Skip to content

audio_drivers.md

Latest commit

 

History

History
63 lines (42 loc) · 2.78 KB

audio_drivers.md

File metadata and controls

63 lines (42 loc) · 2.78 KB
Raw

Audio drivers

This page documents the audio input/output plugins in fp_perception_driver_audio.

Concepts

  • Audio data is represented as fp_perception::audio_data (used inside drivers) and fp_perception_msgs/msg/PerceptionAudio (used on ROS topics/services).
  • Device acquisition uses PortAudio.
  • Plugins are loaded by the perception server using pluginlib and used through typed audio interfaces.

MicrophoneAudioDriver

Class: fp_perception::MicrophoneAudioDriver (PortAudio input)

Device acquisition

  • Device selection uses driver.audio.MicrophoneAudioDriver.device_name first.
  • driver.audio.MicrophoneAudioDriver.device_id is an optional fallback PortAudio device index; leave it unset for name-based or default-device selection.
  • The driver opens and starts a PortAudio input stream.
  • A PortAudio callback captures audio into an internal buffer; readChunk() waits until at least one chunk is available.

Parameters

  • driver.audio.MicrophoneAudioDriver.name (string)
  • driver.audio.MicrophoneAudioDriver.device_name (string)
  • driver.audio.MicrophoneAudioDriver.device_id (int, optional fallback)
  • driver.audio.MicrophoneAudioDriver.chunk_size (int)
  • driver.audio.MicrophoneAudioDriver.sample_rate (int)
  • driver.audio.MicrophoneAudioDriver.channels (int)
  • driver.audio.MicrophoneAudioDriver.capture_buffer_window (int, seconds)

Typical usage

  • The perception server calls readChunk() repeatedly to build a continuous audio feed.
  • If transcription is enabled, the server also keeps an internal rolling buffer for a configurable duration.
  • If use_diagnostics=true, the driver publishes capture health on /diagnostics via diagnostic_updater.

SpeakerAudioDriver

Class: fp_perception::SpeakerAudioDriver (PortAudio output)

Device acquisition

  • Device selection uses driver.audio.SpeakerAudioDriver.device_name first.
  • driver.audio.SpeakerAudioDriver.device_id is an optional fallback PortAudio device index; leave it unset for name-based or default-device selection.
  • The driver opens output streams on-demand keyed by sample format/rate/channels.

Parameters

  • driver.audio.SpeakerAudioDriver.name (string)
  • driver.audio.SpeakerAudioDriver.device_name (string)
  • driver.audio.SpeakerAudioDriver.device_id (int, optional fallback)
  • driver.audio.SpeakerAudioDriver.sample_rate (int)
  • driver.audio.SpeakerAudioDriver.channels (int)
  • driver.audio.SpeakerAudioDriver.test_file_path (string)

Typical usage

  • The perception server can subscribe to an audio topic and forward it to this driver.
  • The perception server can also route synthesized speech audio to this driver when speech requests set use_device_audio=true.
  • If use_diagnostics=true, the driver publishes playback health on /diagnostics via diagnostic_updater.