<s>
MPEG-1	B-Algorithm
is	O
a	O
standard	O
for	O
lossy	B-Algorithm
compression	I-Algorithm
of	O
video	O
and	O
audio	O
.	O
</s>
<s>
Today	O
,	O
MPEG-1	B-Algorithm
has	O
become	O
the	O
most	O
widely	O
compatible	O
lossy	B-Algorithm
audio/video	O
format	O
in	O
the	O
world	O
,	O
and	O
is	O
used	O
in	O
a	O
large	O
number	O
of	O
products	O
and	O
technologies	O
.	O
</s>
<s>
Perhaps	O
the	O
best-known	O
part	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
is	O
the	O
first	O
version	O
of	O
the	O
MP3	B-Application
audio	I-Application
format	O
it	O
introduced	O
.	O
</s>
<s>
The	O
MPEG-1	B-Algorithm
standard	O
is	O
published	O
as	O
ISO/IEC	B-Algorithm
11172	I-Algorithm
–	O
Information	O
technology	O
—	O
Coding	O
of	O
moving	O
pictures	O
and	O
associated	O
audio	O
for	O
digital	O
storage	O
media	O
at	O
up	O
to	O
about	O
1.5	O
Mbit/s	O
.	O
</s>
<s>
The	O
predecessor	O
of	O
MPEG-1	B-Algorithm
for	O
video	O
coding	O
was	O
the	O
H.261	B-Algorithm
standard	I-Algorithm
produced	O
by	O
the	O
CCITT	O
(	O
now	O
known	O
as	O
the	O
ITU-T	O
)	O
.	O
</s>
<s>
The	O
basic	O
architecture	O
established	O
in	O
H.261	B-Algorithm
was	O
the	O
motion-compensated	B-Algorithm
DCT	B-General_Concept
hybrid	O
video	O
coding	O
structure	O
.	O
</s>
<s>
It	O
uses	O
macroblocks	B-General_Concept
of	O
size	O
16×16	O
with	O
block-based	O
motion	O
estimation	O
in	O
the	O
encoder	O
and	O
motion	B-Algorithm
compensation	I-Algorithm
using	O
encoder-selected	O
motion	O
vectors	O
in	O
the	O
decoder	B-General_Concept
,	O
with	O
residual	O
difference	O
coding	O
using	O
a	O
discrete	B-General_Concept
cosine	I-General_Concept
transform	I-General_Concept
(	O
DCT	B-General_Concept
)	O
of	O
size	O
8×8	O
,	O
scalar	B-Algorithm
quantization	I-Algorithm
,	O
and	O
variable-length	O
codes	O
(	O
like	O
Huffman	B-General_Concept
codes	I-General_Concept
)	O
for	O
entropy	B-Algorithm
coding	I-Algorithm
.	O
</s>
<s>
H.261	B-Algorithm
was	O
the	O
first	O
practical	O
video	O
coding	O
standard	O
,	O
and	O
all	O
of	O
its	O
described	O
design	O
elements	O
were	O
also	O
used	O
in	O
MPEG-1	B-Algorithm
.	O
</s>
<s>
Modeled	O
on	O
the	O
successful	O
collaborative	O
approach	O
and	O
the	O
compression	O
technologies	O
developed	O
by	O
the	O
Joint	O
Photographic	O
Experts	O
Group	O
and	O
CCITT	O
's	O
Experts	O
Group	O
on	O
Telephony	O
(	O
creators	O
of	O
the	O
JPEG	O
image	O
compression	O
standard	O
and	O
the	O
H.261	B-Algorithm
standard	I-Algorithm
for	O
video	B-Application
conferencing	I-Application
respectively	O
)	O
,	O
the	O
Moving	O
Picture	O
Experts	O
Group	O
(	O
MPEG	O
)	O
working	O
group	O
was	O
established	O
in	O
January	O
1988	O
,	O
by	O
the	O
initiative	O
of	O
Hiroshi	O
Yasuda	O
(	O
Nippon	O
Telegraph	O
and	O
Telephone	O
)	O
and	O
Leonardo	O
Chiariglione	O
(	O
CSELT	O
)	O
.	O
</s>
<s>
MPEG	O
was	O
formed	O
to	O
address	O
the	O
need	O
for	O
standard	O
video	O
and	O
audio	O
formats	O
,	O
and	O
to	O
build	O
on	O
H.261	B-Algorithm
to	O
get	O
better	O
quality	O
through	O
the	O
use	O
of	O
somewhat	O
more	O
complex	O
encoding	O
methods	O
(	O
e.g.	O
,	O
supporting	O
higher	O
precision	O
for	O
motion	O
vectors	O
)	O
.	O
</s>
<s>
Development	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
began	O
in	O
May	O
1988	O
.	O
</s>
<s>
Fourteen	O
video	O
and	O
fourteen	O
audio	O
codec	B-General_Concept
proposals	O
were	O
submitted	O
by	O
individual	O
companies	O
and	O
institutions	O
for	O
evaluation	O
.	O
</s>
<s>
The	O
codecs	B-General_Concept
were	O
extensively	O
tested	O
for	O
computational	O
complexity	O
and	O
subjective	O
(	O
human	O
perceived	O
)	O
quality	O
,	O
at	O
data	O
rates	O
of	O
1.5Mbit/s	O
.	O
</s>
<s>
This	O
specific	O
bitrate	O
was	O
chosen	O
for	O
transmission	O
over	O
T-1/E	O
-1	O
lines	O
and	O
as	O
the	O
approximate	O
data	O
rate	O
of	O
audio	O
CDs	O
.	O
</s>
<s>
The	O
codecs	B-General_Concept
that	O
excelled	O
in	O
this	O
testing	O
were	O
utilized	O
as	O
the	O
basis	O
for	O
the	O
standard	O
and	O
refined	O
further	O
,	O
with	O
additional	O
features	O
and	O
other	O
improvements	O
being	O
incorporated	O
in	O
the	O
process	O
.	O
</s>
<s>
The	O
reported	O
completion	O
date	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
varies	O
greatly	O
:	O
a	O
largely	O
complete	O
draft	O
standard	O
was	O
produced	O
in	O
September	O
1990	O
,	O
and	O
from	O
that	O
point	O
on	O
,	O
only	O
minor	O
changes	O
were	O
introduced	O
.	O
</s>
<s>
The	O
Berkeley	O
Plateau	O
Multimedia	O
Research	O
Group	O
developed	O
an	O
MPEG-1	B-Algorithm
decoder	B-General_Concept
in	O
November	O
1992	O
.	O
</s>
<s>
In	O
July	O
1990	O
,	O
before	O
the	O
first	O
draft	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
had	O
even	O
been	O
written	O
,	O
work	O
began	O
on	O
a	O
second	O
standard	O
,	O
MPEG-2	B-Algorithm
,	O
intended	O
to	O
extend	O
MPEG-1	B-Algorithm
technology	O
to	O
provide	O
full	O
broadcast-quality	O
video	O
(	O
as	O
per	O
CCIR	O
601	O
)	O
at	O
high	O
bitrates	O
(	O
3	O
–	O
15	O
Mbit/s	O
)	O
and	O
support	O
for	O
interlaced	O
video	O
.	O
</s>
<s>
Due	O
in	O
part	O
to	O
the	O
similarity	O
between	O
the	O
two	O
codecs	B-General_Concept
,	O
the	O
MPEG-2	B-Algorithm
standard	O
includes	O
full	O
backwards	O
compatibility	O
with	O
MPEG-1	B-Algorithm
video	O
,	O
so	O
any	O
MPEG-2	B-Algorithm
decoder	B-General_Concept
can	O
play	O
MPEG-1	B-Algorithm
videos	O
.	O
</s>
<s>
Notably	O
,	O
the	O
MPEG-1	B-Algorithm
standard	O
very	O
strictly	O
defines	O
the	O
bitstream	O
,	O
and	O
decoder	B-General_Concept
function	O
,	O
but	O
does	O
not	O
define	O
how	O
MPEG-1	B-Algorithm
encoding	O
is	O
to	O
be	O
performed	O
,	O
although	O
a	O
reference	O
implementation	O
is	O
provided	O
in	O
ISO/IEC	O
-11172-5	O
.	O
</s>
<s>
This	O
means	O
that	O
MPEG-1	B-Algorithm
coding	B-General_Concept
efficiency	I-General_Concept
can	O
drastically	O
vary	O
depending	O
on	O
the	O
encoder	O
used	O
,	O
and	O
generally	O
means	O
that	O
newer	O
encoders	O
perform	O
significantly	O
better	O
than	O
their	O
predecessors	O
.	O
</s>
<s>
The	O
first	O
three	O
parts	O
(	O
Systems	O
,	O
Video	O
and	O
Audio	O
)	O
of	O
ISO/IEC	B-Algorithm
11172	I-Algorithm
were	O
published	O
in	O
August	O
1993	O
.	O
</s>
<s>
Due	O
to	O
its	O
age	O
,	O
MPEG-1	B-Algorithm
is	O
no	O
longer	O
covered	O
by	O
any	O
essential	O
patents	O
and	O
can	O
thus	O
be	O
