Pulseeffects: Regression causes sound effect distortions

[AsavarTzeth]

Commit 9d8239545fa0f4cda4cc9fc3ef4d3a23dbe07682 and possibly ones that follow causes regressions in some of PulseEffects sound effects. Not sure which plugin, but something causes static in the sound. The music I play is still audible but the static is louder.

[wwmm]

Hmmm... This is strange. I did lots of tests with the new pipeline and everything is fine here. No static.

[wwmm]

In any case all the latest changes are being made to support a feature that people are always asking me for. Supportting turning effects on and off. In other to be able to support this I had to do lots of refactoring. Gstreamer pipeline included. Now each plugin is inside a GstInsertbin. Here I have no problems with this approach but maybe in other machines this is not true...

[AsavarTzeth]

Well I will wait and see then. Perhaps you can continue with your refactoring and give me a call when you feel you have reached a milestone. I would happily help with testing prior to a release.

One idea though. Use feature branches for big changes like this. Because now it might be harder to do a stable release which includes a fix for translations installed in a prefix, or localedir other than the python default.

[wwmm]

Hi! Do you still have this problem with the latest code in master? As I am considering making a new release it would be good to know if this is still happening

[AsavarTzeth]

I will test it when later when I am done studying for the day. Looking forward to test it.

[AsavarTzeth]

Well my testing had to be delayed until the next day. Testing was done on commit d2195f1c033153f6726718014fd651b29487d01a and there is almost no detectable issues. You even seem to have fixed those annoying gtk window size errors.

I think part of the reason why I had distortions were my own incompetence (when it comes to audio manipulation). I strongly suspect that my speakers simply are not powerful enough and I put things like output on too high levels. I say partly because on a rare occasion I still had issues, with the default settings, but that seem to be resolved now.

There was one issue I detected at the end though. If I change latency-time to any other value everything is paused, including sound spectrum analyzer and all.

Output you should get when this happens:

** (pulseeffects:3): CRITICAL **: gst_level_transform_ip: assertion 'num_int_samples % channels == 0' failed

13:17:20.141 - PulseEffects - ERROR - (gerror=GLib.Error('Internal data stream error.', 'gst-stream-error-quark', 1), debug='gstbasesrc.c(2939): gst_base_src_loop (): /GstPipeline:pipeline0/GstPulseSrc:audio_src:\nstreaming stopped, reason error (-5)')

[wwmm]

I also have this error. Both with version 2.3.5 and the latest git. Something interesting is that if we restart pulseeffects using one of the latency values that triggered this error everything is fine. It seems that the problem is not the value itself but the fact it was set on the fly. Strange... I will take a deeper look into this.

[wwmm]

I did some tests and the problem happens when setting the latency of the microphone pipeline. I am setting the same value for both pipelines but clearly values that are ok in the app pipeline crash the microphone pipeline.

Honestly I think I will leave the mic pipeline with its default latency of 10 ms and let the user configure only the latency of the applications pipeline.