Noise with integer mode enabled

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Enable Exclusive access mode & Integer mode
2. Leave forced upsampling to none
3. Play a 44.1/16 bit FLAC file

What is the expected output? What do you see instead?
The file plays with continuous clicking noises/repeateds 'tap'.

What version of the product are you using? On what operating system?
Audirvana 0.9.1 on Osx 10.6.7. Using a Nuforce Icon HDP connected via USB.

Please provide any additional information below.
The issue manifest only when playing using integer mode (normal playing or in 
exclusive access do not cause the issue) on 44.1/16 files (i tested FLAC and 
Apple Lossless). 96/24 or 192/24 FLAC do not present this problem. If i enable 
'Maximum sample rate upsampling' the issue don't appear and the file plays 
correctly, tough the 'loading' bar remains for a long time. Setting to 4x or 2x 
cause to output only noise, similar to static noise.
Attached the debug log.

Original issue reported on code.google.com by feifw...@gmail.com on 24 May 2011 at 4:48

Attachments:

• audirvana_debug.txt

[GoogleCodeExporter]

This clicking is a known Apple CoreAudio bug.
A workaround is to connect the DAC to a powered USB hub.

What are the exact circumstances that lead you to get static noise at 4x or 2x 
oversampling? What SRC are you using (Apple or libSampleRate) ? What is the 
source file type, characteristics ?

Original comment by au.damie...@gmail.com on 24 May 2011 at 9:10

• Changed state: Started

[GoogleCodeExporter]

I didn't know of this CoreAudio bug. As you said, connecting to a powered USB 
hub fixed the clicking issue.
The static at 4x or 2x oversampling remained and it manifest itself with both 
SRC. The tested files are FLAC files ripped from a cd using XLD, a standard 
44.1 khz/16 bit. For testing other things, even if maybe it doesn't really make 
sense, i tried with this setup a 96 khz/24 bit file and there i got a kernel 
panic. This one was a FLAC, from HDTracks (to recap: exclusive access, integer 
mode, 4x or 2x oversampling, with 'none' it plays without problems).
Attached the kernel log.

Original comment by feifw...@gmail.com on 24 May 2011 at 11:08

Attachments:

• Kernel_2011-05-25-004308_mname.panic

[GoogleCodeExporter]

Ok, I think I get it for the 2x oversampling: the HDP does offer 88.2 only in 
16bit. Weird... I'll need to add a check for this. And not select 88.2 when in 
Int mode.

Regarding the kernel panic, when has it happened ?
During device init ? In middle of playback ?
Have you had a "static noise" 2x after the last reboot ?

I'll fix the 2x thing first. Or at least don't use 2x oversampling that doesn't 
except for 48kHz tracks.

Original comment by au.damie...@gmail.com on 25 May 2011 at 5:35

[GoogleCodeExporter]

I've committed the fix to x2/x4 oversampling to r158.

If you still get issues, please tell me.

Original comment by au.damie...@gmail.com on 25 May 2011 at 3:00

• Changed state: Fixed

[GoogleCodeExporter]

First of all, thanks for the fixes!

I have installed 0.9.2 and ran various tests.
Before this, for the kernel panic on 0.9.1 i'm not sure, but i think it has 
occured right after device initialization when the display shows 00:00 time. 
After reboot the 'static noise' persisted, but that changed with the new 
version.

Now for the various test (all on flac files SRC Core Audio) that i ran on 
0.9.2. Some of these are nonsensincal in real use, but i have done them just to 
see how the program will react.
I list the test using this format:
file info: info displayed below audirvana display [ERROR/NOTE]

Using integer mode, no upsampling:
- 44.1/16 : 24/44.1 [OK]
- 88.2/24 : 24/96 [OK, FIXED]
- 96/24 : 24/96 [OK]
- 192/24 : 24/96 [OK]

Using integer mode, 4x or 2x:
- 44.1/16 : 24/44.1 [OK, FIXED]
- 88.2/24 : 16/88.2 [OK]
- 96/24 : 24/96 [OK]
- 192/24 : 24/96 [KERNEL PANIC]

Normal (no exclusive acces, no integer), no upsampling:
- 44.1/16 : 24/44.1 [OK/CHOPPY NOISE]
- 88.2/24 : 24/88.2 [DISTORT]
- 96/24 : 24/96 [OK/CHOPPY NOISE]
- 192/24 : 24/96 [OK/CHOPPY NOISE]

Normal (no exclusive acces, no integer), 4x or 2x:
- 44.1/16 : 24/88.2 [DISTORT]
(other formats not tested)

Now for the details:
The problem at 44.1 using integer and 4x or 2x upsample disappeared. In the 
same way, 88.2 in integer mode plays correctly: in previous version they used 
to be reproduced distorted (intelligible voices but whistling sounds, ecc..). 
Those are marked FIXED here.

Using a 192/24 file in integer mode with 4x or 2x upsampling(ok, no much sense 
in this setting...) i got a kernel panic probably similar of the previous one. 
I attached screenshots of the freezed screen: device was initialized and the 
song was starting. Note that the 96/24 file in this scenario ran. Don't know if 
it was just for pure luck.

Using normal mode no upsampling with a 88.2/24 file or 4x/2x upsample with a 
44.1 file resulted in a distorted sound (DISTORT in the list), identical to the 
whistling noise described previously. Note the values: integer mode upscale to 
96 for 88.2 or degrade to 16 bit with upsampling 4x/2x and plays ok, 'normal' 
mode keeps 88.2 on 24 bits.

For last a problem that i can't reproduce consistently: those marked with 
CHOPPY NOISE show a bug that appear to be unrelated to all of this. Those where 
tested in 'normal' mode no upsample. When played by itself they show no 
problems. 44.1 plays correctly every time in this case. But if i play a 96 or 
192 file and THEN play a 44.1/96/192 file i get a choppy sound (interruptions 
here and there with little pop noises). If i play another file of any bitrate, 
the new file plays correctly. A thing that i noted is that when the 'choppy' 
file run, it doesn't show the gray loading bar that appears normally under the 
scrub bar: it shows if i play another file after this. For testing this i click 
some files on the playlist, just making them plays for a few seconds (i wait 
that they are totally loaded).
Don't know if it was a transient problem, i will report if i can reproduce it 
better.

Original comment by feifw...@gmail.com on 25 May 2011 at 8:02

Attachments:

• audirv1.jpg
• audirv2.jpg
• Kernel_2011-05-25-203600_mname.panic