used	O
without	O
obtaining	O
a	O
licence	O
or	O
paying	O
any	O
fees	O
.	O
</s>
<s>
The	O
near-complete	O
draft	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
was	O
publicly	O
available	O
as	O
ISO	O
CD	O
11172	O
by	O
December	O
6	O
,	O
1991	O
.	O
</s>
<s>
Neither	O
the	O
July	O
2008	O
Kuro5hin	O
article	O
"	O
Patent	O
Status	O
of	O
MPEG-1	B-Algorithm
,	O
H.261	B-Algorithm
and	O
MPEG-2	B-Algorithm
"	O
,	O
nor	O
an	O
August	O
2008	O
thread	O
on	O
the	O
gstreamer-devel	O
mailing	O
list	O
were	O
able	O
to	O
list	O
a	O
single	O
unexpired	O
MPEG-1	B-Algorithm
Video	O
and	O
MPEG-1	B-Algorithm
Audio	O
Layer	O
I/II	O
patent	O
.	O
</s>
<s>
A	O
May	O
2009	O
discussion	O
on	O
the	O
whatwg	O
mailing	O
list	O
mentioned	O
US	O
5,214,678	O
patent	O
as	O
possibly	O
covering	O
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
.	O
</s>
<s>
A	O
full	O
MPEG-1	B-Algorithm
decoder	B-General_Concept
and	O
encoder	O
,	O
with	O
"	O
Layer	O
III	O
audio	O
"	O
,	O
could	O
not	O
be	O
implemented	O
royalty	O
free	O
since	O
there	O
were	O
companies	O
that	O
required	O
patent	O
fees	O
for	O
implementations	O
of	O
MPEG-1	B-Application
Audio	I-Application
Layer	I-Application
III	I-Application
,	O
as	O
discussed	O
in	O
the	O
MP3	B-Application
article	O
.	O
</s>
<s>
All	O
patents	O
in	O
the	O
world	O
connected	O
to	O
MP3	B-Application
expired	O
30	O
December	O
2017	O
,	O
which	O
makes	O
this	O
format	O
totally	O
free	O
for	O
use	O
.	O
</s>
<s>
On	O
23	O
April	O
2017	O
,	O
Fraunhofer	O
IIS	O
stopped	O
charging	O
for	O
Technicolor	O
's	O
MP3	B-Application
licensing	O
program	O
for	O
certain	O
MP3	B-Application
related	O
patents	O
and	O
software	O
.	O
</s>
<s>
The	O
following	O
corporations	O
filed	O
declarations	O
with	O
ISO	O
saying	O
they	O
held	O
patents	O
for	O
the	O
MPEG-1	B-Algorithm
Video	O
(	O
ISO/IEC	O
-11172-2	O
)	O
format	O
,	O
although	O
all	O
such	O
patents	O
have	O
since	O
expired	O
.	O
</s>
<s>
Most	O
popular	O
software	O
for	O
video	O
playback	O
includes	O
MPEG-1	B-Algorithm
decoding	O
,	O
in	O
addition	O
to	O
any	O
other	O
supported	O
formats	O
.	O
</s>
<s>
The	O
popularity	O
of	O
MP3	B-Application
audio	I-Application
has	O
established	O
a	O
massive	O
installed	B-Operating_System
base	I-Operating_System
of	O
hardware	O
that	O
can	O
play	O
back	O
MPEG-1	B-Algorithm
Audio	O
(	O
all	O
three	O
layers	O
)	O
.	O
</s>
<s>
"	O
Virtually	O
all	O
digital	O
audio	O
devices	O
"	O
can	O
play	O
back	O
MPEG-1	B-Algorithm
Audio	O
.	O
</s>
<s>
Before	O
MPEG-2	B-Algorithm
became	O
widespread	O
,	O
many	O
digital	O
satellite/cable	O
TV	O
services	O
used	O
MPEG-1	B-Algorithm
exclusively	O
.	O
</s>
<s>
The	O
widespread	O
popularity	O
of	O
MPEG-2	B-Algorithm
with	O
broadcasters	O
means	O
MPEG-1	B-Algorithm
is	O
playable	O
by	O
most	O
digital	O
cable	O
and	O
satellite	O
set-top	O
boxes	O
,	O
and	O
digital	O
disc	O
and	O
tape	O
players	O
,	O
due	O
to	O
backwards	O
compatibility	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
was	O
used	O
for	O
full-screen	O
video	O
on	O
Green	B-Device
Book	I-Device
CD-i	B-Device
,	O
and	O
on	O
Video	O
CD	O
(	O
VCD	O
)	O
.	O
</s>
<s>
The	O
Super	O
Video	O
CD	O
standard	O
,	O
based	O
on	O
VCD	O
,	O
uses	O
MPEG-1	B-Algorithm
audio	O
exclusively	O
,	O
as	O
well	O
as	O
MPEG-2	B-Algorithm
video	O
.	O
</s>
<s>
The	O
DVD-Video	O
format	O
uses	O
MPEG-2	B-Algorithm
video	O
primarily	O
,	O
but	O
MPEG-1	B-Algorithm
support	O
is	O
explicitly	O
defined	O
in	O
the	O
standard	O
.	O
</s>
<s>
The	O
DVD-Video	O
standard	O
originally	O
required	O
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
for	O
PAL	O
countries	O
,	O
but	O
was	O
changed	O
to	O
allow	O
AC-3/Dolby	O
Digital-only	O
discs	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
is	O
still	O
allowed	O
on	O
DVDs	O
,	O
although	O
newer	O
extensions	O
to	O
the	O
format	O
,	O
like	O
MPEG	B-Algorithm
Multichannel	I-Algorithm
,	O
are	O
rarely	O
supported	O
.	O
</s>
<s>
Most	O
DVD	O
players	O
also	O
support	O
Video	O
CD	O
and	O
MP3	B-Application
CD	O
playback	O
,	O
which	O
use	O
MPEG-1	B-Algorithm
.	O
</s>
<s>
The	O
international	O
Digital	B-Application
Video	I-Application
Broadcasting	I-Application
(	O
DVB	B-Application
)	O
standard	O
primarily	O
uses	O
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
,	O
and	O
MPEG-2	B-Algorithm
video	O
.	O
</s>
<s>
The	O
international	O
Digital	O
Audio	O
Broadcasting	O
(	O
DAB	O
)	O
standard	O
uses	O
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
exclusively	O
,	O
due	O
to	O
its	O
especially	O
high	O
quality	O
,	O
modest	O
decoder	B-General_Concept
performance	O
requirements	O
,	O
and	O
tolerance	O
of	O
errors	O
.	O
</s>
<s>
PASC	O
is	O
an	O
early	O
version	O
of	O
MPEG-1	B-Algorithm
Audio	O
Layer	O
I	O
with	O
a	O
fixed	O
bit	O
rate	O
of	O
384kilobits	O
per	O
second	O
.	O
</s>
<s>
Part	O
1	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
covers	O
systems	O
,	O
and	O
is	O
defined	O
in	O
ISO/IEC	O
-11172-1	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
Systems	O
specifies	O
the	O
logical	O
layout	O
and	O
methods	O
used	O
to	O
store	O
the	O
encoded	O
audio	O
,	O
video	O
,	O
and	O
other	O
data	O
into	O
a	O
standard	O
bitstream	O
,	O
and	O
to	O
maintain	O
synchronization	O
between	O
the	O
different	O
contents	O
.	O
</s>
<s>
This	O
structure	O
was	O
later	O
named	O
an	O
MPEG	O
program	O
stream	O
:	O
"	O
The	O
MPEG-1	B-Algorithm
Systems	O
design	O
is	O
essentially	O
identical	O
to	O
the	O
MPEG-2	B-Algorithm
Program	O
Stream	O
structure.	O
"	O
</s>
<s>
Elementary	O
Streams	O
(	O
ES	O
)	O
are	O
the	O
raw	O
bitstreams	O
of	O
MPEG-1	B-Algorithm
audio	O
and	O
video	O
encoded	O
data	O
(	O
output	O
from	O
an	O
encoder	O
)	O
.	O
</s>
<s>
These	O
files	O
can	O
be	O
distributed	O
on	O
their	O
own	O
,	O
such	O
as	O
is	O
the	O
case	O
with	O
MP3	B-Application
files	I-Application
.	O
</s>
<s>
Packetized	O
Elementary	O
Streams	O
(	O
PES	O
)	O
are	O
elementary	O
streams	O
packetized	O
into	O
packets	B-Protocol
of	O
variable	O
lengths	O
,	O
i.e.	O
,	O
divided	O
ES	O
into	O
independent	O
chunks	O
where	O
cyclic	O
redundancy	O
check	O
(	O
CRC	O
)	O
checksum	B-Algorithm
was	O
added	O
to	O
each	O
packet	B-Protocol
for	O
error	O
detection	O
.	O
</s>
<s>
The	O
PS	O
structure	O
is	O
known	O
as	O
a	O
multiplex	B-Architecture
,	O
or	O
a	O
