Audio Normalization

Audio normalization adjusts the loudness of a program to a specific target level. Audio normalization requires a loudness measurement pass followed by a loudness correction pass.

Audio normalization is applied to an output using the following properties:

Property	Type	Description
`MeteringMode`	enum	Specifies the loudness measurement algorithm: `bt1770_1` ungated loudness, `bt1770_2` gated loudness, `bt1770_3` improved true peak, `bt1770_4` higher channel count.
`DialogIntelligence`	boolean	Indicates whether dialogue intelligence is enabled. Default is `false`.
`SpeechThreshold`	number	Specifies the speech content threshold as a percentage. If the percentage of speech detected in the program exceeds this threshold the speech gated loudness measurment is used for normalization.
`NormalizedLoudness`	number	Specifies the normalized target level in `LUFS`. Default is -23.0 `LUFS`.
`RelativeGateOffset`	number	Specifies the loudness gate to be used as silence. The default value is 0.

For example the following composition normalizes the output audio level to -18.0 LUFS using the BT 1770-2 metering mode. Note that each audio channel must have a speaker label:

---
input:
  container:
  - name: wav
    audio:
    - name: a1
      channels: 2
      label:
      - L
      - R
output:
  container:
  - name: aac
    type: 'mp4 '
    audio:
    - route:
        name: a1
      properties:
        Format: aac
        AudioNormalize:
          MeteringMode: bt1770_2
          DialogInitelligence: true
          SpeechThreshold: 25
          NormalizedLoudness: -18.0
          RelativeGateOffset: -10
      encode:
        properties:
          BitRate: 96000
          MpegVersion: 1
          Profile: lc

{
  "input": {
    "container": [
      {
        "name": "wav",
        "audio": [
          {
            "name": "a1",
            "channels": 2,
            "label": [
              "L",
              "R"
            ]
          }
        ]
      }
    ]
  },
  "output": {
    "container": [
      {
        "name": "aac",
        "type": "mp4 ",
        "audio": [
          {
            "route": {
              "name": "a1"
            },
            "properties": {
              "Format": "aac",
              "AudioNormalize": {
                "MeteringMode": "bt1770_2",
                "DialogInitelligence": true,
                "SpeechThreshold": 25,
                "NormalizedLoudness": -18.0,
                "RelativeGateOffset": -10
              }
            },
            "encode": {
              "properties": {
                "BitRate": 96000,
                "MpegVersion": 1,
                "Profile": "lc"
              }
            }
          }
        ]
      }
    ]
  }
}

Metering Modes

BT.1770-1 measures the average or integrated loudness over the entire duration of the program. The measurement is expressed in LU (loudness units) and loudness units referenced to full scale LUFS.

BT.1770-2 introduced a level based gating function to prevent long periods of silence from disproportionately influencing the integrated loudness measurement.

BT.1770-3 introduced a more accurate method to measure the true peak sample level of the program. The peak sample level prevents subsequent audio normalization from introducing distortion due to peak clipping.

BT.1770-4 includes additional surround channels.

Dialogue Intelligence

Dolby Dialogue Intelligence is an alternative speech based gating function used to measure the integrated loudness of a program.

By default the BT.1770-2 level gated measurement represents the integrated program loudness. When dialog intelligence is enabled and the percentage of speech content exceeds a specific threshold the speech gated measurement is selected as the integrated loudness.