3GP文件格式分析
原文地址 http://blog.csdn.net/dinggo/archive/2007/12/29/2003425.aspx
1. 概述
现在很多智能手机都支持多媒体功能，特别是音频和视频播放功能，而3GP文件格式是手机端普遍支持的视频文件格式。
目前很多手机都支持h263视频编码格式的视频文件播放，还有些手机支持h264。音频文件格式普遍支持amr，有些手机还支持AAC音频格式。
鉴于凤凰卫视节目主要以新闻类为主，音频压缩采用AMR标准，视频压缩采用H263和H264标准。
|
编号
|
文件名
|
文件长度
|
时长
|
|
1
|
news_h263.3gp
|
1.38MB
|
01:21.90
|
|
2
|
zbc1217_h263.3gp
|
1.92MB
|
02:37.98
|
|
3
|
zbc1217_h264.3gp
|
2.44MB
|
02:37.98
|
|
|
|
|
|
2. 3GP文件结构分析示例
AtomicParsley是一个开源的多媒体文件结构分析器。
下面是用该程序分析的两个3GP文件结构：
news_h263.3gp
Atom ftyp @ 0 of size: 28, ends @ 28
Atom mdat @ 28 of size: 65, ends @ 93
Atom mdat @ 93 of size: 130856, ends @ 130949
Atom mdat @ 130949 of size: 1312119, ends @ 1443068
Atom moov @ 1443068 of size: 7136, ends @ 1450204
Atom mvhd @ 1443076 of size: 108, ends @ 1443184
Atom trak @ 1443184 of size: 1546, ends @ 1444730
Atom tkhd @ 1443192 of size: 92, ends @ 1443284
Atom udta @ 1443284 of size: 48, ends @ 1443332
Atom uuid=484d496e-6354-7261-636b-496e666f2020 @ 1443292 of size: 40, ends @ 1443332
Atom mdia @ 1443332 of size: 1398, ends @ 1444730
Atom mdhd @ 1443340 of size: 32, ends @ 1443372
Atom hdlr @ 1443372 of size: 37, ends @ 1443409
Atom minf @ 1443409 of size: 1321, ends @ 1444730
Atom smhd @ 1443417 of size: 16, ends @ 1443433
Atom dinf @ 1443433 of size: 36, ends @ 1443469
Atom dref @ 1443441 of size: 28, ends @ 1443469
Atom stbl @ 1443469 of size: 1261, ends @ 1444730
Atom stts @ 1443477 of size: 24, ends @ 1443501
Atom stsd @ 1443501 of size: 69, ends @ 1443570
Atom samr @ 1443517 of size: 53, ends @ 1443570
Atom damr @ 1443553 of size: 17, ends @ 1443570
Atom stsz @ 1443570 of size: 1112, ends @ 1444682
Atom stsc @ 1444682 of size: 28, ends @ 1444710
Atom stco @ 1444710 of size: 20, ends @ 1444730
Atom trak @ 1444730 of size: 5474, ends @ 1450204
Atom tkhd @ 1444738 of size: 92, ends @ 1444830
Atom tref @ 1444830 of size: 20, ends @ 1444850
Atom sync @ 1444838 of size: 12, ends @ 1444850
Atom udta @ 1444850 of size: 48, ends @ 1444898
Atom uuid=484d496e-6354-7261-636b-496e666f2020 @ 1444858 of size: 40, ends @ 1444898
Atom mdia @ 1444898 of size: 5306, ends @ 1450204
Atom mdhd @ 1444906 of size: 32, ends @ 1444938
Atom hdlr @ 1444938 of size: 37, ends @ 1444975
Atom minf @ 1444975 of size: 5229, ends @ 1450204
Atom vmhd @ 1444983 of size: 20, ends @ 1445003
Atom dinf @ 1445003 of size: 36, ends @ 1445039
Atom dref @ 1445011 of size: 28, ends @ 1445039
Atom stbl @ 1445039 of size: 5165, ends @ 1450204
Atom stts @ 1445047 of size: 24, ends @ 1445071
Atom stsd @ 1445071 of size: 117, ends @ 1445188
Atom s263 @ 1445087 of size: 101, ends @ 1445188
Atom d263 @ 1445173 of size: 15, ends @ 1445188
Atom stsz @ 1445188 of size: 4920, ends @ 1450108
Atom stss @ 1450108 of size: 48, ends @ 1450156
Atom stsc @ 1450156 of size: 28, ends @ 1450184
Atom stco @ 1450184 of size: 20, ends @ 1450204
------------------------------------------------------
Total size: 1450204 bytes; 46 atoms total. AtomicParsley version: 0.9.0 (utf16)
Media data: 1442975 bytes; 7229 bytes all other atoms (0.498% atom overhead).
