W3C

- DRAFT -

Audio Working Group Teleconference

16 Mar 2017

See also: IRC log

Attendees

Present
Joe, Paul, Ray, Jerry, Chris
Regrets
matt
Chair
Joe
Scribe
ChrisL

Contents


<rtoyg_m> G'day!

<rtoyg_m> present rtoyg_m

<padenot> hi all !

<rtoyg_m> Aloha!

hi folks

Review Pull Request status

<scribe> scribenick: ChrisL

joe: some of these PR are mergeable

<joe> https://github.com/WebAudio/web-audio-api/pull/1161

joe: ray had one comment, look good paul?

padenot: need to read in detail first

<joe> https://github.com/WebAudio/web-audio-api/pull/1158

joe: will merge after you have had a chance to read it closely
... ray this is yours actually. will mark assignee

<joe> https://github.com/WebAudio/web-audio-api/pull/1154

(agree to merge)

<joe> https://github.com/WebAudio/web-audio-api/pull/1150

<ghaudiobot> [web-audio-api] rtoy closed pull request #1154: Fix #1116: Make PeriodicWaveOptions optional (gh-pages...1116-fix-periodicwave-constructor) https://github.com/WebAudio/web-audio-api/pull/1154

<ghaudiobot> [web-audio-api] rtoy pushed 2 new commits to gh-pages: https://github.com/WebAudio/web-audio-api/compare/b77740ec5c2b...4977e6c9d6de

<ghaudiobot> web-audio-api/gh-pages a28c946 Raymond Toy: Fix #1116: Make PeriodicWaveOptions optional...

<ghaudiobot> web-audio-api/gh-pages 4977e6c rtoy: Merge pull request #1154 from rtoy/1116-fix-periodicwave-constructor...

ray: chris asked about epsilon rather than zero, need to check

<joe> https://github.com/WebAudio/web-audio-api/pull/1144

padenot: I need to get back to the constructor stuff

ray: looks very good

padenot: couple of typos
... worried about rebasing
... can finish in 10 minutes

<joe> https://github.com/WebAudio/web-audio-api/pull/1143

joe: padenot where is this one?

padenot: need to look at this

joe: lets check the raw git

<joe> https://rawgit.com/WebAudio/web-audio-api/329a623/index.html#playback-AudioBufferSourceNode

ray: comments were small and minor

joe: can finish this in 30 min

ChrisL: simpler and we are all on the same page now

padenot: negative playback rate?

joe: yes
... most issues on negative and sign change playback rate

padenot: clearear

<joe> https://github.com/WebAudio/web-audio-api/pull/1141

joe: does what ray askes, and includes feedback from padenot
... ray approved changes

padenot: lgtm

<ghaudiobot> [web-audio-api] joeberkovitz closed pull request #1141: Clarify language around resuming closed AudioContexts. (gh-pages...1100-resume-closed-ctx) https://github.com/WebAudio/web-audio-api/pull/1141

<joe> https://github.com/WebAudio/web-audio-api/pull/1065

<ghaudiobot> [web-audio-api] joeberkovitz pushed 6 new commits to gh-pages: https://github.com/WebAudio/web-audio-api/compare/4977e6c9d6de...be4dd3d4a212

<ghaudiobot> web-audio-api/gh-pages 6faae40 joeberkovitz: Clarify language around resuming closed AudioContexts....

<ghaudiobot> web-audio-api/gh-pages 2a9d1ff joeberkovitz: Merge branch 'gh-pages' into 1100-resume-closed-ctx

<ghaudiobot> web-audio-api/gh-pages fbb631d joeberkovitz: Respond to @padenot review feedback.

<ghaudiobot> [web-audio-api] joeberkovitz deleted 1100-resume-closed-ctx at d440f06: https://github.com/WebAudio/web-audio-api/commit/d440f06

joe: paul, you and karl were going to look at this

padenot: will ping him, may be on vacation
... has some implications
... if he does not answer, will do myself

Go over new or existing issues

joe: looking at v1 blockers

<joe> https://github.com/WebAudio/web-audio-api/issues?utf8=%E2%9C%93&q=is%3Aopen%20is%3Aissue%20label%3A%22V1%20Blocker%22%20

joe: some do not have open PRs, most do

padenot: 1162 I filed, found in interop testing between chrome & firefox, defined for suspend but not close
... propose to do the same
... need a review for my proposal

ChrisL: makes sense to me

padenot: in chrome it stops the element

ray: will take a look. Not sure what we do on suspend of a context

padenot: is supposed to to let the media element continue, not buffer

ray: chrome makes it stop

joe: if context uses a source node, they ride on top of the stream and should not have an effect

padenot: was reported by someone who was also doing capture

joe: 1135 is what the constructor PR addresses, right?

padenot: no, it is th audio node constructors, was going to add the others like periodic wave,we can do audio buffer
... very small compared to audio node

joe: 1117 has the big PR, right?
... gets closed when PR for 1135 lands

padenot: yes

joe: 1089, matt was involved
... waiting for bbc feedback

jdsmith: was also looking at this, think we return a consistent ordering

padenot: windows uses smpte ordering

joe: jdsmith could you take this?
... matt was going to get feedback from other developers

padenot: talked with some VR devs at facebook, they like this proposal

it sounds good to me

joe: lets go with that

jdsmith: we are good with that

padenot: lack of a good authoritative spec that does not cost money

ChrisL: point to the smpte spec and also say what the ordering is, so developers do not have to buy a spec

joe: put a table into the issue

ray: will bring to our team once it lands

<hongchan> (it was hongchan)

padenot: talked with someone who has the spec, confirmed it is correct. it is the same as windows ordering
... the windows order is public

jdsmith: waveX
... will do that

joe: 5 more bugs
... 1079 has a PR, almost there
... these respec warnings!

padenot: contiguous IDL allows integrating tobies tool that lets us see rendered html of PRs with a diff
... easy to set up with our repo but errs because we have errors in our spec

ray: can ook into it after constructor stuff lands

joe: 95 is about to be closed
... issue 10! the dynamics compressor node. waiting on padenot
... ok so we will wait another week. any questions that need to be answered?

Discuss upcoming Web Audio Conference

joe: matt was going to lead us on these but is not here today. not sure what issues there are

Discuss potential for F2F meeting

joe: this is our traditional public feedback session. it is august. tpac is november.

padenot: so 3 months between them, a bit close

ray: prefer a june f2f

joe: consistent but slow progress, v1 blockers gradually falling. we will be done in may or june. sounds like a good time

jdsmith: goal of f2f is to get to CR?

joe: yes

<scribe> ACTION: joe and matt to fix a date for f2f [recorded in http://www.w3.org/2017/03/16-audio-minutes.html#action01]

<trackbot> Created ACTION-133 - And matt to fix a date for f2f [on Joe Berkovitz - due 2017-03-23].

<joe> https://github.com/WebAudio/web-audio-api/issues?q=is%3Aopen+is%3Aissue+no%3Amilestone

joe: do we have new issues wirth no milestone? yes we have 11
... lets look at the ones flagged as needing review

<joe> https://github.com/WebAudio/web-audio-api/issues/1148

jdsmith: have not looked at this one

joe: this is an architectural issue of no formalism on streaming and playback, so should defer for v2

jdsmith: eme status is really unknown, lots of objections

(agreement to defer to v2)

<joe> https://github.com/WebAudio/web-audio-api/issues/1151

joe: okay to defer?

ray: yes, just a bit inconsistent. can push to v.next

padenot: makes sense

joe: but do in v.1?

padenot: also different that you dont create the normal destination

(goes to v.next)

<joe> https://github.com/WebAudio/web-audio-api/issues/1152

ray: 1152 constructor has no dictionary, seems odd.

joe: makes sense to do as the contexts are so related
... is there much spec work?

ray: small, add a dict like the others

joe: and the additional feature?

ray: can inherit, so fairly easy

joe: can put in v1 but it isn't a v1 blocker

jdsmith: not a blocker
... nice refinement

ChrisL: sgtm

<joe> https://github.com/WebAudio/web-audio-api/issues/1153

ray: it is for reverb modelling, uses gigantic arrays

joe: so auio worklet node

ray: needs thousands
... opposed to doing it here
... in IIR

joe: so we agree this is for audio worklet

ray: we will never allow thousands of coefficients

padenot: no one has asked for a number between 20 and thousands

(agreement)

1155 is a duplicate of 95

<joe> https://github.com/WebAudio/web-audio-api/issues/1157

ray: this is from an IDL linter, like introspection

padenot: makes sense

joe: ok I will spec this

ray: would like to see this, aligns with chrome

padenot: yes, looks good, trivial

jdsmith: yes

<joe> https://github.com/WebAudio/web-audio-api/issues/1159

ray: 1159, output is always stereo even on a mono source

joe: mistake?

ray: consistent, but doubles complexity

joe: it would get upmixed if next node needs stereo

ray: add a diagram with mono in and out

jdsmith: makes sense

padenot: yup

ray: everyone was doing stereo at first

joe: no breakage

ray: only if they put a channel merger on the output

joe: ok so change it?
... easy to fix

ChrisL: yes we should fix it

jdsmith: should fix it

https://github.com/WebAudio/web-audio-api/issues/1165

ray: should this be A rate?

joe: makes no difference to the algo

padenot: great if it works, concern on doing hq resampling if target changes

joe: same issue with k-rate though
... they resample whole block every render quantum

padenot: maybe there is a good fast resample that can change quickly
... perf implications, untested, with be a bit harder to do a-rate

joe: if playback and detune are constant that is when resampling comes inot the picture

padenot: memory vs cpu tradeoff

joe: leave on agenda

(adjourned)

Summary of Action Items

[NEW] ACTION: joe and matt to fix a date for f2f [recorded in http://www.w3.org/2017/03/16-audio-minutes.html#action01]
 

Summary of Resolutions

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.152 (CVS log)
$Date: 2017/03/16 17:04:02 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.152  of Date: 2017/02/06 11:04:15  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/... cl/padenot: clear/
Succeeded: s/removes/stops/
Present: Joe Paul Ray Jerry Chris
Regrets: matt
Found ScribeNick: ChrisL
Inferring Scribes: ChrisL
Found Date: 16 Mar 2017
Guessing minutes URL: http://www.w3.org/2017/03/16-audio-minutes.html
People with action items: joe matt

[End of scribe.perl diagnostic output]