This element is responsible for parsing audio in RawAudio format. The Parser ensures that output buffers have whole samples.

By default, the parser doesn't ensure that each output buffer holds the same number of samples, it only re-aligns buffers to whole samples. When chunk_duration is set, the parser additionally re-chunks the stream, so that every output buffer carries exactly chunk_duration worth of audio.

Element options

Passed via struct Membrane.RawAudioParser.t/0

• assumed_input_stream_format

  RawAudio.t() | nil

  Default value: nil
  The format of the stream to assume if the stream format received on the input pad is Membrane.RemoteStream.t/0. When nil, the format received on the input pad is used. When both formats are instances of Membrane.RawAudio.t/0, the parser expects them to be identical.

• overwrite_pts?

  boolean()

  Default value: false
  If set to true, RawAudioParser will add timestamps based on payload duration. The first buffer's timestamp value can be specified with the pts_offset option and defaults to 0.

• pts_offset

  non_neg_integer()

  Default value: 0
  If set to a value different than 0, RawAudioParser will start timestamps from this offset. Only valid when :overwrite_pts? is set to true.

• chunk_duration

  Membrane.Time.t() | nil

  Default value: nil
  When set, output buffers are re-chunked so that each one carries exactly chunk_duration worth of audio. Bytes that don't fill a whole chunk are buffered until enough data arrives. The trailing remainder is flushed as a (possibly shorter) buffer at end of stream.

  When nil (the default), the parser only re-aligns buffers to whole samples and otherwise passes them through unchanged.

Pads

:input

Accepted formats:

Membrane.RawAudio

Membrane.RemoteStream

:output

Accepted formats:

Membrane.RawAudio