Total free atom space: 0 bytes; 0.000% waste.
------------------------------------------------------
Movie duration: 81.900 seconds (01:21.90) - 140.95* kbp/sec bitrate (*=approximate)
Low-level details. Total tracks: 2
Trk Type Handler Kind Lang Bytes
1 soun soun samr eng 130848 Encoder: NXTR
12.78* kbp/s 81.900 sec AMR Narrow-Band.
Modes: 0, 1, 2, 3, 4, 5, 6, 7, 8. Encoder vendor code: NXTR
2 vide vide s263 eng 1312111 Encoder: NXTR
128.41* kbp/s 81.748 sec H.263 Baseline Profile, Level 10. Encoder vendor code: NXTR
该3GP文件结构如下：
zbc1217_h263.3gp
Atom ftyp @ 0 of size: 24, ends @ 24
Atom mdat @ 24 of size: 1989236, ends @ 1989260
Atom moov @ 1989260 of size: 14951, ends @ 2004211
Atom mvhd @ 1989268 of size: 108, ends @ 1989376
Atom iods @ 1989376 of size: 33, ends @ 1989409
Atom udta @ 1989409 of size: 1069, ends @ 1990478
Atom rtng [eng] @ 1989417 of size: 29, ends @ 1989446
Atom uuid=f78caa0c-36be-4ce9-88d2-03c256dabeb2 @ 1989446 of size: 1032, ends @ 1990478
Atom trak @ 1990478 of size: 9978, ends @ 2000456
Atom tkhd @ 1990486 of size: 92, ends @ 1990578
Atom mdia @ 1990578 of size: 9842, ends @ 2000420
Atom mdhd @ 1990586 of size: 32, ends @ 1990618
Atom hdlr @ 1990618 of size: 33, ends @ 1990651
Atom minf @ 1990651 of size: 9769, ends @ 2000420
Atom vmhd @ 1990659 of size: 20, ends @ 1990679
Atom dinf @ 1990679 of size: 36, ends @ 1990715
Atom dref @ 1990687 of size: 28, ends @ 1990715
Atom stbl @ 1990715 of size: 9705, ends @ 2000420
Atom stsd @ 1990723 of size: 117, ends @ 1990840
Atom s263 @ 1990739 of size: 101, ends @ 1990840
Atom d263 @ 1990825 of size: 15, ends @ 1990840
Atom stts @ 1990840 of size: 32, ends @ 1990872
Atom stsc @ 1990872 of size: 1900, ends @ 1992772
Atom stsz @ 1992772 of size: 6300, ends @ 1999072
Atom stco @ 1999072 of size: 644, ends @ 1999716
Atom stss @ 1999716 of size: 704, ends @ 2000420
Atom udta @ 2000420 of size: 36, ends @ 2000456
Atom uuid=f78caa0c-36be-4ce9-87d2-03c256dabeb2 @ 2000428 of size: 28, ends @ 2000456
Atom trak @ 2000456 of size: 2934, ends @ 2003390
Atom tkhd @ 2000464 of size: 92, ends @ 2000556
Atom mdia @ 2000556 of size: 2798, ends @ 2003354
Atom mdhd @ 2000564 of size: 32, ends @ 2000596
Atom hdlr @ 2000596 of size: 33, ends @ 2000629
Atom minf @ 2000629 of size: 2725, ends @ 2003354
Atom smhd @ 2000637 of size: 16, ends @ 2000653
Atom dinf @ 2000653 of size: 36, ends @ 2000689
Atom dref @ 2000661 of size: 28, ends @ 2000689
Atom stbl @ 2000689 of size: 2665, ends @ 2003354
