Re: [mediacapture-main] Specification to capture unmodified audio

Yes, indeed, the echoCancellation property was the only preprocessing property so far, but meanwhile the specification has changed. Now there are the additionally properties autoGainControl and noiseSuppression.
It would be helpful if an application could rely on unmodified audio data if it is able to set all of those properties to false. As far as I understand the current spec, this should be possible if the MediaTrackCapabilities object returned by the user agent contains a false value in each of the preprocessing property sequence values (e.g.  sequence<boolean>   autoGainControl).

But generally I would prefer to get unmodified audio as default. The  requirements for preprocessing might change in the future and additional processing tasks might be added (beam forming with microphon arrays for instance). If those new tasks would be enabled by default it would change the audio quality of existing applications.

I'm currently using this constraints to get unmodified audio from Chrome:

```javascript
{
 audio: {
   "mandatory": {
      "echoCancellation": false,
      "googEchoCancellation": "false",
      "googAutoGainControl": "false",
      "googNoiseSuppression": "false",
      "googHighpassFilter": "false"
     }
    },
  video: false
 }
```

The current spec contains all preprocessing properties of Chrome except the highpass filter.
You can take this as an example: If Chrome enables the high pass filter by default, an application would unintentionally capture filtered (and therefore altered)  audio data, even if it sets all preprocessing properties of the spec to false. 

-- 
GitHub Notification of comment by klausj
Please view or discuss this issue at https://github.com/w3c/mediacapture-main/issues/457#issuecomment-311315018 using your GitHub account

Received on Tuesday, 27 June 2017 10:12:38 UTC