sound_play

Dependencies

• python-pygame
• festival
• festvox-don
• alsa-base
• alsa-tools

Checking that the speaker/sound card is recognized by the kernel

cat /proc/asound/cards

Your card should be in the list. Make note of the number in front of the card, it will be used to tell alsa where to play sound from.

If your sound device does not show up, your kernel may not support it, or the module may not be loaded. For usb speakers, you may want to try:

modprobe snd-usb-audio

(not sure if this list is exhaustive)

Telling alsa which sound card/speaker to use

Run (replace 75 with the number of the sound device to use):

asoundconf set-default-card 75

This will create .asoundrc.asoundconf in your home directory. To make alsa use these settings, add the following line to ~/.asoundrc

include ".asoundrc.asoundconf"

To set this default to all users, copy this to the system-wide alsa configuration file:

mv ~/.asoundrc.asoundconf /etc/asound.conf

Getting started

Start the sound play node, and have a look at the scripts in the scripts directory that exercise the node’s functionality.

Specify Device via ROS Param

Besides setting default device as system wide settings, you can also specify audio device via rosparam:

<launch>
  <node name="soundplay_node" pkg="sound_play" type="soundplay_node.py">
    <param name="device" value="hw:1,0" />
  </node>
</launch>

or simply run: rosrun sound_play soundplay_node.py _device:="hw:1,0"

In the launch file above, ~device parameter is set to hw:1,0, which tells soundplay_node to use audio device No. 0 connected to audio card No.1. To find card/device number which you want to use, execute:

sudo aplay -l

Changelog for package sound_play

0.3.18 (2024-08-13)

• Merge pull request #249 from peci1/patch-1 festival_plugin: add support for different encodings
• festival_plugin: add support for different encodings
• Contributors: Martin Pecka, Shingo Kitagawa

0.3.17 (2023-06-08)

• Merge pull request #231 from knorth55/no-wait-mode
• dont wait when rospy.Duration(0) is set for timeout
• Merge pull request #229 from furushchev/flite-plugin-lazy-load FlitePlugin: Lazy loading default voice path
• FlitePlugin: Lazy loading default voice path
• Contributors: Shingo Kitagawa, Yuki Furuta

0.3.16 (2022-12-23)

• Merge pull request #203 from nakane11/timeout
• refactor libsoundplay.py
• Add timeout to wait_for_server and wait_for_result
• Contributors: Aoi Nakane, Shingo Kitagawa

0.3.15 (2022-08-29)

• Merge pull request #200 from knorth55/yaml-missing
• show error and skip loading when plugin yaml is missing
• Merge pull request #199 from knorth55/install-plugin-yaml
• fix missing install in CMakeLists.txt
• Contributors: Shingo Kitagawa

0.3.14 (2022-08-18)

• Merge pull request #193 from knorth55/refactor-soundplay-node
• refactor soundplay_node.py
• Merge pull request #192 from knorth55/fix-file-open-issue
• fix typo causing file open issue
• Merge pull request #191 from knorth55/flite-default-voice-dir
• add default voice dir for flite_plugin
• Merge pull request #190 from ros-drivers/knorth55-patch-1
• Update soundplay_node.py
• Merge pull request #187 from knorth55/fix-typo
• fix typo in soundplay_node.py
• Merge pull request #185 from knorth55/sound-play-flite-plugin add flite plugin for sound_play
• Merge pull request #183 from knorth55/sound-play-plugin add soundplay plugin feature
• add flite in sound_play dependency
• refactor FestivalPlugin
• add flite plugin
• change default_voice to None
• add plugin arg in soundplay_node.launch
• refactor codes
• add output screen in soundplay_node.launch
• add soundplay plugin attribute
• Merge pull request #184 from knorth55/default-voice
• add default_voice in soundplay_node.launch
• Merge pull request #182 from iory/is-speaking
• Improve is_speaking by checking goal status
• Merge pull request #181 from knorth55/refactor-is-speaking
• refactor is_speaking.py
• Contributors: JSK fetch user, Shingo Kitagawa, iory

0.3.13 (2022-04-07)

• Merge pull request #176 from iory/is-speeching
• Add is_speaking.py to catkin_install_python
• Fixed name speeching to speaking
• Add is_speeching node for checking robot is speaking
• Contributors: Shingo Kitagawa, iory

0.3.12 (2021-09-01)

