Audio Working Group Teleconference

01 Aug 2013


See also: IRC log


chrislowis, olivier, shepazu, gmandyam, jernoble, ehsan, jussi
Chris, Wilson
Chris Lowis


<trackbot> Date: 01 August 2013

<scribe> Scribe: Olivier

<scribe> ScribeNick: olivier

<jernoble> Zakim: aaaa is me

re- agenda, the issues I was suggesting we tackle were actually https://www.w3.org/Bugs/Public/show_bug.cgi?id=17339 https://www.w3.org/Bugs/Public/show_bug.cgi?id=17366 and https://www.w3.org/Bugs/Public/show_bug.cgi?id=17378

<chrislowis> shepazu: thanks.

chrislowis: convening

Review of action items - https://www.w3.org/2011/audio/track/actions/open


chrislowis: just want to review actions

<gmandyam> +q

olivier: moving to close ACTION-59

gmandyam: agree

Close ACTION-59

<trackbot> Closed ACTION-59 Talk to webrtc chairs, see what their perspective is on web audio API.

chrislowis: wondering about ACTION-47

shepazu: there was a reply off the record
... OpenAL was done in spirit of open source and implementations distributed in LGPL
... most people involved at the time are now gone

chrislowis: do you think we need to take that any further?

shepazu: we'd need to refine exactly what we are looking for from them
... maybe have a discussion about this at some point

<scribe> ACTION: olivier to create a bugzilla issue to look into how web audio relates to openAL, whether we need to dig deeper re- IPR [recorded in http://www.w3.org/2013/08/01-audio-minutes.html#action01]

<trackbot> Created ACTION-65 - Create a bugzilla issue to look into how web audio relates to openAL, whether we need to dig deeper re- IPR [on Olivier Thereaux - due 2013-08-08].

Close ACTION-47

<trackbot> Closed ACTION-47 Reach out to OpenAL folks to get firm commitment on IP status of OpenAL..

chrislowis: looking at a group of action items from May related to OfflineAudioContext

ehsan: there is general agreement on what we want the API to look like

chrislowis: I think we can close these and track any detail on bugzilla

Close ACTION-61

<trackbot> Closed ACTION-61 Ask Ehsan to create a patch to the spec for the changes..

Close ACTION-62

<trackbot> Closed ACTION-62 Propose idea for OfflineAudioContext to list.

Close ACTION-63

<trackbot> Closed ACTION-63 Will reply to Chris's proposed changes to OfflineAudioContext with updated WebIDL..

ACTION-64 is still pending

Web audio API - Issue to close - https://www.w3.org/Bugs/Public/show_bug.cgi?id=17374

olivier: explains https://www.w3.org/Bugs/Public/show_bug.cgi?id=17374
... will apply the patch to remove these sections if there is no objection

ehsan: fine with that

<padenot> Zakim: ??P22 is me

chrislowis: will double-check with editor and go ahead

Web audio API - Issue to close - https://www.w3.org/Bugs/Public/show_bug.cgi?id=20681

<chrislowis> padenot: I muted you, your line was very noisy.

<chrislowis> padenot: Add yourself to the queue if you'd like to speak.

<padenot> chrislowis: thanks, I have a sketchy setup here

<chrislowis> padenot: no problem.

olivier: wondering if we need more work on units

ehsan: would rather have crogers in this conversation

<gmandyam> +q

olivier: crogers on the record as saying he's ok with openal-inspired undefined unit

gmandyam: don't think openAL is the best reference here
... I think we defined units for a lot of other web apis
... will write comment on irc too

chrislowis: want to add comment on bugzilla too

Web Audio API - Open Issues and bugs

<gmandyam> OpenAL is not a good model for deciding when to specify unit measures. It is a native API, and developers many times make the assumption that there is variability in implementations (particularly for handheld devices).

