Audio channel mixing and remixing

AWS Elemental MediaLive provides a two-stage audio processing pipeline that you can use to combine audio from multiple sources and remap channels for your output. You can pre-mix multiple audio PIDs or tracks into a single combined stream, and then remix that stream into your desired output channel layout.

About audio mixing in MediaLive

MediaLive processes audio mixing in two stages. You can use either stage independently or combine both stages in the same channel.

Stage 1: Pre-mixing

When your input contains multiple audio PIDs (in a transport stream) or multiple tracks, you can select multiple PIDs or tracks in a single audio selector. MediaLive independently processes each one—you can remix channels, adjust gain, and normalize loudness—before MediaLive interleaves the results into a combined audio stream.

Stage 2: Output remixing

After pre-mixing (or for single-PID/track inputs), you can remap channels in the audio description by using RemixSettings. This produces the final channel layout, for example downmixing 5.1 surround to stereo.

Note

Pre-mixing is optional. You can use output remixing alone if you only need to remap channels from a single audio source.

Pre-mixing audio from multiple PIDs

To pre-mix audio from multiple PIDs in a transport stream input, use AudioPidSelection with the Pids array. This replaces the legacy single Pid field.

Each entry in the Pids array is an AudioPid object with the following fields:

• Pid (integer, 32–8190) — The PID value.

• PremixSettings (optional) — An AudioPreMixerSettings object that controls how MediaLive processes this PID before interleaving.

The AudioPreMixerSettings object contains the following fields:

• Channels (integer, 1–16) — Target channel count for this PID. If you specify this field, MediaLive remixes the audio to this channel count by using a default matrix.

• RemixSettings — Fine-grained channel mapping. This is mutually exclusive with Channels—you cannot specify both. Uses the same structure as the output-level RemixSettings.

• GainDb (number, -60 to +60) — Gain adjustment in dB applied to this PID's audio.

• AudioNormalizationSettings — Loudness normalization settings (algorithm, target LKFS, and so on).

The following constraints apply:

• You must specify at least 2 PIDs when using PremixSettings.

• You can specify a maximum of 16 PIDs per selector.

• The total output channels across all PIDs with premix settings cannot exceed 16.

The following example shows two PIDs being pre-mixed. PID 101 contains stereo audio remixed to mono (1 channel) with -3 dB gain. PID 102 contains stereo audio kept as stereo (2 channels). The result is a 3-channel interleaved output.

{
  "AudioSelectors": {
    "my_mixed_audio": {
      "SelectorSettings": {
        "AudioPidSelection": {
          "Pids": [
            {
              "Pid": 101,
              "PremixSettings": {
                "Channels": 1,
                "GainDb": -3
              }
            },
            {
              "Pid": 102,
              "PremixSettings": {
                "Channels": 2
              }
            }
          ]
        }
      }
    }
  }
}

Pre-mixing audio from multiple tracks

To pre-mix audio from multiple tracks, use AudioTrackSelection with the Tracks array. This works the same way as PID-based pre-mixing but uses track numbers instead of PID values.

Each entry in the Tracks array is an AudioTrack object with the following fields:

• Track (integer, 1-based) — The track number.

• PremixSettings (optional) — An AudioPreMixerSettings object. This uses the same structure and fields as described in the preceding PID section.

The same constraints apply: you must specify at least 2 tracks when using PremixSettings, and the total output channels cannot exceed 16.

The following example shows two tracks being pre-mixed. Track 1 contains stereo audio remixed to mono. Track 2 contains stereo audio kept as stereo.

{
  "AudioSelectors": {
    "my_mixed_tracks": {
      "SelectorSettings": {
        "AudioTrackSelection": {
          "Tracks": [
            {
              "Track": 1,
              "PremixSettings": {
                "Channels": 1,
                "GainDb": -3
              }
            },
            {
              "Track": 2,
              "PremixSettings": {
                "Channels": 2
              }
            }
          ]
        }
      }
    }
  }
}

Remixing channels in the output

You can configure RemixSettings at the audio description level (output encode) to remap channels from the input stream into your desired output channel layout.

The RemixSettings object contains the following fields:

• ChannelsIn (integer, 1–16) — Number of input channels to map from.

• ChannelsOut (integer, 1–8) — Number of output channels to produce.

• ChannelMappings (array) — One entry per output channel. Each entry contains:

  • OutputChannel (integer, 0-based index) — The output channel index.

  • InputChannelLevels (array) — Contribution from each input channel. Each entry contains:

    • InputChannel (integer, 0-based index) — The input channel index.

    • Gain (integer, -60 to +6) — Gain in dB for this input channel's contribution.

The following example shows a 5.1 surround to stereo downmix. Input channels 0–5 represent the standard 5.1 layout (L, R, C, LFE, LS, RS). The output maps left-facing channels to output channel 0 and right-facing channels to output channel 1.

{
  "RemixSettings": {
    "ChannelsIn": 6,
    "ChannelsOut": 2,
    "ChannelMappings": [
      {
        "OutputChannel": 0,
        "InputChannelLevels": [
          {"InputChannel": 0, "Gain": 0},
          {"InputChannel": 2, "Gain": -3},
          {"InputChannel": 4, "Gain": -6}
        ]
      },
      {
        "OutputChannel": 1,
        "InputChannelLevels": [
          {"InputChannel": 1, "Gain": 0},
          {"InputChannel": 2, "Gain": -3},
          {"InputChannel": 5, "Gain": -6}
        ]
      }
    ]
  }
}

Note

A gain value of -60 effectively mutes that input channel's contribution to the output.

Combining pre-mixing and output remixing

You can use both stages together. First, pre-mix multiple PIDs or tracks into a combined stream. Then, apply RemixSettings on the audio description to produce the final output layout.

For example, consider two stereo PIDs that you pre-mix without channel reduction. MediaLive interleaves them into a 4-channel combined stream. You then configure RemixSettings on the audio description with ChannelsIn set to 4 and ChannelsOut set to 2. This produces a stereo output that blends content from both original PIDs.