Kodi Community Forum
AudioEngine branch - DO NOT REQUEST BINARY BUILDS - Printable Version

+- Kodi Community Forum (https://forum.kodi.tv)
+-- Forum: Development (https://forum.kodi.tv/forumdisplay.php?fid=32)
+--- Forum: Kodi Application (https://forum.kodi.tv/forumdisplay.php?fid=93)
+--- Thread: AudioEngine branch - DO NOT REQUEST BINARY BUILDS (/showthread.php?tid=78289)



RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - cdhesse - 2012-04-01

Question... I've read through this thread, where gnif says AE always uses AE_FMT_FLOAT until the end when it is converted to what is supported. My question is, the "supported" statement, does this refer to only the video card (HDMI), or a combination of the video/sound card PLUS the connected device at the other end of the HDMI?

I noticed in the code...

* @note This is ordered from the worst to best preferred formats
*/
enum AEDataFormat
{
AE_FMT_INVALID = -1,

AE_FMT_U8,
AE_FMT_S8,

AE_FMT_S16BE,
AE_FMT_S16LE,
AE_FMT_S16NE,

AE_FMT_S32BE,
AE_FMT_S32LE,
AE_FMT_S32NE,

AE_FMT_S24BE4,
AE_FMT_S24LE4,
AE_FMT_S24NE4, /* S24 in 4 bytes */

AE_FMT_S24BE3,
AE_FMT_S24LE3,
AE_FMT_S24NE3, /* S24 in 3 bytes */

AE_FMT_DOUBLE,
AE_FMT_FLOAT,

/* Bitstream formats */
AE_FMT_AC3,
AE_FMT_DTS,
AE_FMT_EAC3,
AE_FMT_TRUEHD,
AE_FMT_DTSHD,

AE_FMT_MAX

Are there certain cards that will do AE_FMT_FLOAT, thus resulting in one less audio conversion?

Thanks!


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - DDDamian - 2012-04-01

It will look at the formats supported by the audio device selected in System>Audio Output where the driver is listed.

Yes, if your device supports float there is one less conversion. Some of the Creative cards support float, as do other higher-end soundcards, but most GPU HDMI ports support integer only, whether 16bit (AE_FMT_S16xx) or 24bit packed into 32bit structures (AE_FMT_S24xx4)


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - cdhesse - 2012-04-02

Thanks!

So is 32xx better than the 24xx versions? Does this have any audible impact?


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - DDDamian - 2012-04-02

Heh that's gonna foster arguements lol. But both mathmatically, scientifically and actual content-wise, no, there is no audible difference between the 32 and 24-bit acoustics. Many will argue 16-bit is all you'll ever need. But I hear very audible differences between 16-bit up to about 20-bit. The last bit of 24-bit audio is almost certainly "lost" in the electrical noise of single resistors. So 24-bit really does push the edge in every way.

Don't forget each bit doubles the available data, so the difference between 16-bit and 17-bit is huge. But the difference between 24-bit and 32-bit, while mathematically gigantic is for audible purposes zero.


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - EGOvoruhk - 2012-04-06

(2012-04-02, 03:34)DDDamian Wrote: But I hear very audible differences between 16-bit up to about 20-bit

You're either hearing elements added by the system, and not the source, or you were born with a "deffect" that enables you to hear more than a normal human can

http://www.aes.org/e-lib/browse.cfm?elib=14195
http://people.xiph.org/~xiphmont/demo/neil-young.html

That seems to be the "go to" for the current HD audio argument, and it makes a hell of a lot of sense. You can't really deny those results, especially factoring in who took them, and the hardware they were taken with




RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - DDDamian - 2012-04-06

Yeah, I've read all the studies lol. And I knew someone would pipe up, Guess the AudioEngine branch isn't for you rofl.


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - EGOvoruhk - 2012-04-06

(2012-04-06, 01:04)DDDamian Wrote: Yeah, I've read all the studies lol. And I knew someone would pipe up, Guess the AudioEngine branch isn't for you rofl.

