gstreamer/gst-launch.markdown

# gst-launch

This page last changed on May 30, 2012 by xartigas.

<table>
<tbody>
<tr class="odd">
<td><img src="images/icons/emoticons/information.png" width="16" height="16" /></td>
<td><p>This is the Linux man page for the <code>gst-launch</code> tool. As such, it is very Linux-centric regarding path specification and plugin names. Please be patient while it is rewritten to be more generic.</p></td>
</tr>
</tbody>
</table>

## Name

gst-launch - build and run a GStreamer pipeline

## Synopsis

**gst-launch** *\[OPTION...\]* PIPELINE-DESCRIPTION

## Description

*gst-launch* is a tool that builds and runs basic *GStreamer* pipelines.

In simple form, a PIPELINE-DESCRIPTION is a list of elements separated
by exclamation marks (\!). Properties may be appended to elements, in
the form*property=value*.

For a complete description of possible PIPELINE-DESCRIPTIONS see the
section*pipeline description* below or consult the GStreamer
documentation.

Please note that *gst-launch* is primarily a debugging tool for
developers and users. You should not build applications on top of it.
For applications, use the gst\_parse\_launch() function of the GStreamer
API as an easy way to construct pipelines from pipeline descriptions.

## Options

*gst-launch* accepts the following options:

**--help**

Print help synopsis and available FLAGS

**-v, --verbose**

Output status information and property notifications

**-q, --quiet**

Do not print any progress information

**-m, --messages**

Output messages posted on the pipeline's bus

**-t, --tags**

Output tags (also known as metadata)

**-o FILE, --output=FILE**

Save XML representation of pipeline to FILE and exit

**-f, --no\_fault**

Do not install a fault handler

**-T, --trace**

Print memory allocation traces. The feature must be enabled at compile
time to work.

 

## Gstreamer Options

*gst-launch* also accepts the following options that are common to all
GStreamer applications:

## Pipeline Description

A pipeline consists *elements* and *links*. *Elements* can be put
into *bins* of different sorts. *Elements*, *links* and *bins* can be
specified in a pipeline description in any order.

**Elements**

ELEMENTTYPE *\[PROPERTY1 ...\]*

Creates an element of type ELEMENTTYPE and sets the PROPERTIES.

**Properties**

PROPERTY=VALUE ...

Sets the property to the specified value. You can use **gst-inspect**(1)
to find out about properties and allowed values of different elements.
Enumeration properties can be set by name, nick or value.

**Bins**

*\[BINTYPE.\]* ( *\[PROPERTY1 ...\]* PIPELINE-DESCRIPTION )

Specifies that a bin of type BINTYPE is created and the given properties
are set. Every element between the braces is put into the bin. Please
note the dot that has to be used after the BINTYPE. You will almost
never need this functionality, it is only really useful for applications
using the gst\_launch\_parse() API with 'bin' as bintype. That way it is
possible to build partial pipelines instead of a full-fledged top-level
pipeline.

**Links**

*\[\[SRCELEMENT\].\[PAD1,...\]\]* \! *\[\[SINKELEMENT\].\[PAD1,...\]\]
\[\[SRCELEMENT\].\[PAD1,...\]\]* \! CAPS
\! *\[\[SINKELEMENT\].\[PAD1,...\]\]*

Links the element with name SRCELEMENT to the element with name
SINKELEMENT, using the caps specified in CAPS as a filter. Names can be
set on elements with the name property. If the name is omitted, the
element that was specified directly in front of or after the link is
used. This works across bins. If a padname is given, the link is done
with these pads. If no pad names are given all possibilities are tried
and a matching pad is used. If multiple padnames are given, both sides
must have the same number of pads specified and multiple links are done
in the given order.
So the simplest link is a simple exclamation mark, that links the
element to the left of it to the element right of it.

**Caps**

MIMETYPE *\[, PROPERTY\[, PROPERTY ...\]\]\] \[; CAPS\[; CAPS ...\]\]*

Creates a capability with the given mimetype and optionally with given
properties. The mimetype can be escaped using " or '. If you want to
chain caps, you can add more caps in the same format afterwards.

**Properties**

NAME=*\[(TYPE)\]*VALUE
in lists and ranges: *\[(TYPE)\]*VALUE

Sets the requested property in capabilities. The name is an alphanumeric
value and the type can have the following case-insensitive values:
\- **i** or **int** for integer values or ranges
\- **f** or **float** for float values or ranges
\- **4** or **fourcc** for FOURCC values
\- **b**, **bool** or **boolean** for boolean values
\- **s**, **str** or **string** for strings
\- **fraction** for fractions (framerate, pixel-aspect-ratio)
\- **l** or **list** for lists
If no type was given, the following order is tried: integer, float,
boolean, string.
Integer values must be parsable by **strtol()**, floats by **strtod()**.
FOURCC values may either be integers or strings. Boolean values are
(case insensitive) *yes*, *no*, *true* or *false* and may like strings
be escaped with " or '.
Ranges are in this format: \[ VALUE, VALUE \]
Lists use this format: ( VALUE *\[, VALUE ...\]* )

## Pipeline Control

A pipeline can be controlled by signals. SIGUSR2 will stop the pipeline
(GST\_STATE\_NULL); SIGUSR1 will put it back to play
(GST\_STATE\_PLAYING). By default, the pipeline will start in the
playing state.
There are currently no signals defined to go into the ready or pause
(GST\_STATE\_READY and GST\_STATE\_PAUSED) state explicitely.

