Steve Ball - Learning about Audio in Windows Vista | scobleizer

Manip

Sorry, I thought of another question... This question is admittedly unfair (because it has little or nothing to do with the audio stack) -- I have a 'mute' button on my keyboard, the default action for Explorer (without a driver installed) is to mute the system audio, will it be possible to change this to mute the "in focus" audio stream though the use of something like TweakUI? Smiley

And, if I have three applications loaded, all with windows, am I correct in assuming at only those that use the audio sub-system are listed at the time in the volume control? Interestingly however it is setup it is a royal pain, if you don't list everything that means you can't preempt things like Internet Explorer that fire sounds depending on the content, and if you *do* list everything it is a cluttered interface. In either case I'm writing a UI bug report for the beta 2 Tongue Out

Tom Servo
W-hat?

staceyw wrote:
Very cool. But kinda glanced over the API stuff. I see a need to raise the bar for audio for the .Net developer. Say I want to create a ~simple DJ app that has at least two mp3/wma streams. So I need to be able to manage and mix the two streams, have cue points, some kinda event callbacks to update the audio graph with fine detail. Simple way to enum and select ouput devices and input devices, etc. So simple Load() and Play() on a Media player control does not really cut it anymore. Is the API going to be "at least" this rich for us for .Net? Thanks much.

Wouldn't you want to use Managed DSound in this case, anyway?

--edit: Not to rain on anyone's parade, but why was that WAVE stuff required for audio? I've yet to see an audio application glitch, outside scenarios like a hardware driver or failure temporarily locking up my system. If an audio application was shitty playback code, no system side code will fix it.

LarryOsterman

Tom Servo wrote:

staceyw wrote: Very cool. But kinda glanced over the API stuff. I see a need to raise the bar for audio for the .Net developer. Say I want to create a ~simple DJ app that has at least two mp3/wma streams. So I need to be able to manage and mix the two streams, have cue points, some kinda event callbacks to update the audio graph with fine detail. Simple way to enum and select ouput devices and input devices, etc. So simple Load() and Play() on a Media player control does not really cut it anymore. Is the API going to be "at least" this rich for us for .Net? Thanks much.

Wouldn't you want to use Managed DSound in this case, anyway?

--edit: Not to rain on anyone's parade, but why was that WAVE stuff required for audio? I've yet to see an audio application glitch, outside scenarios like a hardware driver or failure temporarily locking up my system. If an audio application was shitty playback code, no system side code will fix it.

It's actually a good question. Here's a partial answer: On my machine at home, I've NEVER had the startup sound not glitch. There are things that can cause audio to glitch - disk I/O's, long DPC's, memory pressure, etc.

LarryOsterman

Manip wrote:
Sorry, I thought of another question... This question is admittedly unfair (because it has little or nothing to do with the audio stack) -- I have a 'mute' button on my keyboard, the default action for Explorer (without a driver installed) is to mute the system audio, will it be possible to change this to mute the "in focus" audio stream though the use of something like TweakUI?

And, if I have three applications loaded, all with windows, am I correct in assuming at only those that use the audio sub-system are listed at the time in the volume control? Interestingly however it is setup it is a royal pain, if you don't list everything that means you can't preempt things like Internet Explorer that fire sounds depending on the content, and if you *do* list everything it is a cluttered interface. In either case I'm writing a UI bug report for the beta 2

Manip, believe it or not, we've got some of that covered. Not the mute button (although it's an interesting idea), but the multple app issue.

An app remains in the slider for a period of time (30 seconds, I believe) after the app has finished playing audio, to allow the user to interact with the volume control even if it's not playing. Even if the app's exited, we STILL keep it in the volume control for that period of time.

As I said, I think we've covered everything Smiley

Manip

LarryOsterman wrote:
As I said, I think we've covered everything

This quote is guaranteed to come back and bite you.... Tongue Out

I like how the system tray customiser ("Notification Area") in Windows XP works; you can fiddle with everything that is open and running but there is a list of all history below... Maybe you could add a obscure button that shows a history of the last 1 day and allows you to push it onto the volume control list for 30 seconds?

SteveBall
SteveBall

Manip wrote:
Sorry, I thought of another question... This question is admittedly unfair (because it has little or nothing to do with the audio stack) -- I have a 'mute' button on my keyboard, the default action for Explorer (without a driver installed) is to mute the system audio, will it be possible to change this to mute the "in focus" audio stream though the use of something like TweakUI?

Great questions - all fair. We've considered the idea of 'audio focus' but find that it is simply too subjective for us to assume we know what that is at any given time.  This is one reason we built the per app mixer WITH a single, always pinned "Windows Sounds" slider through which we route all system and app registered sound events so that users can set their mix preferences (in real time) between apps and all other notificiations and system sounds.

Regarding a request for a powertoy to customize mute (and or other vol behaviors): not a bad idea, and since Larry is also likely reading this, we'll discuss and consider what some useful powertoy settings might be, and see what Raymond thinks.

There is more subtle complexity in this UX than most people realize at first glance, especially when you consider multi-device scenarios, even simple ones like speakers + headset (not to mention all of the multi-streaming scenarios made possible by our new audio class driver. We're still studying these multi-device scenarios, and frankly, I expect some churn during B2 as we get usability feedback from many people who will certainly have many diverse wishes, intentions and scenarios than we have anticipated thus far. That said, this default UX is designed and optimized with one primary scenario in mind: let me adjust the relative mix of only those apps that are currently playing audio.

And, if I have three applications loaded, all with windows, am I correct in assuming at only those that use the audio sub-system are listed at the time in the volume control?  Interestingly however it is setup it is a royal pain, if you don't list everything that means you can't preempt things like Internet Explorer that fire sounds depending on the content, and if you *do* list everything it is a cluttered interface. In either case I'm writing a UI bug report for the beta 2

Really, how many apps do you ever run which wll be simultaneously playing audio? Our guess is that average will be ONE at a time. Maybe two at most if you are listening to music while playing a game. Maybe three at most if you then browse to a website with an embedded OCX or media playback. Then, maybe four if you accept a VOIP call while doing all of the above - and in this case, we've deisgned the system so that users should never have to adjust a volume control when initiating or accepting a 'communications' audio event while an 'entertainment' audio stream is playing. Also, remember, do not include PlaySound events from those registered with mmsys -- all of those will 'pour' through the pinned Windows slider.

That said, we may have missed some real scenarios that you know about, and if so, we'd love to hear about them. That is one of the primary reasons we're going public with this UX now rather than springing it on all users w/o some extended feedback and usability.

* * *

SteveBall
SteveBall

JChung2006 wrote:
It's funny -- if you think about it, the per-application volume control is like the audio equivalent of Vista glass.

I love this analogy. Thanks for the poetic observation!

* * *

siim04

Hello everybody!

I have a user case (my case) for you:

In the evening and even at night I use my computer to play music till certain time (like radios sleep function). Then the computer will be shut down so I can take it with me to work/university in the morning. And it is extremely important for me to have the sound muted when I start the computer there (I shouldn't disturb others with sounds). So I wrote an application that works almost like any other timed shutdown programs with an option to mute sound before shutting the computer down. The program is freely available on the web and has proven to be very popular in Estonia Smiley .

Now can I use the same kind of functionality in Vista? And will it be difficult to port this application to Vista or will it work without modifications? I guess modifications will be nessecary as otherwise the program would mute just itself.

Siim Karus

Tom Servo
W-hat?

Since the audio guys are monitoring the thread right now, I was wondering if KS runs in usermode, too, now. Just asking because I'm trying to figure out why DSound based playback glitches like hell under stress while KS based playback does just fine.