• Merge pull request #175 from iory/rate Modified loop rate for action execution
• Modified loop rate for action execution
• Merge pull request #131 from yann-bourrigault/master Handle playing sound in loop
• import GObject in try section
• Merge pull request #174 from iory/cache Add arg2 information for cache
• Add arg2 information for cache
• Merge pull request #173 from knorth55/replace-sound-client
• Merge pull request #172 from knorth55/start-action-after-init [sound_play] start ActionServer after initialize in soundplay_node.py
• Merge pull request #171 from knorth55/set-aborted [sound_play] add proper set_aborted in soundplay_node.py
• add replace in sendMsg
• start actionserver after initialize in soundplay_node.py
• add proper set_aborted in soundplay_node.py
• Merge branch 'master' into master
• Handle playing sound repeatedly
• Contributors: Shingo Kitagawa, Yann BOURRIGAULT, iory

0.3.11 (2021-04-08)

• Merge pull request #167 from k-okada/fix_155
• Use rospy.myargv() instead of sys.argv to support remapping
• Contributors: Kei Okada, Shingo Kitagawa

0.3.10 (2021-01-07)

0.3.9 (2020-10-22)

0.3.8 (2020-09-13)

• Merge pull request #155 from garaemon/use-myargv Use rospy.myargv() instead of sys.argv to support remapping
• Use rospy.myargv() instead of sys.argv to support remapping
• Merge pull request #154 from mikaelarguedas/fix_say_python3
• update to support no iso-8859-15 language (#1)
  • support non iso-8859-15 language
  • encode only for python2
• convert items to an iterator
• make cleanup compatible with Python 3
• catch AttributeError to handle python3 strings
• Contributors: Mikael Arguedas, Ryohei Ueda, Shingo Kitagawa

0.3.7 (2020-08-08)

• Merge pull request #149 from garaemon/specify-topic-to-play-sound Support use different topic and actionlib to play sound
• Support use different topic and actionlib to play sound * Add two keywords to the constructor of SoundClient class in order to specify actionlib namespace and topic name to play sound.
  • See #119.
• Merge pull request #144 from ros-drivers/knorth55-patch-1
• add gstreamer1.0-alsa exec_depend in sound_play
• Contributors: Ryohei Ueda, Shingo Kitagawa

0.3.6 (2020-05-29)

• Merge pull request #140 from knorth55/support-python3 fix syntax for python3
• Merge pull request #141 from knorth55/add-maintainer add maintainer
• add maintainer
• fix syntax for python3
• Contributors: Shingo Kitagawa

0.3.5 (2020-04-28)

• Merge pull request #133 from knorth55/noetic-build
• remove unnecessary shebang
• use setuptools instead of distutils.core
• use package format=3 for python3
• refactor CMakeLists.txt
• use catkin_install_python for python shebang
• Merge pull request #135 from knorth55/add-travis
• disable sound_play test
• Contributors: Shingo Kitagawa

0.3.4 (2020-04-02)

• Merge pull request #126 from itohdak/fix-Gstreamer-memory-leak [sound_play/scripts/soundplay_node.py] fix Gstreamer memory leak
• Merge pull request #123 from 708yamaguchi/fix-encode Do not encode text when using langages which ISO-8859-15 does not support
• [sound_play/scripts/soundplay_node.py] fix Gstreamer memory leak
• do not encode text when using langages which ISO-8859-15 does not support
• Merge pull request #118 from v4hn/patch-1 use default audio output by default
• use default audio output by default Not specifying a sound device defaults to the first sound device starting from Ubuntu 16.04., not to the one configured as default. The change is backward compatible and tested on ROS indigo and kinetic on a PR2 robot.
• Merge pull request #110 from gerardcanal/master Encoded text to be said in ISO-8859-15
• Merge branch 'master' of github.com:ros-drivers/audio_common
• Sound play: Encoded file to be said in ISO-8859-15 so that accents in languages such as Spanish, Catalan or French are correctly pronounced (based on http://festcat.talp.cat/en/usage.php which says festival expects ISO-8859-15 encoding)
• Contributors: Austin, Gerard Canal, Michael Görner, Naoya Yamaguchi, Shingo Kitagawa, itohdak

0.3.3 (2018-05-22)

• Fix gstreamer errors. Fixes #108
• Contributors: trainman419

0.3.2 (2018-05-02)

• [sound_play] add option to select audio device to play / record (#87)
  • [sound_play] add option to select audio device to play
  • [sound_play] reformat README to markdown; add usage to set device via rosparam
  • audio_capture: add option for selecting device to use
  • audio_play: add option to select device for playing audio

  * add device argument to launch files Conflicts: audio_capture/launch/capture.launch audio_capture/launch/capture_to_file.launch audio_capture/src/audio_capture.cpp audio_play/launch/play.launch sound_play/scripts/soundplay_node.py