I'll still prefer using it to bitstream my Blu-rays, because even though it's now a waste of space, the lossless audio is going to obviously sound better than a lossy DTS track. But when it comes to making my own rips, I'll be encoding the DTS-HD/TrueHD to 16-bit/48kHz FLAC

In response to the article, I was internally torn up when reading it. On one side I was happy that I could save space in regards to audio, but it was also kind of a bummer knowing that we're both a.) our limit in what we can hear is so "low" and b.) that studios are trying to act as if we aren't as limited. Though I guess that last one shouldn't come as a total surprise. I'm sure they'll be even another audio spec down the road they'll claim we need Laugh


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - DDDamian - 2012-04-06

Just follow your ears, not the studies. If it is placebo me and my dog are both enjoying it immensely Smile

And that new spec is DXD (32bit, 384kbps) - even I'm calling overkill on that lol


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - DDDamian - 2012-04-07

AudioEngine is a complete re-write of core XBMC audio. It is a full install, not a plugin. It's most likely either the driver or a setting in the driver for your USB DAC.


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - spiff - 2012-04-08

No matter what OS you are on it is NOT okay to post a support request in a forum that clearly states that it is not for end user support in a thread that states explicitly in its title that your request will not be tolerated. FFS


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - Martijn - 2012-04-08

(2012-04-08, 11:14)Nemrod Wrote: Of course it is important, someone took time trying to help me ...

not for end user support!

Edit:
Removed your posts because they don't belong here. Go ask in general help and support section


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - rabsid3 - 2012-04-08

hi guys,
what is the latest git version with working passthrough? I'd like to try it on my ubuntu machine.
thx


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - wingrunr21 - 2012-04-09

(2012-04-08, 00:40)spiff Wrote: No matter what OS you are on it is NOT okay to post a support request in a forum that clearly states that it is not for end user support in a thread that states explicitly in its title that your request will not be tolerated. FFS

(2012-04-08, 12:03)Martijn Wrote: not for end user support!

Edit:
Removed your posts because they don't belong here. Go ask in general help and support section

I have started a Linux thread here: http://forum.xbmc.org/showthread.php?tid=128298&pid=1070552#pid1070552
The Windows thread is here: http://forum.xbmc.org/showthread.php?tid=124600

Mods, in order to keep this thread clean perhaps the first post can be updated to reflect these two other threads. Hopefully erhnam won't mind.


RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - gnif - 2012-04-10

(2012-04-01, 19:20)cdhesse Wrote: Are there certain cards that will do AE_FMT_FLOAT, thus resulting in one less audio conversion?
Yes that is correct.

For those interested, over the last day a heap of fixes have gone in to correct pass-through issues, such as:

* random crashing
* choppy/stuttering output on some DTS files
* buffer overflow bug causing a sync calculation issue
* 2.0 streams will no longer be encoded to AC3

A patch from DDDamian has also been merged that should fix things somewhat under windows.

We expect to have AE merged into master very soon as we gear up for peer review and then merge.




RE: AudioEngine branch - DO NOT REQUEST BINARY BUILDS - cdhesse - 2012-04-10

Hi gnif,

DTS-HD (or anything else passthrough) still doesn't work for me.

I've tried HDMI, and custom hw:0,9 as well as hdmi:NVidia for passthrough but always get this error. The applicable part of the log looks to me to be:

14:37:40 T:140249841436416 INFO: CAESinkALSA::Initialize - Attempting to open device hdmi,AES0=0x06,AES1=0x82,AES2=0x00,AES3=0x0e
14:37:40 T:140249841436416 ERROR: CAESinkALSA::Initialize - snd_pcm_open_lconf(-2) - hdmi,AES0=0x06,AES1=0x82,AES2=0x00,AES3=0x0e
14:37:40 T:140249841436416 ERROR: CAESinkOSS::Initialize - Failed to get supported formats, assuming AFMT_S16_NE

Full log can be found at http://pastebin.com/ekSTZy5R

I hope this is a bug and not something I'm doing wrong...