Re: kAudioFormatFlagsAudioUnitCanonical
Re: kAudioFormatFlagsAudioUnitCanonical
- Subject: Re: kAudioFormatFlagsAudioUnitCanonical
- From: Doug Wyatt <email@hidden>
- Date: Thu, 26 Feb 2009 13:08:17 -0800
On Feb 26, 2009, at 11:40 , Chris Adamson wrote:
So, the format of kAudioFormatFlagsAudioUnitCanonical is 8.24 fixed
point (mentioned in some doc, as well as in CoreAudioTypes.h).
on iPhone, yes
Pardon the "still getting my C sensibilities back" question, but are
there any utilities in the iPhone SDK for working with 8.24 fixed,
like the Fixed data type or FloatToFixed function from the Mac SDK?
no
There's a AudioUnitSampleType in CoreAudioTypes.h, but it's just
defined as SFloat32.
Float32 on the desktop, SInt32 on iPhone.
Do I need to plan on doing my own bit-shifting? So far, I'm
generating square waves by setting the high eight bits of each
sample to either 0x80 to 0x00.
um, working with fixed-point you will be doing a lot of bit shifting.
Having brought it up, is it genuinely advantageous to use the AU
canonical format for synthesized audio, or is it just as well to use
something simpler, like 16-bit ints? I know Jens made this point a
few weeks back: <http://lists.apple.com/archives/coreaudio-api/2009/Feb/msg00263.html
>
as a rough rule of thumb, consider 8.24 if you have a relatively
complex signal chain and it's important to maintain precision and
dynamic range. use plain 16 otherwise.
Doug
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden