Interesting.
I'll give that a try somewhere i the next couple of days and I'll let you know how it goes.
Thanks Doug, thanks Bill !
> Subject: Re: Audio queues...again
> From: email@hidden
> Date: Mon, 27 Apr 2009 20:11:43 -0700
> CC: email@hidden; email@hidden
> To: email@hidden
>
>
> On Apr 27, 2009, at 1:53 PM, Etienne Gignac Bouchard wrote:
>
> > Thank for replying.
> >
> > I forgot to mention : iphone sdk 2.2.1, format is wav.
> >
> > I need 2 audio queues to start playing at the exact same host time.
> > I tried enqueuewithparameters on the first buffers of the queues
> > with a absolute hosttime but it is completely ignored.
>
> right, AudioQueueEnqueueBuffer (as the documentation implies by
> talking only about the options that DO work) doesn't currently support
> host time scheduling.
>
> > I have been looking into remoteIO. Although lower level, I heard
> > people have had success with them to play files simultaneously with
> > multiple audio units through a mixer.
> >
> > But if I can find a working case of enqueueing with host times, I
> > might change my mind....do you have code that works in that fashion ?
>
>
> Bill suggested an approach to a couple of us offlist. The following
> ought to work aside from having no error-checking.
>
> // pick an arbitrary point in the near future at which to start
> double secondsInFutureToStart = 0.25; // your constant here
>
> AudioQueueRef q1, q2;
>
> AudioQueueNewOutput(..., &q1);
> AudioQueueNewOutput(..., &q2);
> AudioQueueStart(q1, NULL);
> AudioQueueStart(q2, NULL);
>
> while (1) {
> AudioTimeStamp t1, t2, start1, start2;
> AudioQueueGetCurrentTime(q1, &t1);
> AudioQueueGetCurrentTime(q2, &t2);
>
> const double hostTimeFrequency = CAHostTimeBase::GetFrequency();
> UInt64 startHostTime = CAHostTimeBase::GetTheCurrentTime() +
> secondsInFutureToStart * hostTimeFrequency;
>
> if (startHostTime > t1.mHostTime && startHostTime > t2.mHostTime)
> // good, we picked times that are really in the future
> break;
> // error: the host time is in the past. try again, more conservatively
> secondsInFutureToStart *= 2;
> }
>
> // interpolate to convert the start host time to each queue's sample
> timeline
> start1.mSampleTime = t1.mSampleTime +
> ((startHostTime - t1.mHostTime) / hostTimeFrequency // seconds
> * q1sampleRate);
> start2.mSampleTime = t2.mSampleTime +
> ((startHostTime - t2.mHostTime) / hostTimeFrequency
> * q2sampleRate);
> start1.mFlags = start2.mFlags = kAudioTimeStampSampleTimeValid;
>
> AudioQueueEnqueueBufferWithParameters(q1, buf1, ...., &start1);
> AudioQueueEnqueueBufferWithParameters(q2, buf2, ...., &start2);
>
> To be even more precise you can also factor in the rate scalars in the
> sample times .
>
> hth
> Doug
>
|