• Merge pull request #95 from yujinrobot/volume_check [sound_play] volume check for cached sounds
• [sound_play] checks if sound's Gst instance's volume has changed and resets it
• Contributors: Austin, Naveed Usmani, Yuki Furuta

0.3.1 (2016-08-28)

• Update to new gstreamer rosdeps
• Update sound_play to gstreamer 1.0
• remove chance of uninitialised variable being called in a subscriber callback.
• Add changelogs
• Issue: The error checks for missing publisher/action client in sendMsg were inverted. The non-blocking brach tested the action client while the blocking branch tested the publisher. Fix: Inverted the blocking boolean for both branchs.
• sound_play: Fix build with -DCATKIN_ENABLE_TESTING=OFF. https://bugs.gentoo.org/show_bug.cgi?id=567466
• [soundplay_node] fix resources not being released on dict cleanup This was resulting in the number of sink inputs reaching the maximum threshold, (32 on ubuntu 14.04 with pulseaudio 4.0) after which no more sounds could be played by the node. It would only happen if the rate of sounds being played was slower than the dictionary cleanup.
• depend on actionlib.
• Introduce unit test to ensure soundclient is started correctly.
• Example of using the explicit blocking parameter to override the class setting.
• SoundClient can also explicitly specify whether or not to block while playing the sound. Each play/repeat/say/... method can take an option blocking=True|False argument (using **kwargs), which over-rides the class-wide setting.
• Merge pull request #62 from felixduvallet/set_queue_size Set queue_size in soundplay_node Publisher
• do both in same script.
• Added script showing the various blocking/non-blocking ways of using SoundClient.
• removed trailing whitespace only
• loginfo -> logdebug.
• Slightly more condensed version of thresholding.
• Enable blocking calls inside libsoundplay's SoundClient. This makes use of the actionlib interface provided by soundplay_node, by ensuring SoundClient receives a response before returning. Turn this on by: SoundClient(blocking=true).
• Use new-style python classes (inherits from object).
• removed trailing whitespace.
• Set the volume in each of the sound_play actionlib tests. This makes the script actually play the sounds it requests.
• Specify queue size explicitly. Removed warning message printed each time soundplay_node was started.
• remove trailing whitespace only.
• Change wiki urls
• Fix test target name collision. Fixes #49
• sound_play: cpp header conforms to the style guide
• sound_play: update scripts to allow volume to be set
• sound_play: updated tests to include volume changes
• sound_play: add ability to specify volume at which to play sounds Also changed error to warning as per todo
• sound_play: fix indentation and comment inconsistencies
• sound_play: remove some raw prints cluttering output
• sound_play: added queue_size to SoundClient init Should prevent warning being displayed whenever the client is created. Fixes issue #43
• add simple-actionlib functionality to sound_play
• sound_play: Added functions to play files relative to a package path
• Update maintainer email
• Contributors: Alexis Ballier, Austin, Daniel Stonier, David V. Lu, Felix Duvallet, Matthias Nieuwenhuisen, Michal Staniaszek, Neowizard, aginika, trainman419

0.2.11 (2016-02-16)

• Add changelogs
• Fix bug in say.py. Fixes #72
• Contributors: trainman419

0.2.10 (2016-01-21)

• Add changelogs
• Issue: The error checks for missing publisher/action client in sendMsg were inverted. The non-blocking brach tested the action client while the blocking branch tested the publisher. Fix: Inverted the blocking boolean for both branchs.
• sound_play: Fix build with -DCATKIN_ENABLE_TESTING=OFF. https://bugs.gentoo.org/show_bug.cgi?id=567466
• Contributors: Alexis Ballier, Neowizard, trainman419

0.2.9 (2015-12-02)

