CompressedAudio

A single chunk of a compressed audio bitstream

Panel support

CompressedAudio is used in the Audio panel.

Schema

field	type	description
timestamp	Timestamp	Timestamp of the start of the audio chunk
data	bytes	Compressed audio data. Packet duration is determined by the codec during encoding. Messages should generally contain approximately 20 ms of audio.
format	string	Format of the audio data. See the data field description for supported values. Consumers may support a subset of these formats.

data

For each format value, the data field contains compressed audio data serialized as follows:

• opus:
  • Each message must contain a complete raw Opus packet, without Ogg, WebM, or other container framing, as described in RFC 6716 section 3.
  • Each packet contains all information necessary for decoding, and may be decoded at any sample rate supported by Opus (8, 12, 16, 24, or 48 kHz).
  • A single raw Opus packet represents mono or stereo audio; multichannel Opus requires multistream or container metadata and is not supported by this schema.
• mp4a.40.2:
  • Each message must contain a complete MPEG-4 AAC-LC ADTS frame, including the ADTS header, as described in section 1.A.3.2 of ISO/IEC 14496-3:2019.
  • The ADTS header supplies stream parameters such as sample rate and channel configuration.
• g726-16, g726-24, g726-32, and g726-40:
  • Each message must contain raw G.726 codewords without WAV, RTP, or other container framing. Codewords may not span messages.
  • Audio is mono at 8 kHz. The suffix specifies the bitrate in kbps and therefore the number of bits per sample: 2, 3, 4, or 5 respectively.
  • Codewords are packed beginning with the least-significant bit, as described in RFC 3551 section 4.5.4.
  • Codewords may cross octet boundaries. To keep every codeword within one message, the message length must be a multiple of 1, 3, 1, or 5 octets for g726-16, g726-24, g726-32, and g726-40 respectively, holding 4, 8, 2, or 8 codewords.
  • Messages form a continuous encoded stream and preserve G.726 predictor and quantizer state across message boundaries.
• aal2-g726-16, aal2-g726-24, aal2-g726-32, and aal2-g726-40:
  • These formats have the same requirements as the corresponding g726-* formats, except codewords use the opposite bit-packing order specified for AAL2.

Reference implementations

Foxglove messages are framework-agnostic, and can be implemented using any supported message encoding:

encoding	schema
ROS 1	foxglove_msgs/CompressedAudio
ROS 2	foxglove_msgs/msg/CompressedAudio
JSON	foxglove.CompressedAudio
Protobuf	foxglove.CompressedAudio
FlatBuffers	foxglove.CompressedAudio
OMG IDL	foxglove::CompressedAudio

You must use the schema names specified above for Foxglove to recognize the schema.