Adjusts to KHR_audio PR recommendations to split audio out

[antpb]

This PR adjusts the schema structure to account for feedback provided in KHR_Audio proposal:

KhronosGroup/glTF#2137 (comment)

{
  "emitters": [
    {
      "name": "Positional Emitter",
      "type": "positional",
      "gain": 0.8,
      "inputs": [0, 1],
      "positional": {
        "coneInnerAngle": 6.283185307179586,
        "coneOuterAngle": 6.283185307179586,
        "coneOuterGain": 0.0,
        "distanceModel": "inverse",
        "maxDistance": 10.0,
        "refDistance": 1.0,
        "rolloffFactor": 0.8
      }
    }
  ],
  "sources": [
    {
      "name": "Clip 1",
      "gain": 0.6,
      "playing": true,
      "loop": true,
      "audio": 0
    },
    {
      "name": "Clip 2",
      "gain": 0.6,
      "playing": true,
      "loop": true,
      "audio": 1
    }
  ],
  "audio": [
    {
      "uri": "audio1.mp3"
    },
    {
      "bufferView": 0,
      "mimeType": "audio/mpeg"
    }
  ]
}

[antpb]

in the PR Audio was structured as

  "audio": [
    {
      "uri": "audio1.mp3"
    },
    {
      "bufferView": 0,
      "mimeType": "audio/mpeg"
    }
  ]

Are those objects supposed to be split out? In the readme I mocked it like

      "audio": [
        {
          "uri": "audio1.mp3",
          "bufferView": 0,
          "mimeType": "audio/mpeg"
        }
      ]

Once this is clear, I can update the PR and get the sample asset matching.

[robertlong]

in the PR Audio was structured as

  "audio": [
    {
      "uri": "audio1.mp3"
    },
    {
      "bufferView": 0,
      "mimeType": "audio/mpeg"
    }
  ]

Are those objects supposed to be split out? In the readme I mocked it like

      "audio": [
        {
          "uri": "audio1.mp3",
          "bufferView": 0,
          "mimeType": "audio/mpeg"
        }
      ]

Once this is clear, I can update the PR and get the sample asset matching.

You can use the glTF.image schema as a reference for audio, it's almost the same.

[robertlong]

Also one more change we probably want to make is to change input to sources. This is closer in naming to the audioSources and makes it a bit more obvious.

[robertlong]

Also I think we wanted to change playing back to autoPlay

[oveddan]

Awesome work and I can see myself using this in my project!

Some questions:

• With this extension, how would you specify the actual position of the audio? Is it dependent on the position of the parent element? Or is there a position and rotation attribute?

Suggested future improvements:

• Being able to filter audio channels in each emitter, i.e. only play a single audio channel
• A flag indicating that playback position should be synchronized across user sessions.

[fire]

@robertlong Since you are an implementer of KHR_audio.

We discussed we can choose different horizons of updates. There were opinions of a few days versus stabilization. Wanted to poll.

@antpb requested review the structure of the KHR_audio specification.

[fire]

We didn't want to break existing implementations because of the audio clips structure, but we didn't note any other changes.

Want to try a new strategy where we gather opinion via approvals and request changes to disconnect the approval process from merging.

Since this was discussed in the gltf meeting and there was consensus that it was approved. I wanted to leave the window open and then merge on Friday noon tomorrow PDT time.

[antpb]

three omi to be updated after this is merged