## Pipeline Examples

The examples below assume that you have the correct plug-ins available.
In general, "osssink" can be substituted with another audio output
plug-in such as "directsoundsink", "esdsink", "alsasink",
"osxaudiosink", or "artsdsink". Likewise, "xvimagesink" can be
substituted with "d3dvideosink", "ximagesink", "sdlvideosink",
"osxvideosink", or "aasink". Keep in mind though that different sinks
might accept different formats and even the same sink might accept
different formats on different machines, so you might need to add
converter elements like audioconvert and audioresample (for audio) or
ffmpegcolorspace (for video) in front of the sink to make things work.

**Audio playback**

**gst-launch filesrc location=music.mp3 \! mad \! audioconvert \!
audioresample \! osssink**
Play the mp3 music file "music.mp3" using a libmad-based plug-in and
output to an OSS device

**gst-launch filesrc location=music.ogg \! oggdemux \! vorbisdec \!
audioconvert \! audioresample \! osssink**
Play an Ogg Vorbis format file

**gst-launch gnomevfssrc location=music.mp3 \! mad \! osssink
gst-launch gnomevfssrc location=<http://domain.com/music.mp3> \! mad \!
audioconvert \! audioresample \! osssink**
Play an mp3 file or an http stream using GNOME-VFS

**gst-launch gnomevfssrc location=<smb://computer/music.mp3> \! mad \!
audioconvert \! audioresample \! osssink**
Use GNOME-VFS to play an mp3 file located on an SMB server

**Format conversion**

**gst-launch filesrc location=music.mp3 \! mad \! audioconvert \!
vorbisenc \! oggmux \! filesink location=music.ogg**
Convert an mp3 music file to an Ogg Vorbis file

**gst-launch filesrc location=music.mp3 \! mad \! audioconvert \!
flacenc \! filesink location=test.flac**
Convert to the FLAC format

**Other**

**gst-launch filesrc location=music.wav \! wavparse \! audioconvert \!
audioresample \! osssink**
Plays a .WAV file that contains raw audio data (PCM).

**gst-launch filesrc location=music.wav \! wavparse \! audioconvert \!
vorbisenc \! oggmux \! filesink location=music.ogg
gst-launch filesrc location=music.wav \! wavparse \! audioconvert \!
lame \! filesink location=music.mp3**
Convert a .WAV file containing raw audio data into an Ogg Vorbis or mp3
file

**gst-launch cdparanoiasrc mode=continuous \! audioconvert \! lame \!
id3v2mux \! filesink location=cd.mp3**
rips all tracks from compact disc and convert them into a single mp3
file

**gst-launch cdparanoiasrc track=5 \! audioconvert \! lame \! id3v2mux
\! filesink location=track5.mp3**
rips track 5 from the CD and converts it into a single mp3 file

Using **gst-inspect**(1), it is possible to discover settings like the
above for cdparanoiasrc that will tell it to rip the entire cd or only
tracks of it. Alternatively, you can use an URI and gst-launch-0.10 will
find an element (such as cdparanoia) that supports that protocol for
you, e.g.: **gst-launch [cdda://5]() \! lame vbr=new vbr-quality=6 \!
filesink location=track5.mp3**

**gst-launch osssrc \! audioconvert \! vorbisenc \! oggmux \! filesink
location=input.ogg**
records sound from your audio input and encodes it into an ogg file

**Video**

**gst-launch filesrc location=JB\_FF9\_TheGravityOfLove.mpg \! dvddemux
\! mpeg2dec \! xvimagesink**
Display only the video portion of an MPEG-1 video file, outputting to an
X display window

**gst-launch filesrc location=/flflfj.vob \! dvddemux \! mpeg2dec \!
sdlvideosink**
Display the video portion of a .vob file (used on DVDs), outputting to
an SDL window

**gst-launch filesrc location=movie.mpg \! dvddemux name=demuxer
demuxer. \! queue \! mpeg2dec \! sdlvideosink demuxer. \! queue \! mad
\! audioconvert \! audioresample \! osssink**
Play both video and audio portions of an MPEG movie

**gst-launch filesrc location=movie.mpg \! mpegdemux name=demuxer
demuxer. \! queue \! mpeg2dec \! ffmpegcolorspace \! sdlvideosink
demuxer. \! queue \! mad \! audioconvert \! audioresample \! osssink**
Play an AVI movie with an external text subtitle stream

This example also shows how to refer to specific pads by name if an
element (here: textoverlay) has multiple sink or source pads.