container	O
format	O
.	O
</s>
<s>
90kHz	O
PTS	O
values	O
in	O
the	O
PS	O
header	O
tell	O
the	O
decoder	B-General_Concept
which	O
video	O
SCR	O
values	O
match	O
which	O
audio	O
SCR	O
values	O
.	O
</s>
<s>
PTS	O
determines	O
when	O
to	O
display	O
a	O
portion	O
of	O
an	O
MPEG	O
program	O
,	O
and	O
is	O
also	O
used	O
by	O
the	O
decoder	B-General_Concept
to	O
determine	O
when	O
data	O
can	O
be	O
discarded	O
from	O
the	O
buffer	B-General_Concept
.	O
</s>
<s>
Either	O
video	O
or	O
audio	O
will	O
be	O
delayed	O
by	O
the	O
decoder	B-General_Concept
until	O
the	O
corresponding	O
segment	O
of	O
the	O
other	O
arrives	O
and	O
can	O
be	O
decoded	O
.	O
</s>
<s>
Such	O
PTS	O
wraparound	O
disparities	O
can	O
cause	O
timing	O
issues	O
that	O
must	O
be	O
specially	O
handled	O
by	O
the	O
decoder	B-General_Concept
.	O
</s>
<s>
DTS	O
is	O
quite	O
similar	O
to	O
PTS	O
,	O
but	O
instead	O
of	O
just	O
handling	O
sequential	O
frames	O
,	O
it	O
contains	O
the	O
proper	O
time-stamps	O
to	O
tell	O
the	O
decoder	B-General_Concept
when	O
to	O
decode	O
and	O
display	O
the	O
next	O
B-frame	O
(	O
types	O
of	O
frames	O
explained	O
below	O
)	O
,	O
ahead	O
of	O
its	O
anchor	O
(	O
P	O
-	O
or	O
I	O
-	O
)	O
frame	O
.	O
</s>
<s>
This	O
is	O
done	O
so	O
the	O
packets	B-Protocol
of	O
the	O
simultaneous	O
streams	O
can	O
be	O
transferred	O
over	O
the	O
same	O
channel	O
and	O
are	O
guaranteed	O
to	O
both	O
arrive	O
at	O
the	O
decoder	B-General_Concept
at	O
precisely	O
the	O
same	O
time	O
.	O
</s>
<s>
This	O
is	O
a	O
case	O
of	O
time-division	B-Protocol
multiplexing	I-Protocol
.	O
</s>
<s>
Improper	O
interleaving	O
will	O
result	O
in	O
buffer	B-General_Concept
underflows	O
or	O
overflows	O
,	O
as	O
the	O
receiver	O
gets	O
more	O
of	O
one	O
stream	O
than	O
it	O
can	O
store	O
(	O
e.g.	O
</s>
<s>
The	O
MPEG	O
Video	O
Buffering	O
Verifier	O
(	O
VBV	O
)	O
assists	O
in	O
determining	O
if	O
a	O
multiplexed	B-Architecture
PS	O
can	O
be	O
decoded	O
by	O
a	O
device	O
with	O
a	O
specified	O
data	O
throughput	O
rate	O
and	O
buffer	B-General_Concept
size	O
.	O
</s>
<s>
This	O
offers	O
feedback	O
to	O
the	O
multiplexer	O
and	O
the	O
encoder	O
,	O
so	O
that	O
they	O
can	O
change	O
the	O
multiplex	B-Architecture
size	O
or	O
adjust	O
bitrates	O
as	O
needed	O
for	O
compliance	O
.	O
</s>
<s>
Part	O
2	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
covers	O
video	O
and	O
is	O
defined	O
in	O
ISO/IEC	O
-11172-2	O
.	O
</s>
<s>
The	O
design	O
was	O
heavily	O
influenced	O
by	O
H.261	B-Algorithm
.	O
</s>
<s>
MPEG-1	B-Algorithm
Video	O
exploits	O
perceptual	O
compression	O
methods	O
to	O
significantly	O
reduce	O
the	O
data	O
rate	O
required	O
by	O
a	O
video	O
stream	O
.	O
</s>
<s>
It	O
also	O
exploits	O
temporal	O
(	O
over	O
time	O
)	O
and	O
spatial	O
(	O
across	O
a	O
picture	O
)	O
redundancy	O
common	O
in	O
video	O
to	O
achieve	O
better	O
data	B-General_Concept
compression	I-General_Concept
than	O
would	O
be	O
possible	O
otherwise	O
.	O
</s>
<s>
Before	O
encoding	O
video	O
to	O
MPEG-1	B-Algorithm
,	O
the	O
color-space	O
is	O
transformed	O
to	O
Y′CbCr	O
(	O
Y′=Luma	O
,	O
Cb	O
=	O
Chroma	O
Blue	O
,	O
Cr	O
=	O
Chroma	O
Red	O
)	O
.	O
</s>
<s>
This	O
use	O
of	O
higher	O
resolution	O
for	O
some	O
color	O
components	O
is	O
similar	O
in	O
concept	O
to	O
the	O
Bayer	B-Algorithm
pattern	I-Algorithm
filter	I-Algorithm
that	O
is	O
commonly	O
used	O
for	O
the	O
image	O
capturing	O
sensor	O
in	O
digital	O
color	O
cameras	O
.	O
</s>
<s>
However	O
,	O
on	O
videos	O
with	O
fine	O
detail	O
(	O
high	O
spatial	O
complexity	O
)	O
this	O
can	O
manifest	O
as	O
chroma	O
aliasing	B-Error_Name
artifacts	B-Algorithm
.	O
</s>
<s>
Compared	O
to	O
other	O
digital	O
compression	B-Algorithm
artifacts	I-Algorithm
,	O
this	O
issue	O
seems	O
to	O
very	O
rarely	O
be	O
a	O
source	O
of	O
annoyance	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
supports	O
resolutions	O
up	O
to	O
4095×4095	O
(	O
12	O
bits	O
)	O
,	O
and	O
bit	O
rates	O
up	O
to	O
100Mbit/s	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
videos	O
are	O
most	O
commonly	O
seen	O
using	O
Source	O
Input	O
Format	O
(	O
SIF	O
)	O
resolution	O
:	O
352×240	O
,	O
352×288	O
,	O
or	O
320×240	O
.	O
</s>
<s>
These	O
relatively	O
low	O
resolutions	O
,	O
combined	O
with	O
a	O
bitrate	O
less	O
than	O
1.5	O
Mbit/s	O
,	O
make	O
up	O
what	O
is	O
known	O
as	O
a	O
constrained	O
parameters	O
bitstream	O
(	O
CPB	O
)	O
,	O
later	O
renamed	O
the	O
"	O
Low	O
Level	O
"	O
(	O
LL	O
)	O
profile	O
in	O
MPEG-2	B-Algorithm
.	O
</s>
<s>
This	O
is	O
the	O
minimum	O
video	O
specifications	O
any	O
decoder	B-General_Concept
should	O
be	O
able	O
to	O
handle	O
,	O
to	O
be	O
considered	O
MPEG-1	B-Algorithm
compliant	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
has	O
several	O
frame/picture	O
types	O
that	O
serve	O
different	O
purposes	O
.	O
</s>
<s>
High-speed	O
seeking	O
through	O
an	O
MPEG-1	B-Algorithm
video	O
is	O
only	O
possible	O
to	O
the	O
nearest	O
I-frame	O
.	O
</s>
<s>
I-frame	O
only	O
compression	O
is	O
very	O
fast	O
,	O
but	O
produces	O
very	O
large	O
file	O
sizes	O
:	O
a	O
factor	O
of	O
3	O
(	O
or	O
more	O
)	O
larger	O
than	O
normally	O
encoded	O
MPEG-1	B-Algorithm
video	O
,	O
depending	O
on	O
how	O
temporally	O
complex	O
a	O
specific	O
video	O
is	O
.	O
</s>
<s>
I-frame	O
only	O
MPEG-1	B-Algorithm
video	O
is	O
very	O
similar	O
to	O
MJPEG	B-Application
video	O
.	O
</s>
<s>
So	O
much	O
so	O
that	O
very	O
high-speed	O
and	O
theoretically	O
lossless	O
(	O
in	O
reality	O
,	O
there	O
are	O
rounding	O
errors	O
)	O
conversion	O
can	O
be	O
made	O
from	O
one	O
format	O
to	O
the	O
other	O
,	O
provided	O
a	O
couple	O
of	O
restrictions	O
(	O
color	O
space	O
and	O
quantization	B-Algorithm
matrix	O
)	O
are	O
followed	O
in	O
the	O
creation	O
of	O
the	O
bitstream	O
.	O
</s>
<s>
The	O
length	O
between	O
I-frames	O
is	O
known	O
as	O
the	O
group	B-General_Concept
of	I-General_Concept
pictures	I-General_Concept
(	O
GOP	O
)	O
size	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
most	O
commonly	O
uses	O
a	O
GOP	O
size	O
of	O
15	O
–	O
18	O
.	O
</s>
<s>
Limits	O
are	O
placed	O
on	O
the	O
maximum	O
number	O
of	O
frames	O
between	O
I-frames	O
due	O
to	O
decoding	O
complexing	O
,	O
decoder	B-General_Concept
buffer	B-General_Concept
size	O
,	O
recovery	O
time	O
after	O
data	O
errors	O
