Thanks for the report.

Your repro case seems wrong.  Your graph, as you say, looks like sound -> panner -> gain -> destination.

But you set the gain of the gainNode to 0.

Please provide another test case.  Or correct the following, which was based on your repro case:

let context = new AudioContext();
let panner = new PannerNode(context, {
  maxDistance: 1,
  panningModel: "HRTF",
  rolloffFactor: 1,
  coneInnerAngle: 360,
  coneOuterAngle: 0,
  coneOuterGain: 0
});

panner.setPosition(0,0,0);
panner.refDistance = 6697.47512;

let gain = new GainNode(context);
let s = new OscillatorNode(context);

s.connect(gain).connect(panner).connect(context.destination);
context.listener.setPosition( 17763, 25650, 11887 );

I hear a sound.

I made a hopefully reproducable jsbin of it: http://jsbin.com/sezena/edit?html,js,output

at maxDistance = 1 I get a glitch then no sound
at maxDistance = 10000 I get the sound

Thank you for providing more feedback. Adding requester "rtoy@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Thanks for the repro case. I don't get the same results as you, but there's something definitely wrong.  With maxDistance = 1, I hear audio, but it's totally distorted.

Ok, there are a couple of issues here. First, the actual distance was clamped between 0 and maxDistance. That, as you say, is incorrect for the inverse distance model.  And then, since maxDistance = 1, the inverse distance formula did not compute max(distance, refDistance), so the resulting gain was huge (6697.48).  This causes the distortion.

Great, thanks for the progress report.

If I am to overcome this in current build, does just always making sure maxDistance > refDistance fixes it ?
like always setting "maxDistance = refDistance * 1e5;" each time I change refDistance ?

Yes, making sure maxDistance > refDistance is enough. I don't know your actual use case, so I can't say if maxDistance = 1e5 * refDistance is right.

However, you said you change refDistance.  I'm curious what the use case for that is.  Could you explain why you need to change refDistance? (It sounds like you change it a lot, instead of just once and keeping it fixed.)

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/c8faec73855c122f94f00c54549ee6542da59e9b

commit c8faec73855c122f94f00c54549ee6542da59e9b
Author: rtoy <rtoy@chromium.org>
Date: Sat May 06 00:14:41 2017

Correctly clamp in distance formulas

The clamping of the distance formulas was incorrect.  Implement
clamping according to the spec.

BUG= 718069 
TEST=panner-distance-clamping.html

Review-Url: https://codereview.chromium.org/2859143004
Cr-Commit-Position: refs/heads/master@{#469824}

[modify] https://crrev.com/c8faec73855c122f94f00c54549ee6542da59e9b/third_party/WebKit/LayoutTests/webaudio/Panner/panner-distance-clamping.html
[modify] https://crrev.com/c8faec73855c122f94f00c54549ee6542da59e9b/third_party/WebKit/LayoutTests/webaudio/resources/panner-formulas.js
[modify] https://crrev.com/c8faec73855c122f94f00c54549ee6542da59e9b/third_party/WebKit/Source/platform/audio/Distance.cpp

Thanks a lot for the fix!

RefDistance is for 3D sound positioning/range editing in an "editor".
not changing in the "viewer" ;)