<gmandyam> (cont.) Web developers will expect some level of consistency among implementations.




chrislowis: let's look at https://www.w3.org/Bugs/Public/show_bug.cgi?id=17339 - De-zippering is not defined
... spec doesn't go far enough defining dezippering
... ehsan, any comment on how you've interpreted that in implementation

ehsan: spec mentions smoothing for gain nodes, [scribe missed]
... would have like to exactly how to implement this for each type of parameter
... I believe that we don't do any dezippering in gain node
... it wasn't clear what we should be doing
... for delaynode, I think I looked at webkit source, ended up using similar value but think it should be in the spec

<scribe> ACTION: chrislowis to look into what dezippering in webkit was based on [recorded in http://www.w3.org/2013/08/01-audio-minutes.html#action02]

<trackbot> Created ACTION-66 - Look into what dezippering in webkit was based on [on Chris Lowis - due 2013-08-08].

chrislowis: I will look at whether there is an accepted precedent or reference literature we can point to
... don't think this was added empirically for webkit, I'll try and find out what that is
... looking at https://www.w3.org/Bugs/Public/show_bug.cgi?id=17366 - Oscillator types are not defined
... I don't think the definition is controversial
... and phase starts at 0 at time 0

ehsan: I may be lacking necessary background
... finding the definition hard to understand
... don't think the definitions are very controversial
... but there could be different interpretations for phase

chrislowis: we could say in spec that the implementation should support this kind of wave forms
... but how to generate them isn't really a solved problem
... I know the way webkit does is quite standard and complicated in how it avoids aliasing
... should we define mathematica ideal?

ehsan: not sure where we should draw the line
... would prefer a spec that tries to define exactly the outcome
... that's my general preference but I understand this may not be ideal for audio
... oscillatornode is the most underspecified node in the spec
... it has made it difficult to implement

<gmandyam> +q

olivier: will make it a high priority in bugzilla... don't have a perfect solution either

gmandyam: in practice there will be cases where waveform diverges from mathematical definition
... should we add some language such as "ideally"
... so that developers don't expect a perfect level of precision

chrislowis: reminds me of discussion about biquadfilternode, we found this reference doc which would be a possible reference guide for implementors

shepazu: want to share experience with SVG
... very similar situation with graphics experts defining the spec and use conventions from the industry
... second wave of implementors were not experts in graphics and so we had very similar pain points with things left undefined
... made it hard to create interoperable implementations
... web developers were very unsatisfied with variations in implementations
... and I suspect that they would find it equally unsatisfied with audio rendering
... we should keep in mind what expectations are
... so I think more definitions could be helpful to match developer expectations

chrislowis: https://www.w3.org/Bugs/Public/show_bug.cgi?id=17378 - AudioBufferSourceNode.playbackRate not strictly defined

ehsan: right now the way implementation work is that negative playback rate are interpreted as playing backward
... there are a number of details of how resampling is achieved

chrislowis: is this an urgent or important one to solve?

ehsan: there is at least one demo relying on it
... this is definitely a concept that has a correspondence in the real world

chrislowis: the question of how we handle resampling will come up again

<Zakim> olivier, you wanted to suggest a patch at least defining effect of negative values

<ehsan> olivier: we seem to have lost you

<ehsan> chrislowis: olivier: are you on the call?

<chrislowis> ehsan: seems like it's at your end?

<ehsan> could be

<ehsan> let me redial in

chrislowis: will put a call out on the ML to ask for a volunteer to suggest patch for 17378

Any other business

chrislowis: Any Other Business?™

Next meeting

olivier: no objection for next meeting in 2 weeks?

[no objection]

chrislowis: meeting adjourned

Summary of Action Items

[NEW] ACTION: chrislowis to look into what dezippering in webkit was based on [recorded in http://www.w3.org/2013/08/01-audio-minutes.html#action02]
[NEW] ACTION: olivier to create a bugzilla issue to look into how web audio relates to openAL, whether we need to dig deeper re- IPR [recorded in http://www.w3.org/2013/08/01-audio-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2013/08/01 16:59:03 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.138  of Date: 2013-04-25 13:59:11  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Found Scribe: Olivier
Inferring ScribeNick: olivier
Found ScribeNick: olivier
Default Present: +1.408.772.aaaa, jernoble, gmandyam\, chrislowis, olivier, ehsan, Doug_Schepers, jussi, [Mozilla]
Present: chrislowis olivier shepazu gmandyam jernoble ehsan jussi
Regrets: Chris Wilson
Agenda: http://lists.w3.org/Archives/Public/public-audio/2013JulSep/0365.html
Found Date: 01 Aug 2013
Guessing minutes URL: http://www.w3.org/2013/08/01-audio-minutes.html
People with action items: chrislowis olivier

[End of scribe.perl diagnostic output]