Atom stsd @ 2000697 of size: 69, ends @ 2000766
Atom samr @ 2000713 of size: 53, ends @ 2000766
Atom damr @ 2000749 of size: 17, ends @ 2000766
Atom stts @ 2000766 of size: 24, ends @ 2000790
Atom stsc @ 2000790 of size: 1900, ends @ 2002690
Atom stsz @ 2002690 of size: 20, ends @ 2002710
Atom stco @ 2002710 of size: 644, ends @ 2003354
Atom udta @ 2003354 of size: 36, ends @ 2003390
Atom uuid=f78caa0c-36be-4ce9-87d2-03c256dabeb2 @ 2003362 of size: 28, ends @ 2003390
Atom trak @ 2003390 of size: 420, ends @ 2003810
Atom tkhd @ 2003398 of size: 92, ends @ 2003490
Atom mdia @ 2003490 of size: 296, ends @ 2003786
Atom mdhd @ 2003498 of size: 32, ends @ 2003530
Atom hdlr @ 2003530 of size: 33, ends @ 2003563
Atom minf @ 2003563 of size: 223, ends @ 2003786
Atom nmhd @ 2003571 of size: 12, ends @ 2003583
Atom dinf @ 2003583 of size: 36, ends @ 2003619
Atom dref @ 2003591 of size: 28, ends @ 2003619
Atom stbl @ 2003619 of size: 167, ends @ 2003786
Atom stsd @ 2003627 of size: 67, ends @ 2003694
Atom mp4s @ 2003643 of size: 51, ends @ 2003694
Atom esds @ 2003659 of size: 35, ends @ 2003694
Atom stts @ 2003694 of size: 24, ends @ 2003718
Atom stsc @ 2003718 of size: 28, ends @ 2003746
Atom stsz @ 2003746 of size: 20, ends @ 2003766
Atom stco @ 2003766 of size: 20, ends @ 2003786
Atom tref @ 2003786 of size: 24, ends @ 2003810
Atom mpod @ 2003794 of size: 16, ends @ 2003810
Atom trak @ 2003810 of size: 401, ends @ 2004211
Atom tkhd @ 2003818 of size: 92, ends @ 2003910
Atom mdia @ 2003910 of size: 301, ends @ 2004211
Atom mdhd @ 2003918 of size: 32, ends @ 2003950
Atom hdlr @ 2003950 of size: 33, ends @ 2003983
Atom minf @ 2003983 of size: 228, ends @ 2004211
Atom nmhd @ 2003991 of size: 12, ends @ 2004003
Atom dinf @ 2004003 of size: 36, ends @ 2004039
Atom dref @ 2004011 of size: 28, ends @ 2004039
Atom stbl @ 2004039 of size: 172, ends @ 2004211
Atom stsd @ 2004047 of size: 72, ends @ 2004119
Atom mp4s @ 2004063 of size: 56, ends @ 2004119
Atom esds @ 2004079 of size: 40, ends @ 2004119
Atom stts @ 2004119 of size: 24, ends @ 2004143
Atom stsc @ 2004143 of size: 28, ends @ 2004171
Atom stsz @ 2004171 of size: 20, ends @ 2004191
Atom stco @ 2004191 of size: 20, ends @ 2004211
------------------------------------------------------
Total size: 2004211 bytes; 82 atoms total. AtomicParsley version: 0.9.0 (utf16)
Media data: 1989236 bytes; 14975 bytes all other atoms (0.747% atom overhead).
Total free atom space: 0 bytes; 0.000% waste.