• Add changelogs
• [soundplay_node] fix resources not being released on dict cleanup This was resulting in the number of sink inputs reaching the maximum threshold, (32 on ubuntu 14.04 with pulseaudio 4.0) after which no more sounds could be played by the node. It would only happen if the rate of sounds being played was slower than the dictionary cleanup.
• depend on actionlib.
• Introduce unit test to ensure soundclient is started correctly.
• Example of using the explicit blocking parameter to override the class setting.
• SoundClient can also explicitly specify whether or not to block while playing the sound. Each play/repeat/say/... method can take an option blocking=True|False argument (using **kwargs), which over-rides the class-wide setting. Conflicts: sound_play/src/sound_play/libsoundplay.py
• do both in same script.
• Added script showing the various blocking/non-blocking ways of using SoundClient.
• removed trailing whitespace only Conflicts: sound_play/scripts/say.py
• loginfo -> logdebug.
• Enable blocking calls inside libsoundplay's SoundClient. This makes use of the actionlib interface provided by soundplay_node, by ensuring SoundClient receives a response before returning. Turn this on by: SoundClient(blocking=true). Conflicts: sound_play/src/sound_play/libsoundplay.py
• Use new-style python classes (inherits from object). Conflicts: sound_play/src/sound_play/libsoundplay.py
• removed trailing whitespace. Conflicts: sound_play/src/sound_play/libsoundplay.py
• Revert "Set the volume in each of the sound_play actionlib tests." This reverts commit 55ab08c882809fc6d21affb849a7dac9f1901867. Indigo-devel does not have the volume API
• Set the volume in each of the sound_play actionlib tests. This makes the script actually play the sounds it requests.
• Specify queue size explicitly. Removed warning message printed each time soundplay_node was started.
• remove trailing whitespace only.
• Fix wiki links
• Contributors: David V. Lu, Felix Duvallet, Michal Staniaszek, trainman419

0.2.8 (2015-10-02)

• Fix test target name collision. Fixes #49
• sound_play: remove some raw prints cluttering output
• sound_play: added queue_size to SoundClient init Should prevent warning being displayed whenever the client is created. Fixes issue #43
• add simple-actionlib functionality to sound_play
• sound_play: Added functions to play files relative to a package path
• Update maintainer email
• Contributors: Matthias Nieuwenhuisen, Michal Staniaszek, aginika, trainman419

0.2.7 (2014-07-25)

0.2.6 (2014-02-26)

• Fix path resolution in python soundplay lib.
• now importing roslib. closes #33
• Contributors: Piyush Khandelwal, trainman419

0.2.5 (2014-01-23)

• "0.2.5"
• Install sounds. Fixes #29.
• install sound_play.h and export include folder
• Contributors: ahendrix, trainman419, v4hn

0.2.4 (2013-09-10)

• Fix cmake ordering.
• Contributors: Austin Hendrix

0.2.3 (2013-07-15)

• Fix python.
• Contributors: Austin Hendrix

0.2.2 (2013-04-10)

• Actually add proper dependency on message generation.
• Reorder CMakeLists.txt.
• Contributors: Austin Hendrix

0.2.1 (2013-04-08 13:59)

0.2.0 (2013-04-08 13:49)

• Finish catkinizing audio_common.
• Start catkinizing sound_play.
• Fix typo in package.xml
• Versions and more URLs.
• Convert manifests to package.xml
• Ditch old makefiles.
• Use festival default voice from libsoundplay.
• Set myself as the maintainer.
• Fix filehandle leak and add debug statements.
• Updates manifest
• Updated manifests for rodep2
• Fixed sound_play
• Added test wave
• Cleaned up the test script
• Added default voice to say command
• Updated the gstreamer rosdeps
• Removed comment
• Added diagnostic_msgs to sound_play
• Added a rosdep.yaml file
• Added ability to use different festival voices
• Added exit(1) when import of pygame fails. This makes the error message easier to notice.
• Added Ubuntu platform tags to manifest
• Added a link to the troubleshooting wiki page in the diagnostic message as requested by #4070.
• Took out the deprecated API.
• Sound play now publishes header timestamp in message. #3822
• Cleaned up temp file generation when doing text to speach. Now uses the tempfile module.
• Adding missing export of headers for sound_play C++ API
• Changing node name for sound play diagnostics, #3599
• Added test.launch to run sound server and a test client.
• Remove use of deprecated rosbuild macros
• Replaced review tag with standardized message
• Updated review status
• Added a launch file to start soundplay_node.py
• Made the sound_play client libraries be more explicit about what to do when the node is not running.
• Updated manifest description
• Updated copyright year
• fixed XML typo
• updated package description
• Added a copyright message.
• Removed debugging message from sound_play node.
• Added tests for new sound_play python API and fixed a few bugs.
• Fixed missing self arguments in sound_play libsoundplay.py
• Upgraded the python sound_play API
• Converted non-camelCase methods to camelCase in sound_play C++ API
• Changed Lock to RLock to fix #2801
• Made the deprecation of SoundHandle into a warning.
• Added debug messages
• Updated soundplay_node to publish diagnostics and increased the number of active channels.
• Added diagnostic_msgs dependency to sound_play
• sound_play: Renamed SoundHandle to SoundClient. Added Sound-centric C++ API. Changed byte to int8 in msg file. Updated documentation.
• migration part 1
• Contributors: Austin Hendrix, Nate Koenig, blaise, blaisegassend, eitan, gerkey, kwc, nkoenig, watts, wheeler