Your company here — click to reach over 10,000 unique daily visitors

roc-send - Man Page

send real-time audio


roc-send Options


Read audio stream from an audio device or file and send it to remote receiver.


-h,  --help

Print help and exit

-V,  --version

Print version and exit

-v,  --verbose

Increase verbosity level (may be used multiple times)

-L,  --list-supported

list supported schemes and formats


Input file or device URI


Force input file format


Remote source endpoint


Remote repair endpoint


Remote control endpoint


enable So_reuseaddr when binding sockets


Recording target latency, TIME units


Number of source packets in FEC block


Number of repair packets in FEC block


Outgoing packet length, TIME units


Maximum packet size, in bytes


Maximum internal frame size, in bytes


Duration of the internal frames, TIME units


Override input sample rate, Hz


Resampler backend  (possible values="default", "builtin", "speex", "speexdec" default=`default')


Resampler profile  (possible values="low", "medium", "high" default=`medium')


Enable packet interleaving  (default=off)


Enable self profiling  (default=off)


Set colored logging mode for stderr output (possible values="auto", "always", "never" default=`auto')

Endpoint URI

--source, --repair, and --control options define network endpoints to which to send the traffic.

ENDPOINT_URI should have the following form:



  • rtsp://localhost:123/path?query
  • rtp+rs8m://localhost:123
  • rtp://
  • rtp://[::1]:123
  • rtcp://

The list of supported protocols can be retrieved using --list-supported option.

The host field should be either FQDN (domain name), or IPv4 address, or IPv6 address in square brackets.

The port field can be omitted if the protocol defines standard port. Otherwise, it is mandatory.

The path and query fields are allowed only for protocols that support them, e.g. for RTSP.

If FEC is enabled, a pair of a source and repair endpoints should be provided. The two endpoints should use compatible protocols, e.g. rtp+rs8m:// for source endpoint, and rs8m:// for repair endpoint. If FEC is disabled, a single source endpoint should be provided.

Supported source and repair protocols:

  • source rtp://, repair none (bare RTP without FEC)
  • source rtp+rs8m://, repair rs8m:// (RTP with Reed-Solomon FEC)
  • source rtp+ldpc://, repair ldpc:// (RTP with LDPC-Staircase FEC)

In addition, it is recommended to provide control endpoint. It is used to exchange non-media information used to identify session, carry feedback, etc. If no control endpoint is provided, session operates in reduced fallback mode, which may be less robust and may not support all features.

Supported control protocols:

  • rtcp://


--input option requires a device or file URI in one of the following forms:

  • DEVICE_TYPE://DEVICE_NAME -- audio device
  • DEVICE_TYPE://default -- default audio device for given device type
  • file:///ABS/PATH -- absolute file path
  • file://localhost/ABS/PATH -- absolute file path (alternative form; only "localhost" host is supported)
  • file:/ABS/PATH -- absolute file path (alternative form)
  • file:REL/PATH -- relative file path
  • file://- -- stdin
  • file:- -- stdin (alternative form)


  • pulse://default
  • pulse://alsa_input.pci-0000_00_1f.3.analog-stereo
  • alsa://hw:1,0
  • file:///home/user/test.wav
  • file://localhost/home/user/test.wav
  • file:/home/user/test.wav
  • file:./test.wav
  • file:-

The list of supported schemes and file formats can be retrieved using --list-supported option.

If the --input is omitted, the default driver and device are selected.

The --input-format option can be used to force the input file format. If it is omitted, the file format is auto-detected. This option is always required when the input is stdin.

The path component of the provided URI is percent-decoded. For convenience, unencoded characters are allowed as well, except that % should be always encoded as %25.

For example, the file named /foo/bar%/[baz] may be specified using either of the following URIs: file:///foo%2Fbar%25%2F%5Bbaz%5D and file:///foo/bar%25/[baz].

Multiple slots

Multiple sets of endpoints can be specified to send media to multiple addresses.

Such endpoint sets are called slots. All slots should have the same set of endpoint types (source, repair, etc) and should use the same protocols for them.


If --reuseaddr option is provided, SO_REUSEADDR socket option will be enabled for all sockets.

For TCP, it allows immediately reusing recently closed socket in TIME_WAIT state, which may be useful you want to be able to restart server quickly.

For UDP, it allows multiple processes to bind to the same address, which may be useful if you're using systemd socket activation.

Regardless of the option, SO_REUSEADDR is always disabled when binding to ephemeral port.

Time units

TIME should have one of the following forms:

123ns, 123us, 123ms, 123s, 123m, 123h


Endpoint examples

Send file to receiver with one bare RTP endpoint:

$ roc-send -vv -i file:./input.wav -s rtp://

Send file to receiver with IPv4 source, repair, and control endpoints:

$ roc-send -vv -i file:./input.wav -s rtp+rs8m:// \
    -r rs8m:// -c rtcp://

Send file to receiver with IPv6 source, repair, and control endpoints:

$ roc-send -vv -i file:./input.wav -s rtp+rs8m://[2001:db8::]:10001 \
    -r rs8m://[2001:db8::]:10002 -r rtcp://[2001:db8::]:10003

Send file to two receivers, each with three endpoints:

$ roc-send -vv \
    -i file:./input.wav \
    -s rtp+rs8m:// -r rs8m:// \
        -c rtcp:// \
    -s rtp+rs8m:// -r rs8m:// \
        -c rtcp://

I/O examples

Capture sound from the default device (omit -i):

$ roc-send -vv -s rtp://

Capture sound from the default ALSA device:

$ roc-send -vv -s rtp:// -i alsa://default

Capture sound from a specific PulseAudio device:

$ roc-send -vv -s rtp:// -i pulse://alsa_input.pci-0000_00_1f.3.analog-stereo

Send WAV file (guess format by extension):

$ roc-send -vv -s rtp:// -i file:./input.wav

Send WAV file (specify format manually):

$ roc-send -vv -s rtp:// -i file:./input.file --input-format wav

Send WAV from stdin:

$ roc-send -vv -s rtp:// -i file:- --input-format wav <./input.wav

Send WAV file (specify absolute path):

$ roc-send -vv -s rtp:// -i file:///home/user/input.wav

Tuning examples

Force a specific rate on the input device:

$ roc-send -vv -s rtp:// --rate=44100

Select the LDPC-Staircase FEC scheme and a larger block size:

$ roc-send -vv -i file:./input.wav -s rtp+ldpc:// \
    -r ldpc:// -c ldpc:// \
    --nbsrc=1000 --nbrpr=500

Select lower packet length:

$ roc-send -vv -i file:./input.wav -s rtp+ldpc:// \
    --packet-length 2500us

Select lower I/O latency and frame length:

$ roc-send -vv -s rtp:// \
    --io-latency=20ms --frame-length 4ms

Manually specify resampling parameters:

$ roc-send -vv -s rtp:// \
    --resampler-backend=speex --resampler-profile=high

See Also

roc-recv(1), and the Roc web site at https://roc-streaming.org/


Please report any bugs found via GitHub (https://github.com/roc-streaming/roc-toolkit/).


See authors page on the website for a list of maintainers and contributors.

Referenced By

roc-copy(1), roc-recv(1).

2024 Roc Toolkit 0.3