TamTam activities sounds staticky on XO 1.5 A2 laptops

[wad]

On an XO-1.5 A2 laptop (SN 14) running Q3e05 firmware and a linux distro with a 2.6.30_xo1.5-20090717.0115.1.olpc.ba8f22b kernel, the audio on Tam Tam activites sounds horrible.

When a sound is played, it comes out with occasional/regular static (buffer underflow ?).

This happens over both the headphones and the speakers.

When OFW and other activities play sounds, this problem does not appear to be present.

As the Tam Tam Activities are one of the premier ways of playing with sound on the XO, having them sound bad makes the new hardware sound bad...

[cjb]

Would be good to see a dmesg.

I'm sure we're not using the same version of csound as 802; perhaps something's up with the new version.

[wad]

This is a dmesg log relelvant to the TamTam bug on XO-1.5

[sayamindu]

Setting the value of PLUGIN_RATE to 16010 in ~/Activities/TamtTamMini.activity/common/Config.py seems to get rid of at least some of the cracking sound for me. Can anyone confirm this ?

[sayamindu]

https://bugs.launchpad.net/ubuntu/+source/sdl-mixer1.2/+bug/66483 has a long discussion on a similar problem, and a custom ~/.asoundrc mentioned in the discussion seems to act as a workaround for us as well:

pcm.!default {
type plugin
card 0
}

ctl.!default {
type hw
card 0
}

pcm.!default {
 type plug
 slave.pcm "dmixer"
}

pcm.dmixer {
 type dmix
 ipc_key 1024
 slave {
  pcm {
   type hw
   card 0
  }
  rate 44100
  period_time 0
  period_size 2048
  buffer_size 32768
 }
 slowptr 1
}

[cjb]

Tested fixed in Q3A07 and os26.

[dsd]

This broke recording (#9526)

The .asoundrc is mostly redundant: dmix is being used by default anyway, as is slowptr. I'm not sure about the period and buffer size values, but through some experimentation, the relevant thing seems to be the change of rate from the default 48000 to 44100.

so I'm going to drop this .asoundrc file and instead we'll change the dmix default rate to 44100 (inside /usr/share/alsa/alsa.conf, modified from the build script). but i'm still hopeful that we can figure this out properly, this is still a hack...

[Quozl]

symptom has returned on os39. reopening.

[cjb]

Fixed for the next build, I think -- the sed lines were added to %post --nochroot instead of %post, so they were outside the chroot.

[cjb]

test in os40

[RafaelOrtiz]

Works for me,

[anonymous]

Milestone 1.5-software deleted

[dsd]

In #12651 we finally found the root cause of this problem, so this hack will be dropped shortly.