,	O
seeking	O
ability	O
,	O
and	O
accumulation	O
of	O
IDCT	B-General_Concept
errors	O
in	O
low-precision	O
implementations	O
most	O
common	O
in	O
hardware	O
decoders	O
(	O
See	O
:	O
IEEE-1180	O
)	O
.	O
</s>
<s>
The	O
difference	O
between	O
a	O
P-frame	O
and	O
its	O
anchor	O
frame	O
is	O
calculated	O
using	O
motion	O
vectors	O
on	O
each	O
macroblock	B-General_Concept
of	O
the	O
frame	O
(	O
see	O
below	O
)	O
.	O
</s>
<s>
Such	O
motion	O
vector	O
data	O
will	O
be	O
embedded	O
in	O
the	O
P-frame	O
for	O
use	O
by	O
the	O
decoder	B-General_Concept
.	O
</s>
<s>
This	O
means	O
decoding	O
B-frames	O
requires	O
larger	O
data	B-General_Concept
buffers	I-General_Concept
and	O
causes	O
an	O
increased	O
delay	O
on	O
both	O
decoding	O
and	O
during	O
encoding	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
has	O
a	O
unique	O
frame	O
type	O
not	O
found	O
in	O
later	O
video	O
standards	O
.	O
</s>
<s>
"	O
D-frames	O
"	O
or	O
DC-pictures	O
are	O
independently	O
coded	O
images	O
(	O
intra-frames	O
)	O
that	O
have	O
been	O
encoded	O
using	O
DC	O
transform	O
coefficients	O
only	O
(	O
AC	O
coefficients	O
are	O
removed	O
when	O
encoding	O
D-framessee	O
DCT	B-General_Concept
below	O
)	O
and	O
hence	O
are	O
very	O
low	O
quality	O
.	O
</s>
<s>
For	O
this	O
reason	O
,	O
D-frames	O
are	O
seldom	O
actually	O
used	O
in	O
MPEG-1	B-Algorithm
video	O
encoding	O
,	O
and	O
the	O
D-frame	O
feature	O
has	O
not	O
been	O
included	O
in	O
any	O
later	O
video	O
coding	O
standards	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
operates	O
on	O
video	O
in	O
a	O
series	O
of	O
8×8	O
blocks	O
for	O
quantization	B-Algorithm
.	O
</s>
<s>
This	O
set	O
of	O
6	O
blocks	O
,	O
with	O
a	O
resolution	O
of	O
16×16	O
,	O
is	O
processed	O
together	O
and	O
called	O
a	O
macroblock	B-General_Concept
.	O
</s>
<s>
A	O
macroblock	B-General_Concept
is	O
the	O
smallest	O
independent	O
unit	O
of	O
(	O
color	O
)	O
video	O
.	O
</s>
<s>
Motion	O
vectors	O
(	O
see	O
below	O
)	O
operate	O
solely	O
at	O
the	O
macroblock	B-General_Concept
level	O
.	O
</s>
<s>
If	O
the	O
height	O
or	O
width	O
of	O
the	O
video	O
are	O
not	O
exact	O
multiples	O
of	O
16	O
,	O
full	O
rows	O
and	O
full	O
columns	O
of	O
macroblocks	B-General_Concept
must	O
still	O
be	O
encoded	O
and	O
decoded	O
to	O
fill	O
out	O
the	O
picture	O
(	O
though	O
the	O
extra	O
decoded	O
pixels	O
are	O
not	O
displayed	O
)	O
.	O
</s>
<s>
The	O
encoder	O
compares	O
the	O
current	O
frame	O
with	O
adjacent	O
parts	O
of	O
the	O
video	O
from	O
the	O
anchor	O
frame	O
(	O
previous	O
I	O
-	O
or	O
P	O
-	O
frame	O
)	O
in	O
a	O
diamond	O
pattern	O
,	O
up	O
to	O
a	O
(	O
encoder-specific	O
)	O
predefined	O
radius	B-Protocol
limit	O
from	O
the	O
area	O
of	O
the	O
current	O
macroblock	B-General_Concept
.	O
</s>
<s>
the	O
vector	O
of	O
the	O
motion	O
)	O
from	O
the	O
previous	O
video	O
area	O
to	O
the	O
current	O
macroblock	B-General_Concept
need	O
to	O
be	O
encoded	O
into	O
the	O
inter-frame	O
(	O
P	O
-	O
or	O
B	O
-	O
frame	O
)	O
.	O
</s>
<s>
The	O
reverse	O
of	O
this	O
process	O
,	O
performed	O
by	O
the	O
decoder	B-General_Concept
to	O
reconstruct	O
the	O
picture	O
,	O
is	O
called	O
motion	B-Algorithm
compensation	I-Algorithm
.	O
</s>
<s>
A	O
predicted	O
macroblock	B-General_Concept
rarely	O
matches	O
the	O
current	O
picture	O
perfectly	O
,	O
however	O
.	O
</s>
<s>
The	O
differences	O
between	O
the	O
estimated	O
matching	O
area	O
,	O
and	O
the	O
real	O
frame/macroblock	O
is	O
called	O
the	O
prediction	O
error	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
video	O
uses	O
a	O
motion	O
vector	O
(	O
MV	O
)	O
precision	O
of	O
one	O
half	O
of	O
one	O
pixel	O
,	O
or	O
half-pel	O
.	O
</s>
<s>
Finer	O
MV	O
precision	O
results	O
in	O
using	O
a	O
larger	O
amount	O
of	O
data	O
to	O
represent	O
the	O
MV	O
,	O
as	O
larger	O
numbers	O
must	O
be	O
stored	O
in	O
the	O
frame	O
for	O
every	O
single	O
MV	O
,	O
increased	O
coding	O
complexity	O
as	O
increasing	O
levels	O
of	O
interpolation	O
on	O
the	O
macroblock	B-General_Concept
are	O
required	O
for	O
both	O
the	O
encoder	O
and	O
decoder	B-General_Concept
,	O
and	O
diminishing	O
returns	O
(	O
minimal	O
gains	O
)	O
with	O
higher	O
precision	O
MVs	O
.	O
</s>
<s>
Because	O
neighboring	O
macroblocks	B-General_Concept
are	O
likely	O
to	O
have	O
very	O
similar	O
motion	O
vectors	O
,	O
this	O
redundant	O
information	O
can	O
be	O
compressed	O
quite	O
effectively	O
by	O
being	O
stored	O
DPCM-encoded	O
.	O
</s>
<s>
Only	O
the	O
(	O
smaller	O
)	O
amount	O
of	O
difference	O
between	O
the	O
MVs	O
for	O
each	O
macroblock	B-General_Concept
needs	O
to	O
be	O
stored	O
in	O
the	O
final	O
bitstream	O
.	O
</s>
<s>
P-frames	O
have	O
one	O
motion	O
vector	O
per	O
macroblock	B-General_Concept
,	O
relative	O
to	O
the	O
previous	O
anchor	O
frame	O
.	O
</s>
<s>
Partial	O
macroblocks	B-General_Concept
,	O
and	O
black	O
borders/bars	O
encoded	O
into	O
the	O
video	O
that	O
do	O
not	O
fall	O
exactly	O
on	O
a	O
macroblock	B-General_Concept
boundary	O
,	O
cause	O
havoc	O
with	O
motion	O
prediction	O
.	O
</s>
<s>
The	O
block	O
padding/border	O
information	O
prevents	O
the	O
macroblock	B-General_Concept
from	O
closely	O
matching	O
with	O
any	O
other	O
area	O
of	O
the	O
video	O
,	O
and	O
so	O
,	O
significantly	O
larger	O
prediction	O
error	O
information	O
must	O
be	O
encoded	O
for	O
every	O
one	O
of	O
the	O
several	O
dozen	O
partial	O
macroblocks	B-General_Concept
along	O
the	O
screen	O
border	O
.	O
</s>
<s>
DCT	B-General_Concept
encoding	O
and	O
quantization	B-Algorithm
(	O
see	O
below	O
)	O
also	O
is	O
n't	O
nearly	O
as	O
effective	O
when	O
there	O
is	O
large/sharp	O
picture	O
contrast	O
in	O
a	O
block	O
.	O
</s>
<s>
An	O
even	O
more	O
serious	O
problem	O
exists	O
with	O
macroblocks	B-General_Concept
that	O
contain	O
significant	O
,	O
random	O
,	O
edge	O
noise	O
,	O
where	O
the	O
picture	O
transitions	O
to	O
(	O
typically	O
)	O
black	O
.	O
</s>
<s>
Each	O
8×8	O
block	O
is	O
encoded	O
by	O
first	O
applying	O
a	O
forward	O
discrete	B-General_Concept
cosine	I-General_Concept
transform	I-General_Concept
(	O
FDCT	O
)	O
and	O
then	O
a	O
quantization	B-Algorithm
process	O
.	O
</s>
<s>
The	O
FDCT	O
process	O
(	O
by	O
itself	O
)	O
is	O
theoretically	O
lossless	O
,	O
and	O
can	O
be	O
reversed	O
by	O
applying	O
an	O
Inverse	O
DCT	B-General_Concept
