets begun on this page but continuing on to the next page do not
247 count). The rationale here is that the position specified in the
248 frame header of the last page tells how long the data coded by the
249 bitstream is. A truncated stream will still return the proper number
250 of samples that can be decoded fully.
252 A special value of '-1' (in two's complement) indicates that no packets
268 <h4>stream serial number</h4>
270 Ogg allows for separate logical bitstreams to be mixed at page
271 granularity in a physical bitstream. The most common case would be
272 sequential arrangement, but it is possible to interleave pages for
273 two separate bitstreams to be decoded concurrently. The serial
274 number is the means by which pages physical pages are associated with
275 a particular logical stream. Each logical stream must have a unique
276 serial number within a physical stream:
287 <h4>page sequence no</h4>
289 Page counter; lets us know if a page is lost (useful where packets
290 span page boundaries).
301 <h4>page checksum</h4>
303 32 bit CRC value (direct algorithm, initial val and final XOR = 0,
304 generator polynomial=0x04c11db7). The value is computed over the
305 entire header (with the CRC field in the header set to zero) and then
306 continued over the page. The CRC field is then filled with the
309 (A thorough discussion of CRC algorithms can be found in <a
310 href="ftp://ftp.rocksoft.com/papers/crc_v3.txt">"A
311 Painless Guide to CRC Error Detection Algorithms"</a> by Ross
313 href="mailto:ross@guest.adelaide.edu.au">ross@guest.adelaide.edu.au</a>.)
324 <h4>page_segments</h4>
326 The number of segment entries to appear in the segment table. The
327 maximum number of 255 segments (255 bytes each) sets the maximum
328 possible physical page size at 65307 bytes or just under 64kB (thus
329 we know that a header corrupted so as destroy sizing/alignment
330 information will not cause a runaway bitstream. We'll read in the
331 page according to the corrupted size information that's guaranteed to
332 be a reasonable size regardless, notice the checksum mismatch, drop
333 sync and then look for recapture).<p>
341 <h4>segment_table (containing packet lacing values)</h4>
343 The lacing values for each packet segment physically appearing in
344 this page are listed in contiguous order.
351 n 0x00-0xff (0-255, n=page_segments+26)
354 Total page size is calculated directly from the known header size and
355 lacing values in the segment table. Packet data segments follow
356 immediately after the header.<p>
358 Page headers typically impose a flat .25-.5% space overhead assuming
359 nominal ~8k page sizes. The segmentation table needed for exact
360 packet recovery in the streaming layer adds approximately .5-1%
361 nominal assuming expected encoder behavior in the 44.1kHz, 128kbps
365 <a href="http://www.xiph.org/">
366 <img src="white-xifish.png" align=left border=0>
368 <font size=-2 color=#505050>
370 Ogg is a <a href="http://www.xiph.org">Xiph.org Foundation</a> effort
371 to protect essential tenets of Internet multimedia from corporate
372 hostage-taking; Open Source is the net's greatest tool to keep
373 everyone honest. See <a href="http://www.xiph.org/about.html">About
374 the Xiph.org Foundation</a> for details.
377 Ogg Vorbis is the first Ogg audio CODEC. Anyone may freely use and
378 distribute the Ogg and Vorbis specification, whether in a private,
379 public or corporate capacity. However, the Xiph.org Foundation and
380 the Ogg project (xiph.org) reserve the right to set the Ogg Vorbis
381 specification and certify specification compliance.<p>
383 Xiph.org's Vorbis software CODEC implementation is distributed under a
384 BSD-like license. This does not restrict third parties from
385 distributing independent implementations of Vorbis software under
388 Ogg, Vorbis, Xiph.org Foundation and their logos are trademarks (tm)
389 of the <a href="http://www.xiph.org/">Xiph.org Foundation</a>. These
390 pages are copyright (C) 1994-2002 Xiph.org Foundation. All rights
projects / lugaru.git / blob