W3C home > Mailing lists > Public > public-audio@w3.org > April to June 2012

Re: Suggestion for minimizing audio glitches

From: Chris Rogers <crogers@google.com>
Date: Tue, 17 Apr 2012 11:29:26 -0700
Message-ID: <CA+EzO0=H+KqX5zAZnffHacOcqPze5hD6e1ppv31qqcF-8voU1A@mail.gmail.com>
To: Randell Jesup <randell-ietf@jesup.org>
Cc: public-audio@w3.org
On Tue, Apr 17, 2012 at 11:27 AM, Randell Jesup <randell-ietf@jesup.org>wrote:

> On 4/17/2012 1:44 PM, Alistair MacDonald wrote:
>
>> OK, that's great.
>>
>> Assuming there is be a WebRTC event for when a stream looses
>> connectivity, I guess you would just setValueCurveAtTime() to the gain.
>> Then ramp up quickly when the stream re-connects.
>>
>
> Generally there is no notification; WebRTC runs over UDP.  There are a few
> cases where we do know we've lost connectivity temporarily (IP change,
> etc), plus call-end.
>
> WebRTC's internal audio processing will generally conceal lost audio
> packets (which would likely include IP change), so really the only thing
> left would be end-of-call or close-tab.
>

Sounds good to me.



>
>  Is that along the lines of what you were thinking Chris?
>>
>>
>>
>>
>> On Tue, Apr 17, 2012 at 1:01 PM, Chris Rogers <crogers@google.com
>> <mailto:crogers@google.com>> wrote:
>>
>>
>>
>>    On Tue, Apr 17, 2012 at 9:05 AM, Alistair MacDonald <al@signedon.com
>>    <mailto:al@signedon.com>> wrote:
>>
>>        Randell's 1-5 suggestions are very interesting.
>>
>>        I would think putting this behavior on the destination node
>>        might be odd. But I wonder if adding this kind of behavior to a
>>        something like the gain node might be useful?
>>
>>        For example: if I wanted to combine Video-Chat with a DAW (UC-1
>>        & UC3), then the following issues would be in play...
>>
>>        1) If a VOIP stream stops suddenly, the user might think there
>>        was a pop/click in their audio track. Adding a tail/decay would
>>        be a solution. (Randell's Option 3)
>>        2) Being a DAW, we would need as much CPU as possible. So
>>        avoiding the tail calculation in JavaScript would be ideal.
>>
>>
>>    We should be able to do this today with a fade-out using an
>>    AudioGainNode.
>>
>
>
> --
> Randell Jesup
> randell-ietf@jesup.org
>
>
Received on Tuesday, 17 April 2012 18:30:00 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 17 April 2012 18:30:00 GMT