Audio WG Teleconference -- 17 Oct 2012

<chris> zakim: what is conference code?

<gcardoso> zakim ??P18 is me

<chris> anybody know the conference code?

audio

<gcardoso> 28346

yup

<chris> thanks!

Publishing FPWD of MIDI spec

<tmichel> https://dvcs.w3.org/hg/audio/raw-file/tip/midi/specification.html

CWilson: still work to do

… but now the spec has a bit of intro, which was missing

… not concerned about going to FPWD

Olivier: no showstopper issue?

CWilson: no

CRogers: one of the things which is confusing, there isn't really much of an introduction

CWilson: have you looked at it in the past week and half

CRogers: intro section now is short - one sentence
... not a showstopper

… but even the structure is a little confusing, starts directly with getMIDIAccess

… would be good to start with consideration about security security and privacy

CWilson: I can take another round explaining the character of the API in the intro

… like the webaudio API, explain how it works

… re security and privacy, there's some big disconnects and need more thoughts

Doug: in general, FPWD can be quite sketchy

<gmandyam> Giri Mandyam, Qualcomm Innovation Center, has joined

… this one would be quite complete already

… would benefit from having a little more explanation

… expanding intro would be goo

… as would some caveat in security section

… but I think current draft probably sufficient for FPWD

Olivier: would be good to publish before TPAC

CWilson: we've had discussion internally about section for privacy/security

… the whole design has implication about privacy

… but e.g Java exposes such MIDI info without warning

… hence "this may be a privacy concern" rather than "this is a privacy concern" or "you must pop a dialog"

… happy to publish before TPAC

… and could work on the two sections today

Olivier: if we can do edits today, we can have a CfC and move to publish if no objection

Doug: changes required are mostly editorial, not normative

… and we can play with later

… so we can resolve to publish as is, ask that you make the changes

… about security/privacy, the draft should ack that we are still thinking about the implications, that the section is work in progress

Doug: as side note, just because Java does thing in a specific way doesn't mean it's a good privacy model

CWilson: agree, it just means this info has been exposed and we need to think about it

CRogers: might be a difference between MIDI input and output

no need to get permission for e.g joystick API

CRogers: but for writing MIDI it would probably need permission

Doug: could I control my home security/audio/lighting system? or even in a facility?

CWilson: if you have a MIDI system hooked up that way

… e.g stage control

Doug: wondering if the scope is implicitly/explicitly mentioned

[CWilson explains how you could plug in a whole MIDI system]

Doug: given the focus on IoT, we should at some point try to look at that use case

CWilson: there is a section about this, explaining the API gives you access to MIDI devices

… what you do with them is up to you

Olivier: may be worth mentioning that in the intro

Doug: also, may be worth dispelling the notion that "MIDI is crappy audio playback"

CRogers: explain that it has nothing to do with .mid files

Doug: exactly right

<gmandyam> +q

Giri: is a call for exclusion going to come with FPWD?

Doug: yes, that will be the first opportunity for exclusion

… which is why it's important to have most features in FPWD

Web Audio API: Channel handling / naming

https://www.w3.org/Bugs/Public/show_bug.cgi?id=17396

https://www.w3.org/Bugs/Public/show_bug.cgi?id=17351

https://www.w3.org/Bugs/Public/show_bug.cgi?id=17379

Looking at https://www.w3.org/Bugs/Public/show_bug.cgi?id=17396 AudioDestinationNode.numberOfChannels

CRogers: we haven't defined the channel layout

… we have a few, mono, stereo, quad etc

… but devices can have unusual channel mappings

… so automated downmixing doesn't make sense

… you could take the first 2 and drop all the rest

… would be better when we have implementations going beyond 2 channels

… expecting that even for e.g mono smartphones, the audiocontext could still say 2, and the implementation would mix down to one

Olivier: so do we want more than 2 channels in "v1"?

CRogers: you can process large numbers of channels in the middle of the graph

… at the minimum the representation should probably say stereo

Olivier: that leads us to https://www.w3.org/Bugs/Public/show_bug.cgi?id=17379

CRogers: yes, we need to add something to describe the channel layout

… we should expose that information through a channel layout attribute

… so that developers can mix accordingly

Olivier: any thought on how to describe the channel layout?

CRogers: for example, CoreAudio has a way to get a list of channels and tag

… but there is a simpler approach where typical setups are numbered

… the approach with an array of channels and label for each one feels overkill ATM

… could be a readonly value for the time being

… has to be readonly since it maps the physical output

Olivier: how would implementation deal with plugging a sound card with 16 channels

CRogers: max number of channels would be 16, and there would be a constant for "undefined" or "not applicable" layout

<scribe> ACTION: CRogers to summarise channel layout discussion into Bug 17379 [recorded in http://www.w3.org/2012/10/17-audio-minutes.html#action01]

<trackbot> Created ACTION-50 - Summarise channel layout discussion into Bug 17379 [on Chris Rogers - due 2012-10-24].

reminding members to rejoin Audio WG

<scribe> ACTION: Olivier to send reminder for group participants to re-join [recorded in http://www.w3.org/2012/10/17-audio-minutes.html#action02]

<trackbot> Sorry, ambiguous username (more than one match) - Olivier

<trackbot> Try using a different identifier, such as family name or username (eg. ot, folivier3)

<scribe> ACTION: OT to send reminder to send to all AC reps for group participants to re-join [recorded in http://www.w3.org/2012/10/17-audio-minutes.html#action03]

<trackbot> Created ACTION-51 - Send reminder to send to all AC reps for group participants to re-join [on Olivier Thereaux - due 2012-10-24].

<chrislowis> Zakim: p10 is chrislowis

Giri: there is a comment on recording media streams in GetUserMedia

<gmandyam> See http://lists.w3.org/Archives/Public/public-media-capture/2012Oct/0066.html

- DRAFT -

Audio WG Teleconference

17 Oct 2012

Attendees

Contents

Publishing FPWD of MIDI spec

Web Audio API: Channel handling / naming

reminding members to rejoin Audio WG

Summary of Action Items

Scribe.perl diagnostic output