AudioEngine branch - DO NOT REQUEST BINARY BUILDS

gnif Wrote:When another stream starts, it is not the only one, so it just gets mixed in to whatever is already open. This will only occur in the paplayer for gapless and cross-fade reasons.

This sounds perfect.That's the best i could think about. This is the way we can minimalize resampling occasions.

Right now to play two streams together we had to use Alsa mixing which allows only one fixed sampling rate.

Thanks for the info. I am really waiting to try it out

Another update...

* Re-open of audio device to match sample rate now implemented
* Visualizations now take float data (less format conversions)
* Lots of crash fixes
* Initial pass-through has been implemented (does not function yet & un-tested)

I have managed to remove most of the unnecessary conversions between bit-depth formats now, this means a less load on the CPU since we are working in float everywhere now. In some cases the conversion could have been as as bad as float->s16->float->s16->float->s16 for the audio, and then another float->s16->float if you add visualizations to the equation.

The worst case now is, X->float, and float->s16 for some visualizations, most though work in float too, so we are down to one conversion in most cases provided your sound card supports float output. If the audio codec returns float data, we are down to 0 conversions, which is really nice IMHO.

Whats left:
* paplayer seek/ff/rr
* paplayer stop on end of playlist
* passthrough
* transcoding to ac3

and if someone is kind enough to donate the hardware, true-hd, etc... (i only have spdif hardware).

At this rate it AE should be ready to merge into trunk fairly soon depending on my available time. In the meantime, the adventurous can give it a go, just be sure to configure with --disable-pulse as only ALSA works atm.

Again, I am not accepting ANY feature requests or bug reports

gnif Wrote:and if someone is kind enough to donate the hardware, true-hd, etc... (i only have spdif hardware).

Hi!

What would you need ? I can donate an nvidia video card with hdmi lpcm out gt220, probably capable of true-hd passthrough.

If you need an AV receiver, that's pretty expensive. The best i can do is make some donation through the xbmc donation system (but of course just with a less sum). Which would be a help for you ?

This sounds great (Pardon the pun)

My main interest with XBMC is delivering audio to my dac. I am happy with the sound now, but the possibility of bit perfect is great.

Can I ask if there will be any ability of delivering hires audio in any sense of the word?

As I understand how OSX works is that the audio/midi setup sets the rate and thats that. But certain softwares can change audio output on the fly such as amarra.

But anyhow, I look forward to it and hope are skinning folk will consider more focus into the audio section as well.

garyi, I am not sure what you are asking for as I am unfamiliar with the mac system, but if you are referring to outputting at the sample rate the system expects, yes, AE is able to handle that.

alanwww1, I have sent you a PM, but to be clear to everyone...

If you have faulty (and yes, I mean faulty) HDMI equipment, such as a smashed HDMI capable TV that still turns on, or an amplifier that powers up but produces no sound as it is burnt out, etc... that would be enough, I can hack it to work with it, as long as your willing to donate it to me for the greater good.

I am not looking for a free ride, but I can not afford the equipment either, so if people want it, and can't develop it themselves, id be happy to do it if I can procure the equipment. My aim is to get XBMC's audio up to scratch so people have more reasons to use XBMC over other media centers.

gnif. The output of audio from a mac is determined by the OS, which is set via audio/midi applications in the utilities folder. If I set it to say output at 44100 at 24bit then that is what will be outputed, if I play a higher rated audio file then it will be downsampled to the audio preference.

A number of applications on the market now, including ammara and pure music can change these settings.

In other words unless you ensure this happens (or someone does on the OSX branch) then what comes out of the mac won't necessarily be what you think you specified.

gnif, I suggest spending some time here http://developer.apple.com/mac/library/d...ction.html

I know little about CoreAudio, I spend my time with video decoding and rendering, there's lot to be done there so aside from testing, I won't be much help with porting AE to OSX. Phi handled the audio under OSX. One thing to watch out for, CoreAudio API is dependent on which SDK you target. The current SDk we target is 10.4.

gnif, I can only donate. I don't have extra HDMI equipment available. Would donations directly help you somehow?

Hi ashlar,

Please send any and all donations to me personally via my source-forge account here:
https://sourceforge.net/donate/?user_id=2785077

Good to see you guys are starting to clean and up working on fixes rather than adding much more features.

one little question....does this mean that sound settings in xbmc will be more compatible to most systems in the future? I still see a lot systems that require tuning of the asound.conf (e.g. tripple outputting sound to all channels or adding an asound.conf for outputting menu sounds).

So I guess it will be more user friendly in the near future?