(	O
IDCT	B-General_Concept
)	O
to	O
reproduce	O
the	O
original	O
values	O
(	O
in	O
the	O
absence	O
of	O
any	O
quantization	B-Algorithm
and	O
rounding	O
errors	O
)	O
.	O
</s>
<s>
In	O
reality	O
,	O
there	O
are	O
some	O
(	O
sometimes	O
large	O
)	O
rounding	O
errors	O
introduced	O
both	O
by	O
quantization	B-Algorithm
in	O
the	O
encoder	O
(	O
as	O
described	O
in	O
the	O
next	O
section	O
)	O
and	O
by	O
IDCT	B-General_Concept
approximation	O
error	O
in	O
the	O
decoder	B-General_Concept
.	O
</s>
<s>
The	O
minimum	O
allowed	O
accuracy	O
of	O
a	O
decoder	B-General_Concept
IDCT	B-General_Concept
approximation	O
is	O
defined	O
by	O
ISO/IEC	O
23002-1	O
.	O
</s>
<s>
The	O
FDCT	O
process	O
converts	O
the	O
8×8	O
block	O
of	O
uncompressed	B-General_Concept
pixel	O
values	O
(	O
brightness	O
or	O
color	O
difference	O
values	O
)	O
into	O
an	O
8×8	O
indexed	O
array	O
of	O
frequency	O
coefficient	O
values	O
.	O
</s>
<s>
Since	O
the	O
DC	O
coefficient	O
value	O
is	O
statistically	O
correlated	O
from	O
one	O
block	O
to	O
the	O
next	O
,	O
it	O
is	O
compressed	O
using	O
DPCM	B-Algorithm
encoding	O
.	O
</s>
<s>
Additionally	O
,	O
the	O
frequency	O
conversion	O
performed	O
by	O
applying	O
the	O
DCT	B-General_Concept
provides	O
a	O
statistical	O
decorrelation	O
function	O
to	O
efficiently	O
concentrate	O
the	O
signal	O
into	O
fewer	O
high-amplitude	O
values	O
prior	O
to	O
applying	O
quantization	B-Algorithm
(	O
see	O
below	O
)	O
.	O
</s>
<s>
Quantization	B-Algorithm
is	O
,	O
essentially	O
,	O
the	O
process	O
of	O
reducing	O
the	O
accuracy	O
of	O
a	O
signal	O
,	O
by	O
dividing	O
it	O
by	O
some	O
larger	O
step	O
size	O
and	O
rounding	O
to	O
an	O
integer	O
value	O
(	O
i.e.	O
</s>
<s>
The	O
frame-level	O
quantizer	B-Algorithm
is	O
a	O
number	O
from	O
0	O
to	O
31	O
(	O
although	O
encoders	O
will	O
usually	O
omit/disable	O
some	O
of	O
the	O
extreme	O
values	O
)	O
which	O
determines	O
how	O
much	O
information	O
will	O
be	O
removed	O
from	O
a	O
given	O
frame	O
.	O
</s>
<s>
The	O
frame-level	O
quantizer	B-Algorithm
is	O
typically	O
either	O
dynamically	O
selected	O
by	O
the	O
encoder	O
to	O
maintain	O
a	O
certain	O
user-specified	O
bitrate	O
,	O
or	O
(	O
much	O
less	O
commonly	O
)	O
directly	O
specified	O
by	O
the	O
user	O
.	O
</s>
<s>
A	O
"	O
quantization	B-Algorithm
matrix	O
"	O
is	O
a	O
string	O
of	O
64	O
numbers	O
(	O
ranging	O
from	O
0	O
to	O
255	O
)	O
which	O
tells	O
the	O
encoder	O
how	O
relatively	O
important	O
or	O
unimportant	O
each	O
piece	O
of	O
visual	O
information	O
is	O
.	O
</s>
<s>
An	O
example	O
quantization	B-Algorithm
matrix	O
:	O
</s>
<s>
Quantization	B-Algorithm
is	O
performed	O
by	O
taking	O
each	O
of	O
the	O
64	O
frequency	O
values	O
of	O
the	O
DCT	B-General_Concept
block	O
,	O
dividing	O
them	O
by	O
the	O
frame-level	O
quantizer	B-Algorithm
,	O
then	O
dividing	O
them	O
by	O
their	O
corresponding	O
values	O
in	O
the	O
quantization	B-Algorithm
matrix	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
actually	O
uses	O
two	O
separate	O
quantization	B-Algorithm
matrices	O
,	O
one	O
for	O
intra-blocks	O
(	O
I-blocks	O
)	O
and	O
one	O
for	O
inter-block	O
(	O
P	O
-	O
and	O
B	O
-	O
blocks	O
)	O
so	O
quantization	B-Algorithm
of	O
different	O
block	O
types	O
can	O
be	O
done	O
independently	O
,	O
and	O
so	O
,	O
more	O
effectively	O
.	O
</s>
<s>
This	O
quantization	B-Algorithm
process	O
usually	O
reduces	O
a	O
significant	O
number	O
of	O
the	O
AC	O
coefficients	O
to	O
zero	O
,	O
(	O
known	O
as	O
sparse	O
data	O
)	O
which	O
can	O
then	O
be	O
more	O
efficiently	O
compressed	O
by	O
entropy	B-Algorithm
coding	I-Algorithm
(	O
lossless	O
compression	O
)	O
in	O
the	O
next	O
step	O
.	O
</s>
<s>
An	O
example	O
quantized	O
DCT	B-General_Concept
block	O
:	O
</s>
<s>
Quantization	B-Algorithm
eliminates	O
a	O
large	O
amount	O
of	O
data	O
,	O
and	O
is	O
the	O
main	O
lossy	B-Algorithm
processing	O
step	O
in	O
MPEG-1	B-Algorithm
video	O
encoding	O
.	O
</s>
<s>
This	O
is	O
also	O
the	O
primary	O
source	O
of	O
most	O
MPEG-1	B-Algorithm
video	B-Algorithm
compression	I-Algorithm
artifacts	I-Algorithm
,	O
like	O
blockiness	O
,	O
color	B-Algorithm
banding	I-Algorithm
,	O
noise	O
,	O
ringing	B-Algorithm
,	O
discoloration	O
,	O
et	O
al	O
.	O
</s>
<s>
This	O
happens	O
when	O
video	O
is	O
encoded	O
with	O
an	O
insufficient	O
bitrate	O
,	O
and	O
the	O
encoder	O
is	O
therefore	O
forced	O
to	O
use	O
high	O
frame-level	O
quantizers	B-Algorithm
(	O
strong	O
quantization	B-Algorithm
)	O
through	O
much	O
of	O
the	O
video	O
.	O
</s>
<s>
Several	O
steps	O
in	O
the	O
encoding	O
of	O
MPEG-1	B-Algorithm
video	O
are	O
lossless	O
,	O
meaning	O
they	O
will	O
be	O
reversed	O
upon	O
decoding	O
,	O
to	O
produce	O
exactly	O
the	O
same	O
(	O
original	O
)	O
values	O
.	O
</s>
<s>
Since	O
these	O
lossless	O
data	B-General_Concept
compression	I-General_Concept
steps	O
do	O
n't	O
add	O
noise	O
into	O
,	O
or	O
otherwise	O
change	O
the	O
contents	O
(	O
unlike	O
quantization	B-Algorithm
)	O
,	O
it	O
is	O
sometimes	O
referred	O
to	O
as	O
noiseless	B-Protocol
coding	I-Protocol
.	O
</s>
<s>
Since	O
lossless	O
compression	O
aims	O
to	O
remove	O
as	O
much	O
redundancy	O
as	O
possible	O
,	O
it	O
is	O
known	O
as	O
entropy	B-Algorithm
coding	I-Algorithm
in	O
the	O
field	O
of	O
information	O
theory	O
.	O
</s>
<s>
The	O
coefficients	O
of	O
quantized	O
DCT	B-General_Concept
blocks	O
tend	O
to	O
zero	O
towards	O
the	O
bottom-right	O
.	O
</s>
<s>
Maximum	O
compression	O
can	O
be	O
achieved	O
by	O
a	O
zig-zag	O
scanning	O
of	O
the	O
DCT	B-General_Concept
block	O
starting	O
from	O
the	O
top	O
left	O
and	O
using	O
Run-length	B-Algorithm
encoding	I-Algorithm
techniques	O
.	O
</s>
<s>
The	O
DC	O
coefficients	O
and	O
motion	O
vectors	O
are	O
DPCM-encoded	O
.	O
</s>
<s>
Run-length	B-Algorithm
encoding	I-Algorithm
(	O
RLE	O
)	O
is	O
a	O
simple	O
method	O
of	O
compressing	O
repetition	O
.	O
</s>
<s>
RLE	O
is	O
particularly	O
effective	O
after	O
quantization	B-Algorithm
,	O
as	O
a	O
significant	O
number	O
of	O
the	O
AC	O
coefficients	O
are	O
now	O
zero	O
(	O
called	O
sparse	O
data	O
)	O
,	O
and	O
can	O
be	O
represented	O
with	O
just	O
a	O
couple	O
of	O
bytes	O
.	O
</s>
<s>
This	O
is	O
stored	O
in	O
a	O
special	O
2-dimensional	O
Huffman	O
table	O
that	O
codes	O
the	O
run-length	B-Algorithm
and	O
the	O
run-ending	O
character	O
.	O
</s>
<s>
Huffman	B-General_Concept
Coding	I-General_Concept
is	O
a	O
very	O
popular	O
and	O
relatively	O
simple	O
method	O
of	O
entropy	B-Algorithm
coding	I-Algorithm
,	O
and	O
used	O
in	O
MPEG-1	B-Algorithm
video	O
to	O
reduce	O
the	O
data	O
size	O
.	O
</s>
<s>
The	O
decoder	B-General_Concept
simply	O
reverses	O
this	O
process	O
to	O
produce	O
the	O
original	O
data	O
.	O
</s>
<s>
This	O
is	O
the	O
final	O
step	O
in	O
the	O
video	O
encoding	O
process	O
,	O
so	O
the	O
result	O
of	O
Huffman	B-General_Concept
coding	I-General_Concept
is	O
known	O
as	O
the	O
MPEG-1	B-Algorithm
video	O
"	O
bitstream.	O
"	O
</s>
<s>
Therefore	O
,	O
processing	O
of	O
B-frames	O
requires	O
more	O
buffer	B-General_Concept
on	O
the	O
decoded	O
side	O
.	O
</s>
<s>
A	O
configuration	O
of	O
the	O
Group	B-General_Concept
of	I-General_Concept
Pictures	I-General_Concept
(	O
GOP	O
)	O
should	O
be	O
selected	O
based	O
on	O
these	O
factors	O
.	O
</s>
<s>
Such	O
configurations	O
are	O
therefore	O
not	O
suited	O
for	O
video-telephony	O
or	O
video-conferencing	B-Application
applications	O
.	O
</s>
<s>
Part	O
3	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
covers	O
audio	O
and	O
is	O
defined	O
in	O
ISO/IEC	O
-11172-3	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
Audio	O
utilizes	O
psychoacoustics	O
to	O
significantly	O
reduce	O
the	O
data	O
rate	O
required	O
by	O
an	O
audio	O
stream	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
Audio	O
is	O
divided	O
into	O
3	O
layers	O
.	O
</s>
<s>
A	O
"	O
Full	O
"	O
Layer	O
II	O
decoder	B-General_Concept
can	O
also	O
play	O
Layer	O
I	O
audio	O
,	O
but	O
not	O
Layer	O
III	O
audio	O
,	O
although	O
not	O
all	O
higher	O
level	O
players	O
are	O
"	O
full	O
"	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
Audio	O
Layer	O
I	O
is	O
a	O
simplified	O
version	O
of	O
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
.	O
</s>
<s>
It	O
has	O
lower	O
complexity	O
than	O
Layer	O
II	O
to	O
facilitate	O
real-time	B-General_Concept
encoding	O
on	O
the	O
hardware	O
available	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
(	O
the	O
first	O
version	O
of	O
MP2	O
,	O
often	O
informally	O
called	O
MUSICAM	O
)	O
is	O
a	O
lossy	B-Algorithm
audio	O
format	O
designed	O
to	O
provide	O
high	O
quality	O
at	O
about	O
192	O
kbit/s	O
for	O
stereo	O
sound	O
.	O
</s>
<s>
Decoding	O
MP2	O
audio	O
is	O
computationally	O
simple	O
relative	O
to	O
MP3	B-Application
,	O
AAC	B-Application
,	O
etc	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
was	O
derived	O
from	O
the	O
MUSICAM	O
(	O
Masking	O
pattern	O
adapted	O
Universal	O
Subband	O
Integrated	O
Coding	O
And	O
Multiplexing	B-Architecture
)	O
audio	O
codec	B-General_Concept
,	O
developed	O
by	O
Centre	O
commun	O
d'études	O
de	O
télévision	O
et	O
télécommunications	O
(	O
CCETT	O
)	O
,	O
Philips	O
,	O
and	O
Institut	O
für	O
Rundfunktechnik	O
(	O
IRT/CNET	O
)	O
as	O
part	O
of	O
the	O
EUREKA	O
147	O
pan-European	O
inter-governmental	O
research	O
and	O
development	O
initiative	O
for	O
the	O
development	O
of	O
digital	O
audio	O
broadcasting	O
.	O
</s>
<s>
Most	O
key	O
features	O
of	O
MPEG-1	B-Algorithm
Audio	O
were	O
directly	O
inherited	O
from	O
MUSICAM	O
,	O
including	O
the	O
filter	O
bank	O
,	O
time-domain	O
processing	O
,	O
audio	O
frame	O
sizes	O
,	O
etc	O
.	O
</s>
<s>
However	O
,	O
improvements	O
were	O
made	O
,	O
and	O
the	O
actual	O
MUSICAM	O
algorithm	O
was	O
not	O
used	O
in	O
the	O
final	O
MPEG-1	B-Algorithm
Audio	O
Layer	O
II	O
standard	O
.	O
</s>
<s>
polyphased	O
)	O
to	O
prevent	O
aliasing	B-Error_Name
.	O
</s>
<s>
Time	O
domain	O
refers	O
to	O
how	O
analysis	O
and	O
quantization	B-Algorithm
is	O
performed	O
on	O
short	O
,	O
discrete	O
samples/chunks	O
of	O
the	O
audio	O
waveform	O
.	O
</s>
<s>
This	O
offers	O
low	O
delay	O
as	O
only	O
a	O
small	O
number	O
of	O
samples	O
are	O
analyzed	O
before	O
encoding	O
,	O
as	O
opposed	O
to	O
frequency	O
domain	O
encoding	O
(	O
like	O
MP3	B-Application
)	O
which	O
must	O
analyze	O
many	O
times	O
more	O
samples	O
before	O
it	O
can	O
decide	O
how	O
to	O
transform	O
and	O
output	O
encoded	O
audio	O
.	O
</s>
<s>
This	O
also	O
offers	O
higher	O
performance	O
on	O
complex	O
,	O
random	O
and	O
transient	O
impulses	O
(	O
such	O
as	O
percussive	O
instruments	O
,	O
and	O
applause	O
)	O
,	O
offering	O
avoidance	O
of	O
artifacts	B-Algorithm
like	O
pre-echo	O
.	O
</s>
<s>
The	O
32	O
sub-band	O
filter	O
bank	O
returns	O
32	O
amplitude	B-Application
coefficients	O
,	O
one	O
for	O
each	O
equal-sized	O
frequency	O
band/segment	O
of	O
the	O
audio	O
,	O
which	O
is	O
about	O
700Hz	O
wide	O
(	O
depending	O
on	O
the	O
audio	O
's	O
sampling	O
frequency	O
)	O
.	O
</s>
<s>
The	O
encoder	O
then	O
utilizes	O
the	O
psychoacoustic	O
model	O
to	O
determine	O
which	O
sub-bands	O
contain	O
audio	O
information	O
that	O
is	O
less	O
important	O
,	O
and	O
so	O
,	O
where	O
quantization	B-Algorithm
will	O
be	O
inaudible	O
,	O
or	O
at	O
least	O
much	O
less	O
noticeable	O
.	O
</s>
<s>
The	O
psychoacoustic	O
model	O
uses	O
an	O
empirically	O
determined	O
masking	O
model	O
to	O
determine	O
which	O
sub-bands	O
contribute	O
more	O
to	O
the	O
masking	O
threshold	O
,	O
and	O
how	O
much	O
quantization	B-Algorithm
noise	O
each	O
can	O
contain	O
without	O
being	O
perceived	O
.	O
</s>
<s>
Less	O
significant	O
sub-bands	O
are	O
reduced	O
in	O
accuracy	O
by	O
quantization	B-Algorithm
.	O
</s>
<s>
This	O
basically	O
involves	O
compressing	O
the	O
frequency	O
range	O
(	O
amplitude	B-Application
of	O
the	O
coefficient	O
)	O
,	O
i.e.	O
</s>
<s>
Then	O
computing	O
an	O
amplification	O
factor	O
,	O
for	O
the	O
decoder	B-General_Concept
to	O
use	O
to	O
re-expand	O
each	O
sub-band	O
to	O
the	O
proper	O
frequency	O
range	O
.	O
</s>
<s>
This	O
means	O
that	O
the	O
frequencies	O
above	O
6kHz	O
of	O
both	O
channels	O
are	O
combined/down	O
-mixed	O
into	O
one	O
single	O
(	O
mono	O
)	O
channel	O
,	O
but	O
the	O
"	O
side	O
channel	O
"	O
information	O
on	O
the	O
relative	O
intensity	O
(	O
volume	O
,	O
amplitude	B-Application
)	O
of	O
each	O
channel	O
is	O
preserved	O
and	O
encoded	O
into	O
the	O
bitstream	O
separately	O
.	O
</s>
<s>
This	O
can	O
allow	O
further	O
reduction	O
of	O
the	O
audio	O
bitrate	O
without	O
much	O
perceivable	O
loss	O
of	O
fidelity	O
,	O
but	O
is	O
generally	O
not	O
used	O
with	O
higher	O
bitrates	O
as	O
it	O
does	O
not	O
provide	O
very	O
high	O
quality	O
(	O
transparent	B-Algorithm
)	O
audio	O
.	O
</s>
<s>
Subjective	O
audio	O
testing	O
by	O
experts	O
,	O
in	O
the	O
most	O
critical	O
conditions	O
ever	O
implemented	O
,	O
has	O
shown	O
MP2	O
to	O
offer	O
transparent	B-Algorithm
audio	O
compression	O
at	O
256	O
kbit/s	O
for	O
16-bit	O
44.1kHz	O
CD	O
audio	O
using	O
the	O
earliest	O
reference	O
implementation	O
(	O
more	O
recent	O
encoders	O
should	O
presumably	O
perform	O
even	O
better	O
)	O
.	O
</s>
<s>
MP2	O
remains	O
a	O
favoured	O
lossy	B-Algorithm
audio	O
coding	O
standard	O
due	O
to	O
its	O
particularly	O
high	O
audio	O
coding	O
performances	O
on	O
important	O
audio	O
material	O
such	O
as	O
castanet	O
,	O
symphonic	O
orchestra	O
,	O
male	O
and	O
female	O
voices	O
and	O
particularly	O
complex	O
and	O
high	O
energy	O
transients	O
(	O
impulses	O
)	O
like	O
percussive	O
sounds	O
:	O
triangle	O
,	O
glockenspiel	O
and	O
audience	O
applause	O
.	O
</s>
<s>
More	O
recent	O
testing	O
has	O
shown	O
that	O
MPEG	B-Algorithm
Multichannel	I-Algorithm
(	O
based	O
on	O
MP2	O
)	O
,	O
despite	O
being	O
compromised	O
by	O
an	O
inferior	O
matrixed	O
mode	O
(	O
for	O
the	O
sake	O
of	O
backwards	O
compatibility	O
)	O
rates	O
just	O
slightly	O
lower	O
than	O
much	O
more	O
recent	O
audio	O
codecs	B-General_Concept
,	O
such	O
as	O
Dolby	O
Digital	O
(	O
AC-3	O
)	O
and	O
Advanced	B-Application
Audio	I-Application
Coding	I-Application
(	O
AAC	B-Application
)	O
(	O
mostly	O
within	O
the	O
margin	O
of	O
errorand	O
substantially	O
superior	O
in	O
some	O
cases	O
,	O
such	O
as	O
audience	O
applause	O
)	O
.	O
</s>
<s>
The	O
MPEG-2	B-Application
AAC	I-Application
Stereo	O
verification	O
tests	O
reached	O
a	O
vastly	O
different	O
conclusion	O
,	O
however	O
,	O
showing	O
AAC	B-Application
to	O
provide	O
superior	O
performance	O
to	O
MP2	O
at	O
half	O
the	O
bitrate	O
.	O
</s>
<s>
MPEG-1	B-Application
Audio	I-Application
Layer	I-Application
III	I-Application
(	O
the	O
first	O
version	O
of	O
MP3	B-Application
)	O
is	O
a	O
lossy	B-Algorithm
audio	O
format	O
designed	O
to	O
provide	O
acceptable	O
quality	O
at	O
about	O
64	O
kbit/s	O
for	O
monaural	O
audio	O
over	O
single-channel	O
(	O
BRI	B-Protocol
)	O
ISDN	B-Protocol
links	O
,	O
and	O
128	O
kbit/s	O
for	O
stereo	O
sound	O
.	O
</s>
<s>
MPEG-1	B-Application
Audio	I-Application
Layer	I-Application
III	I-Application
was	O
derived	O
from	O
the	O
Adaptive	O
Spectral	O
Perceptual	O
Entropy	B-Algorithm
Coding	I-Algorithm
(	O
ASPEC	O
)	O
codec	B-General_Concept
developed	O
by	O
Fraunhofer	O
as	O
part	O
of	O
the	O
EUREKA	O
147	O
pan-European	O
inter-governmental	O
research	O
and	O
development	O
initiative	O
for	O
the	O
development	O
of	O
digital	O
audio	O
broadcasting	O
.	O
</s>
<s>
ASPEC	O
was	O
itself	O
based	O
on	O
Multiple	O
adaptive	O
Spectral	O
audio	O
Coding	O
(	O
MSC	O
)	O
by	O
E	O
.	O
F	O
.	O
Schroeder	O
,	O
Optimum	O
Coding	O
in	O
the	O
Frequency	O
domain	O
(	O
OCF	O
)	O
the	O
doctoral	O
thesis	O
by	O
Karlheinz	O
Brandenburg	O
at	O
the	O
University	O
of	O
Erlangen-Nuremberg	O
,	O
Perceptual	O
Transform	B-Algorithm
Coding	I-Algorithm
(	O
PXFM	B-Application
)	O
by	O
J	O
.	O
D	O
.	O
Johnston	O
at	O
AT&T	O
Bell	O
Labs	O
,	O
and	O
Transform	B-Algorithm
coding	I-Algorithm
of	O
audio	O
signals	O
by	O
Y	O
.	O
Mahieux	O
and	O
J	O
.	O
Petit	O
at	O
Institut	O
für	O
Rundfunktechnik	O
(	O
IRT/CNET	O
)	O
.	O
</s>
<s>
MP3	B-Application
is	O
a	O
frequency-domain	O
audio	O
transform	B-Algorithm
encoder	I-Algorithm
.	O
</s>
<s>
Even	O
though	O
it	O
utilizes	O
some	O
of	O
the	O
lower	O
layer	O
functions	O
,	O
MP3	B-Application
is	O
quite	O
different	O
from	O
MP2	O
.	O
</s>
<s>
MP3	B-Application
works	O
on	O
1152	O
samples	O
like	O
MP2	O
,	O
but	O
needs	O
to	O
take	O
multiple	O
frames	O
for	O
analysis	O
before	O
frequency-domain	O
(	O
MDCT	O
)	O
processing	O
and	O
quantization	B-Algorithm
can	O
be	O
effective	O
.	O
</s>
<s>
It	O
outputs	O
a	O
variable	O
number	O
of	O
samples	O
,	O
using	O
a	O
bit	O
buffer	B-General_Concept
to	O
enable	O
this	O
variable	O
bitrate	O
(	O
VBR	O
)	O
encoding	O
while	O
maintaining	O
1152	O
sample	O
size	O
output	O
frames	O
.	O
</s>
<s>
This	O
causes	O
a	O
significantly	O
longer	O
delay	O
before	O
output	O
,	O
which	O
has	O
caused	O
MP3	B-Application
to	O
be	O
considered	O
unsuitable	O
for	O
studio	O
applications	O
where	O
editing	O
or	O
other	O
processing	O
needs	O
to	O
take	O
place	O
.	O
</s>
<s>
MP3	B-Application
does	O
not	O
benefit	O
from	O
the	O
32	O
sub-band	O
polyphased	O
filter	O
bank	O
,	O
instead	O
just	O
using	O
an	O
18-point	O
MDCT	O
transformation	O
on	O
each	O
output	O
to	O
split	O
the	O
data	O
into	O
576	O
frequency	O
components	O
,	O
and	O
processing	O
it	O
in	O
the	O
frequency	O
domain	O
.	O
</s>
<s>
This	O
extra	O
granularity	O
allows	O
MP3	B-Application
to	O
have	O
a	O
much	O
finer	O
psychoacoustic	O
model	O
,	O
and	O
more	O
carefully	O
apply	O
appropriate	O
quantization	B-Algorithm
to	O
each	O
band	O
,	O
providing	O
much	O
better	O
low-bitrate	O
performance	O
.	O
</s>
<s>
This	O
causes	O
quantization	B-Algorithm
artifacts	B-Algorithm
,	O
due	O
to	O
transient	O
sounds	O
like	O
percussive	O
events	O
and	O
other	O
high-frequency	O
events	O
that	O
spread	O
over	O
a	O
larger	O
window	O
.	O
</s>
<s>
MP3	B-Application
uses	O
pre-echo	O
detection	O
routines	O
,	O
and	O
VBR	O
encoding	O
,	O
which	O
allows	O
it	O
to	O
temporarily	O
increase	O
the	O
bitrate	O
during	O
difficult	O
passages	O
,	O
in	O
an	O
attempt	O
to	O
reduce	O
this	O
effect	O
.	O
</s>
<s>
It	O
is	O
also	O
able	O
to	O
switch	O
between	O
the	O
normal	O
36	O
sample	O
quantization	B-Algorithm
window	O
,	O
and	O
instead	O
using	O
3	O
short	O
12	O
sample	O
windows	O
instead	O
,	O
to	O
reduce	O
the	O
temporal	O
(	O
time	O
)	O
length	O
of	O
quantization	B-Algorithm
artifacts	B-Algorithm
.	O
</s>
<s>
And	O
yet	O
in	O
choosing	O
a	O
fairly	O
small	O
window	O
size	O
to	O
make	O
MP3	B-Application
's	O
temporal	O
response	O
adequate	O
enough	O
to	O
avoid	O
the	O
most	O
serious	O
artifacts	B-Algorithm
,	O
MP3	B-Application
becomes	O
much	O
less	O
efficient	O
in	O
frequency	O
domain	O
compression	O
of	O
stationary	O
,	O
tonal	O
components	O
.	O
</s>
<s>