------------------------------------------------------
Movie duration: 156.980 seconds (02:36.98) - 101.38* kbp/sec bitrate (*=approximate)
Low-level details. Total tracks: 4
Trk Type Handler Kind Lang Bytes
1 vide [none listed] s263 und 1738018 Encoder: ViVi
88.58* kbp/s 156.960 sec H.263 Baseline Profile, Level 20. Encoder vendor code: ViVi
2 soun [none listed] samr und 251168 Encoder: ViVi
12.80* kbp/s 156.980 sec AMR Narrow-Band. Modes: 7. Encoder vendor code: ViVi
3 odsm [none listed] mp4s und 18
4 sdsm [none listed] mp4s und 24
该3GP文件结构如下：
其实3GP文件就是一个对象（box）容器，对象还允许包含子对象。
对象格式如下：
struct
{
UINT32
size; // 对象大小
char
type; // 对象类型
void*
data; // 对象数据（属性），大小为(size-8)
};
};
《
ISO base media file format
》说明不同
box
对象的结构定义细节。
3. 3GP文件对象实例分析
l
ftype
File Type（该对象仅有一个）
记录该文件使用的规范和版本等信息。
|
属性
|
大小
|
值
|
说明
|
|
size
|
4
|
0x0000001c = 28
|
|
|
ftyp
|
4
|
ftyp
|
|
|
major_brand
|
4
|
3gp4
|
主类型
|
|
minor_version
|
4
|
|
最小版本号
|
|
compatible_brands[]
|
string
|
3gp4 | 3gp5 | 3g2a
|
兼容类型
|
l
mdat
Media Data
mdat
存放媒体数据（
media data
）。一般包括文字、音频和视频等。这些
mdat
的顺序不固定。
当前
mdat
对象包含的是纯文字。
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
0x00000041 = 65
|
|
|
mdat
|
4
|
|
|
|
|
string
|
|
|
这个
box
可以忽略。
l
mdat
音频
数据块
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
0x0001FF28 = 130856
|
|
|
mdat
|
4
|
|
|
|
|
32*N
|
3C
打头
|
每个音频帧
32
字节
|
amr
音频数据以
32
字节一帧，每帧
20
毫秒。所以根据
amr
数据包大小就能估算出语音的时间长度。
(
130856
– 8) / 32 = 4089 * 20 = 81780 ms = 817.80
秒
= 01:21.90
这个时间与播放器上显示的时间基本相同。
注意：
这里真实的视频数据起始位置为
0x000005D
。
在音频
trak
里面的
stso
对象里面的
chunk_offset
记录的就是这个值。
在
RFC3267
中有如下描述：
The sampling frequency used in AMR is 8000 Hz and the speech encoding is
performed on 20 ms speech frames. Therefore, each encoded AMR speech
frame represents 160 samples of the original speech.
AMR
语音采样频率为
8Khz
，每帧对应
20ms
。
l
mdat
视频
数据块
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
0x004946F1 = 4802289
|
|
|
mdat
|
4
|
|
|
|
|
4
|
|
H263
视频数据
|
实际视频数据大小：
4802289 – 8 = 4802281
《
H263
协议中文版》第
8
页
5.1
图象层中描述：
每帧图象的数据包含一个图象头，并紧跟着块组数据，最后是一个
end-of-sequence
码和填塞位。
图象开始码
(PSC) (22 bits)
PSC
是一个
22
比特的字。它的值是
0000 0000 0000 0000 1000
00
-
（
00 00 80 00 …z
）
时域参考
(TR) (8 bits)
00
-00 0000 00-
类型信息
(PTYPE) (13 bits)
00-10 0000 1000 000-0
位
6-8
：
010
：
表示图像为
QCIF
。
位
9
：图象编码类型，
“0” INTRA (I-picture)
，
“1” INTER (P-picture)
。关键帧标志位，
0
：关键帧，
1
：
p
帧。
l
moov
Movie Box
该对象保留了
3GP
视频的
metadata
，即该
3GP
的全部描述信息。
这个对象一般在文件开始或者末尾。
下面详细分析
moov
的各个子对象。
l
mvhd
Movie Head Box
