[Minutes] Audio WG teleconference, 2012-09-12

Hi group,

The minutes of the Audio WG teleconference for 12th September 2012 are now online:
http://www.w3.org/2012/09/12-audio-minutes.html

We met for an hour yesterday to discuss the following topics:

• 1. Interface naming (update)

Our last meeting had seen a resolution to rename setTargetValueAtTime (to setTargetAtTime) and noteOn/noteOff to start/stop, but hadn't addressed the question of noteGrainOn. The group unanimously liked the idea proposed by Chris R to not have a separate method for this, but instead to overload the start() method with extra arguments.

https://www.w3.org/Bugs/Public/show_bug.cgi?id=17344

Chris hopes to have these changes included in the spec within a couple of weeks.


• 2. Revisiting the circular graph issue

https://www.w3.org/Bugs/Public/show_bug.cgi?id=17326
This is an old issue we'd discussed at length in May but had not yet reached a conclusion on two questions. During the call this week, we reached a tentative conclusion on all three. 

 - How to deal with delay in feedback loops? 
   Suggested resolution: implementations should deal with delays in feedback loops by clamping it to the smallest block of sample they support, without throwing an exception.
 - What should be the minimum? Should it be in samples or milliseconds? 
   There were two suggestions: one to specify the minimum as a block of 128 samples (which would make things sound differently on platforms with different sample sizes, but that was deemed acceptable in the same way as graphics look different on platforms with different frame rates and power). The other was to not specify it and let implementations choose (and, I assume, expose?) their minimum delay.
 

• 3. Use Cases and Requirements review

The group is happy with the shape of the draft. We resolved to publish it as a working draft to become a working group note once we incorporate public feedback.


• 4. Loop start/stop points

https://www.w3.org/Bugs/Public/show_bug.cgi?id=17390

The question here was about defining loop start and stop points as well as loop mode. The proposed resolution is that:
 - the AudioBufferSourceNode will have (optional) start and end point values
 - defaults of 0 and 0 will signify that the whole buffer is to be looped
 - the loop attribute stays
 - start/end values should be floats
 - for now, we will not support sub-sample values and implementations will be expected to round values down to the nearest sample
 -  We will not add a loop mode attribute, but there may be one in the future (reverse, etc)


• 5. Logistics

A quick update on the rechartering (should be in effect soon) and on TPAC (remember to register, and ping me if you know you won't be making it there - we will try to have a session or two with teleconference equipment).


Next meeting will be on the 26th of September, same time.

Cheers,
Olivier