**gst-launch textoverlay name=overlay \! ffmpegcolorspace \! videoscale
\! autovideosink filesrc location=movie.avi \! decodebin2 \!
ffmpegcolorspace \! overlay.video\_sink filesrc location=movie.srt \!
subparse \! overlay.text\_sink**

Play an AVI movie with an external text subtitle stream using playbin2

**gst-launch playbin2 uri=<file:///path/to/movie.avi>
suburi=<file:///path/to/movie.srt>**

**Network streaming**

Stream video using RTP and network elements.

**gst-launch v4l2src \!
video/x-raw-yuv,width=128,height=96,format='(fourcc)'UYVY \!
ffmpegcolorspace \! ffenc\_h263 \! video/x-h263 \! rtph263ppay pt=96 \!
udpsink host=192.168.1.1 port=5000 sync=false**
Use this command on the receiver

**gst-launch udpsrc port=5000 \! application/x-rtp,
clock-rate=90000,payload=96 \! rtph263pdepay queue-delay=0 \!
ffdec\_h263 \! xvimagesink**
This command would be run on the transmitter

**Diagnostic**

**gst-launch -v fakesrc num-buffers=16 \! fakesink**
Generate a null stream and ignore it (and print out details).

**gst-launch audiotestsrc \! audioconvert \! audioresample \!
osssink**
Generate a pure sine tone to test the audio output

**gst-launch videotestsrc \! xvimagesink
gst-launch videotestsrc \! ximagesink**
Generate a familiar test pattern to test the video output

**Automatic linking**

You can use the decodebin element to automatically select the right
elements to get a working pipeline.

**gst-launch filesrc location=musicfile \! decodebin \! audioconvert \!
audioresample \! osssink**
Play any supported audio format

**gst-launch filesrc location=videofile \! decodebin name=decoder
decoder. \! queue \! audioconvert \! audioresample \! osssink decoder.
\! ffmpegcolorspace \! xvimagesink**
Play any supported video format with video and audio output. Threads are
used automatically. To make this even easier, you can use the playbin
element:

**gst-launch playbin uri=<file:///home/joe/foo.avi>**

**Filtered connections**

These examples show you how to use filtered caps.

**gst-launch videotestsrc \!
'video/x-raw-yuv,format=(fourcc)YUY2;video/x-raw-yuv,format=(fourcc)YV12'
\! xvimagesink**
Show a test image and use the YUY2 or YV12 video format for this.

**gst-launch osssrc \!
'audio/x-raw-int,rate=\[32000,64000\],width=\[16,32\],depth={16,24,32},signed=(boolean)true'
\! wavenc \! filesink location=recording.wav**
record audio and write it to a .wav file. Force usage of signed 16 to 32
bit samples and a sample rate between 32kHz and 64KHz.

## Environment Variables

**GST\_DEBUG**

Comma-separated list of debug categories and levels, e.g. GST\_DEBUG=
totem:4,typefind:5

**GST\_DEBUG\_NO\_COLOR**[](http://totem:4,typefind:5)

When this environment variable is set, coloured debug output is
disabled.

**GST\_DEBUG\_DUMP\_DOT\_DIR**

When set to a filesystem path, store dot files of pipeline graphs there.

**GST\_REGISTRY**

Path of the plugin registry file. Default is
~/.gstreamer-0.10/registry-CPU.xml where CPU is the machine/cpu type
GStreamer was compiled for, e.g. 'i486', 'i686', 'x86-64', 'ppc', etc.
(check the output of "uname -i" and "uname -m" for details).

**GST\_REGISTRY\_UPDATE**

Set to "no" to force GStreamer to assume that no plugins have changed,
been added or been removed. This will make GStreamer skip the initial
check whether a rebuild of the registry cache is required or not. This
may be useful in embedded environments where the installed plugins never
change. Do not use this option in any other setup.

**GST\_PLUGIN\_PATH**

Specifies a list of directories to scan for additional plugins. These
take precedence over the system plugins.

**GST\_PLUGIN\_SYSTEM\_PATH**

Specifies a list of plugins that are always loaded by default. If not
set, this defaults to the system-installed path, and the plugins
installed in the user's home directory

**OIL\_CPU\_FLAGS**

Useful liboil environment variable. Set OIL\_CPU\_FLAGS=0 when valgrind
or other debugging tools trip over liboil's CPU detection (quite a few
important GStreamer plugins like videotestsrc, audioconvert or
audioresample use liboil).

**G\_DEBUG**

Useful GLib environment variable. Set G\_DEBUG=fatal\_warnings to make
GStreamer programs abort when a critical warning such as an assertion
failure occurs. This is useful if you want to find out which part of the
code caused that warning to be triggered and under what circumstances.
Simply set G\_DEBUG as mentioned above and run the program in gdb (or
let it core dump). Then get a stack trace in the usual way

<!-- end list -->

Document generated by Confluence on Oct 08, 2015 10:28