Being	O
forced	O
to	O
use	O
a	O
hybrid	O
time	O
domain	O
(	O
filter	O
bank	O
)	O
/frequency	O
domain	O
(	O
MDCT	O
)	O
model	O
to	O
fit	O
in	O
with	O
Layer	O
II	O
simply	O
wastes	O
processing	O
time	O
and	O
compromises	O
quality	O
by	O
introducing	O
aliasing	B-Error_Name
artifacts	B-Algorithm
.	O
</s>
<s>
MP3	B-Application
has	O
an	O
aliasing	B-Error_Name
cancellation	O
stage	O
specifically	O
to	O
mask	O
this	O
problem	O
,	O
but	O
which	O
instead	O
produces	O
frequency	O
domain	O
energy	O
which	O
must	O
be	O
encoded	O
in	O
the	O
audio	O
.	O
</s>
<s>
Layer	O
II	O
's	O
1024	O
point	O
FFT	O
does	O
n't	O
entirely	O
cover	O
all	O
samples	O
,	O
and	O
would	O
omit	O
several	O
entire	O
MP3	B-Application
sub-bands	O
,	O
where	O
quantization	B-Algorithm
factors	O
must	O
be	O
determined	O
.	O
</s>
<s>
MP3	B-Application
instead	O
uses	O
two	O
passes	O
of	O
FFT	O
analysis	O
for	O
spectral	O
estimation	O
,	O
to	O
calculate	O
the	O
global	O
and	O
individual	O
masking	O
thresholds	O
.	O
</s>
<s>
In	O
addition	O
to	O
Layer	O
II	O
's	O
intensity	O
encoded	O
joint	O
stereo	O
,	O
MP3	B-Application
can	O
use	O
middle/side	O
(	O
mid/side	O
,	O
m/s	O
,	O
MS	O
,	O
matrixed	O
)	O
joint	O
stereo	O
.	O
</s>
<s>
When	O
combined	O
with	O
quantization	B-Algorithm
,	O
however	O
,	O
it	O
can	O
exaggerate	O
artifacts	B-Algorithm
.	O
</s>
<s>
An	O
MP3	B-Application
encoder	I-Application
can	O
switch	O
between	O
m/s	O
stereo	O
and	O
full	O
stereo	O
on	O
a	O
frame-by-frame	O
basis	O
.	O
</s>
<s>
Unlike	O
Layers	O
I	O
and	O
II	O
,	O
MP3	B-Application
uses	O
variable-length	O
Huffman	B-General_Concept
coding	I-General_Concept
(	O
after	O
perceptual	O
)	O
to	O
further	O
reduce	O
the	O
bitrate	O
,	O
without	O
any	O
further	O
quality	O
loss	O
.	O
</s>
<s>
MP3	B-Application
's	O
more	O
fine-grained	O
and	O
selective	O
quantization	B-Algorithm
does	O
prove	O
notably	O
superior	O
to	O
MP2	O
at	O
lower-bitrates	O
.	O
</s>
<s>
128	O
kbit/s	O
is	O
considered	O
the	O
"	O
sweet	O
spot	O
"	O
for	O
MP3	B-Application
;	O
meaning	O
it	O
provides	O
generally	O
acceptable	O
quality	O
stereo	O
sound	O
on	O
most	O
music	O
,	O
and	O
there	O
are	O
diminishing	O
quality	O
improvements	O
from	O
increasing	O
the	O
bitrate	O
further	O
.	O
</s>
<s>
MP3	B-Application
is	O
also	O
regarded	O
as	O
exhibiting	O
artifacts	B-Algorithm
that	O
are	O
less	O
annoying	O
than	O
Layer	O
II	O
,	O
when	O
both	O
are	O
used	O
at	O
bitrates	O
that	O
are	O
too	O
low	O
to	O
possibly	O
provide	O
faithful	O
reproduction	O
.	O
</s>
<s>
Layer	O
III	O
audio	O
files	O
use	O
the	O
extension	O
"	O
.mp3	B-Application
"	O
.	O
</s>
<s>
The	O
MPEG-2	B-Algorithm
standard	O
includes	O
several	O
extensions	O
to	O
MPEG-1	B-Algorithm
Audio	O
.	O
</s>
<s>
These	O
are	O
known	O
as	O
MPEG-2	B-Algorithm
BC	O
–	O
backwards	O
compatible	O
with	O
MPEG-1	B-Algorithm
Audio	O
.	O
</s>
<s>
MPEG-2	B-Algorithm
Audio	O
is	O
defined	O
in	O
ISO/IEC	B-Algorithm
13818-3	I-Algorithm
.	O
</s>
<s>
MPEG	B-Algorithm
Multichannel	I-Algorithm
–	O
Backward	O
compatible	O
5.1-channel	O
surround	O
sound	O
.	O
</s>
<s>
These	O
sampling	O
rates	O
are	O
exactly	O
half	O
that	O
of	O
those	O
originally	O
defined	O
for	O
MPEG-1	B-Algorithm
Audio	O
.	O
</s>
<s>
The	O
even-lower	O
bitrates	O
were	O
introduced	O
because	O
tests	O
showed	O
that	O
MPEG-1	B-Algorithm
Audio	O
could	O
provide	O
higher	O
quality	O
than	O
any	O
existing	O
(	O
)	O
very	O
low	O
bitrate	O
(	O
i.e.	O
</s>
<s>
speech	B-Application
)	O
audio	O
codecs	B-General_Concept
.	O
</s>
<s>
Part	O
4	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
covers	O
conformance	O
testing	O
,	O
and	O
is	O
defined	O
in	O
ISO/IEC	O
-11172-4	O
.	O
</s>
<s>
Provides	O
two	O
sets	O
of	O
guidelines	O
and	O
reference	O
bitstreams	O
for	O
testing	O
the	O
conformance	O
of	O
MPEG-1	B-Algorithm
audio	O
and	O
video	O
decoders	O
,	O
as	O
well	O
as	O
the	O
bitstreams	O
produced	O
by	O
an	O
encoder	O
.	O
</s>
<s>
Part	O
5	O
of	O
the	O
MPEG-1	B-Algorithm
standard	O
includes	O
reference	O
software	O
,	O
and	O
is	O
defined	O
in	O
ISO/IEC	O
TR	O
11172	O
–	O
5	O
.	O
</s>
<s>
C	B-Language
reference	O
code	O
for	O
encoding	O
and	O
decoding	O
of	O
audio	O
and	O
video	O
,	O
as	O
well	O
as	O
multiplexing	B-Architecture
and	O
demultiplexing	B-Architecture
.	O
</s>
<s>
This	O
includes	O
the	O
ISO	O
Dist10	O
audio	O
encoder	O
code	O
,	O
which	O
LAME	B-Algorithm
and	O
TooLAME	B-Algorithm
were	O
originally	O
based	O
upon	O
.	O
</s>
<s>
.mpg	O
is	O
one	O
of	O
a	O
number	O
of	O
file	O
extensions	O
for	O
MPEG-1	B-Algorithm
or	O
MPEG-2	B-Algorithm
audio	O
and	O
video	O
compression	O
.	O
</s>
<s>
MPEG-1	B-Algorithm
Part	O
2	O
video	O
is	O
rare	O
nowadays	O
,	O
and	O
this	O
extension	O
typically	O
refers	O
to	O
an	O
MPEG	O
program	O
stream	O
(	O
defined	O
in	O
MPEG-1	B-Algorithm
and	O
MPEG-2	B-Algorithm
)	O
or	O
MPEG	O
transport	O
stream	O
(	O
defined	O
in	O
MPEG-2	B-Algorithm
)	O
.	O
</s>
<s>
Other	O
suffixes	O
such	O
as	O
.m2ts	O
also	O
exist	O
specifying	O
the	O
precise	O
container	O
,	O
in	O
this	O
case	O
MPEG-2	B-Algorithm
TS	O
,	O
but	O
this	O
has	O
little	O
relevance	O
to	O
MPEG-1	B-Algorithm
media	O
.	O
</s>
<s>
.mp3	B-Application
is	O
the	O
most	O
common	O
extension	O
for	O
files	O
containing	O
MP3	B-Application
audio	I-Application
(	O
typically	O
MPEG-1	B-Algorithm
Audio	O
,	O
sometimes	O
MPEG-2	B-Algorithm
Audio	O
)	O
.	O
</s>
<s>
An	O
MP3	B-Application
file	I-Application
is	O
typically	O
an	O
uncontained	O
stream	O
of	O
raw	O
audio	O
;	O
the	O
conventional	O
way	O
to	O
tag	O
MP3	B-Application
files	I-Application
is	O
by	O
writing	O
data	O
to	O
"	O
garbage	O
"	O
segments	O
of	O
each	O
frame	O
,	O
which	O
preserve	O
the	O
media	O
information	O
but	O
are	O
discarded	O
by	O
the	O
player	O
.	O
</s>
<s>
This	O
is	O
similar	O
in	O
many	O
respects	O
to	O
how	O
raw	O
.AAC	B-Application
files	O
are	O
tagged	O
(	O
but	O
this	O
is	O
less	O
supported	O
nowadays	O
,	O
e.g.	O
</s>
<s>
iTunes	B-Operating_System
)	O
.	O
</s>
<s>
Note	O
that	O
although	O
it	O
would	O
apply	O
,	O
.mpg	O
does	O
not	O
normally	O
append	O
raw	O
AAC	B-Application
or	O
AAC	B-Application
in	O
MPEG-2	B-Application
Part	I-Application
7	I-Application
Containers	O
.	O
</s>
<s>
The	O
.aac	B-Application
extension	O
normally	O
denotes	O
these	O
audio	O
files	O
.	O
</s>
