See also: IRC log
<trackbot> Date: 19 September 2013
<padenot> yes
trackbot, start meeting
<trackbot> Meeting: Audio Working Group Teleconference
<trackbot> Date: 19 September 2013
olivier: fix was trivial
cwilso: hope it was ok I didn't use a fork/branch
<cwilso> Normally, I would use a branch.
<cwilso> (for substantive changes)
https://github.com/WebAudio/web-audio-api/issues/127#issuecomment-24507931
padenot: discussing whether to
filter high frequency
... can be useful not to filter in some cases
cwilso: it may be difficult to tell whether it's a use case where you wouldn't filter
padenot: when you use an
LFO?
... below 20Hz you know you are not going to use it to output
sound
cwilso: we don't have a switch to
say whether to use a mathematical square wave
... is there a strong reason for doing that?
joe: "that" means always or never filter?
cwilso: the suggested change is to not require (MUST) to filter
joe: think that's OK if we are
not saying when filtering is optional or required
... let developers indicate explicitely what they want - down
the line
shepazu: worried about
discrepancy between implementation
... it bothers me as a developer
... would want to know why implementations would filter or
not
... that's a general opinion, not specific to this case
cwilso: not sure we have had strong case for not filtering
joe: if you are using as
modulation source, you could be doing the wrong thing by
filtering
... no perfect solution
cwilso: maybe we should have a
switch on oscillators
... audio oscillator vs mathematical oscillator
shepazu: +1
padenot: seems like a good solution to require filtering but allow addition of mathematical oscillator
<scribe> ACTION: padenot to document resolution of issue 127 (keep MUST) and add issue for the addition of a mathematically accurate oscillator [recorded in http://www.w3.org/2013/09/19-audio-minutes.html#action01]
<trackbot> Created ACTION-73 - Document resolution of issue 127 (keep must) and add issue for the addition of a mathematically accurate oscillator [on Paul Adenot - due 2013-09-26].
rtoyg_: discussion about disconnecting a source node when disconnected and reconnected
<padenot> you are very voipy here
rtoyg_: what does the audio source do. continue as paused, as it it had been playing, ...
cwilso: split consensus on
whether there was a per-sourcenode timeline or universal
timeline
... ROC suggesting a patch-cable model where you have a single
timeline always progressing
... Jer suggesting pausing model because data not going
through
... (rough summary)
shepazu: would this be a way people would use to pause a sound?
cwilso: there was an article
which was suggesting so
... I mentioned to them it was a bug, not a feature
... challenge with pausing the node is that for some sources
it's not easy to pause
... e.g live mediastream
<cwilso> (or live audio input)
<scribe> ACTION: cwilso to create an issue in github, summarise thread at http://lists.w3.org/Archives/Public/public-audio/2013JulSep/thread.html#msg1812 [recorded in http://www.w3.org/2013/09/19-audio-minutes.html#action02]
<trackbot> Created ACTION-74 - Create an issue in github, summarise thread at http://lists.w3.org/archives/public/public-audio/2013julsep/thread.html#msg1812 [on Chris Wilson - due 2013-09-26].
olivier: find it interesting that developers use the spec that way
cwilso: no ideal answer
... best answer probably the last round with ROC
... problem with leaks
... it's hard to avoid that and avoid observability of GC
ehsan: wanted to second what ROC
said on the thread
... always an issue with eventtarget on the platform
... scriptprocessornode should probably not be an exception
here
cwilso: agree this is probably
the best we can do
... problem is we made it so all nodes would just disappear
when done
... but I have no better solution
... this will be something to highlight in spec and developer
doc
ehsan: there are other nodes
cwilso: true, but more likely here. No point in using scriptprocessornode if you don't set the oncomplete event
ehsan: counter-example is looping audiobuffersourcenode
cwilso: but it wouldn't release
<cwilso> In short: we should carefully determine places where leaks are likely to happen, and properly evangelize anti-leaking patterns.
olivier: guess we won't be doing anything about it but need to ack that it is imperfect
cwilso: we need to add mention in the spec
<scribe> ACTION: ehsan to add issue in github about disconnected scriptprocessornode, mention to be added in the spec [recorded in http://www.w3.org/2013/09/19-audio-minutes.html#action03]
<trackbot> Created ACTION-75 - Add issue in github about disconnected scriptprocessornode, mention to be added in the spec [on Ehsan Akhgari - due 2013-09-26].
ehsan: discussed a long time ago with crogers
cwilso: agree that readonly float
is rational given the use today
... want to introduce side chaining
... use signal to duck the level of another signal
... used a lot in music
... having a side chain compression would be useful
... but would require changes to the spec
... can see having reduction as direct output
... would still like for us to add side chain compression or
have it as audionode so you can connect it
... way that I re-built vocoder was emulating that
olivier: relevant for v1 or wishlist?
cwilso: some exposure of this
should be v1, but don't have a plan on how to do it
... could do as scriptprocessornode but then adding latency
olivier: asking ehsan - does this sound like an additional feature?
ehsan: plannning to ship in 6 weeks, would rather not have any new feature in the API
cwilso: not arguing this is a new
feature for dynamic compression
... still think it is an important one
... don't think this should be done in 6 weeks but then don't
think everything else will
ehsan: would like a solution that
could be backward compatible
... we could just make it a float value
cwilso: if I understand ehsan -
you're saying that we're not solving the use case. Exposing it
as readonly float for now, add side chaining as an additional
feature later
... agree that making it a readonly float is the right thing to
do now
ehsan: trust your judgement on usefulness of the use case
<scribe> ACTION: cwilso to add issue to make compressor node reduction a readonly float [recorded in http://www.w3.org/2013/09/19-audio-minutes.html#action04]
<trackbot> Created ACTION-76 - Add issue to make compressor node reduction a readonly float [on Chris Wilson - due 2013-09-26].
<scribe> ACTION: cwilso to document use case for side chain compression as issue in github, [recorded in http://www.w3.org/2013/09/19-audio-minutes.html#action05]
<trackbot> Created ACTION-77 - Document use case for side chain compression as issue in github, [on Chris Wilson - due 2013-09-26].
next meeting is scheduled for 3rd of October
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) No ScribeNick specified. Guessing ScribeNick: olivier Inferring Scribes: olivier WARNING: No "Present: ... " found! Possibly Present: Doug_Schepers IPcaller aaaa aabb aacc audio aut0mata chrislowis colinbdclark cwilso eXhumed ehsan from gmandyam has heath https inserted joe johnwbyrd1 joined marcosc marcosc_ marcosc__ mdjp olivier padenot paul___irish rtoyg rtoyg_ rtyog shepazu tobie toyoshiAw trackbot You can indicate people for the Present list like this: <dbooth> Present: dbooth jonathan mary <dbooth> Present+ amy Agenda: http://lists.w3.org/Archives/Public/public-audio/2013JulSep/1907.html Found Date: 19 Sep 2013 Guessing minutes URL: http://www.w3.org/2013/09/19-audio-minutes.html People with action items: cwilso ehsan padenot[End of scribe.perl diagnostic output]