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
0x0000006C = 108
|
|
|
mvhd
|
4
|
|
|
|
version
|
1
|
0
|
|
|
flag
|
3
|
0
|
|
|
creation_time
|
4
|
|
|
|
modification_time
|
4
|
|
|
|
timescale
|
4
|
0x000003E8 = 1000
|
时间缩放因子
|
|
duration
|
4
|
0x00013FEC = 81900 = 81.9s
|
持续时间
|
|
rate
|
4
|
0x00010000 1.0
（即全速播放）
|
播放比率
|
|
volume
|
2
|
0x0100 1.0
（音量）
|
音量
|
|
reserved
|
|
|
保留
|
|
reserved
|
|
|
保留
|
|
matrix
|
|
|
视频转换矩阵
|
|
pre_defined
|
|
|
预定义
(
全
0)
|
|
next_track_ID
|
4
|
0x000000CA = 202
|
下一个
track_id
|
trak
(
音频
)
Track Box
l
tkhd
Track Header Box
定义了该
track
的属性。
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
0x0000005C = 92
|
|
|
tkhd
|
4
|
|
|
|
version
|
1
|
0
|
|
|
flag
|
3
|
1
|
|
|
creation_time
|
4
|
|
|
|
modification_time
|
4
|
|
|
|
track_id
|
4
|
0x00000065 = 101
|
|
|
reserved
|
4
|
全
0
|
|
|
duration
|
4
|
0x00013FEC = 81900 = 81.9s
|
音频播放时间
|
|
reserved
|
8
|
|
|
|
layer
|
2
|
|
|
|
alternate_group
|
2
|
|
|
|
volume
|
2
|
0x0100
|
如果
track
是音频则为
0x0100
，视频则为
0x0000
。
|
|
reserved
|
2
|
0
|
|
|
matrix
|
4 * 9
|
|
视频转换矩阵
|
|
width
|
4
|
0
|
音频的
w
和
h
为
0
|
|
height
|
4
|
0
|
|
注意：如果
track
是音频
volume
则为
0x0100
，视频
volume
则为
0x0000
。
l
mdhd
Media Header Box
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
32
|
|
|
mdhd
|
4
|
|
|
|
version
|
1
|
|
|
|
flag
|
3
|
|
|
|
creation_time
|
4
|
|
|
|
modification_time
|
4
|
|
|
|
timescale
|
4
|
0x00001F40 = 8000
|
时间缩放因子
|
|
duration
|
4
|
0x0009FF60 = 655200/8000=81.9
|
音频播放时长
|
|
pad
|
1bit
|
|
|
|
language
|
15bit
|
0x15c7 ISO-639-2/T language code
|
|
|
pre_defined
|
2
|
|
|
l
hdlr
Handler Reference Box
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
0x00000025 = 37
|
|
|
hdlr
|
4
|
|
|
|
version
|
1
|
|
|
|
flag
|
3
|
|
|
|
pre_defined
|
4
|
0
|
|
|
handler_type
|
4
|
soun
|
vide
：
Video track
soun
：
Audio track
|
|
reserved
|
4 * 3
|
0
|
|
|
handler_type
|
string
（
4+1
）
|
soun
|
以
’/0’
结束
|
l
stbl
Sample Table Box
l
stts
Decoding Time to Sample Box
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
0x00000018 = 24
|
|
|
stts
|
4
|
|
|
|
version
|
1
|
|
|
|
flag
|
3
|
|
|
|
entry_count
|
4
|
1
|
|
|
sample_count
|
4
|
0x00000111 = 273
|
|
|
sample_delta
|
4
|
0x00000960 = 2400
|
|
|
|
|
|
|
说明：该
amr
音频数据里面有
273
个
sample
，每个
sample
的时间为
2400
，而音频的时间因子为
8000
（见
5.2.3
.1 mdhd
的
timescale
），所以每个
sample
的时长
= 2400 / 8000s = 0.3s
。
l
stsd
Sample Description Box
|
|
大小
|
值
|
其他说明
|
|
size
|
4
|
0x00000045 = 69
|
|
|
stsd
|
4
|
|
|
|
version
|