[CiNcH]

BUT - it is everything far lower than in a 5 ms range.

5ms translates to 200Hz, right? I consider that audible, very much depending on the content though.

It is not something you permanently hear. But sometimes it is just very obvious (like a melody with a rather constant tone). It is actually hard to capture the error with an iPhone...

Code:

NOTICE: ----- error: -5.917242, rr: 0.996575, prop: -0.002959, int: -0.000466
NOTICE: ----- error: 8.063054, rr: 1.003727, prop: 0.004032, int: -0.000305

That looks like a rather high P jump, no?

[fritsch]

Not really. The 5ms is a distance that is compensated over one entire second.

[CiNcH]

Currently experimenting with test patterns (constant tone). I can inject a TransportStream from a file via DVBViewer Recording Service. It will limit datarate based on the PCR. I am reading it via DVBViewer PVR Add-on which declares the stream as realtime within Kodi...

My experiment worked. rr had a very high jitter the first few seconds. And the frequency changes are very audible. But I assume that the algorithm needs those seconds to settle, so it is not really relevant. Afterwards rr is pretty close to 1.0 all the time.

[fritsch]

If there is a high error at the beginning (under certain circumstances), then you will hear the rr adjusting, the I will increase more heavily and the P wants to half the error in every step. Yes for stream start it sounds quite ugly (you can just start a youtube music video). But it should settle after max 5 seconds.

[CiNcH]

I am providing a video of the test that I did with constant tone. What I did was putting a constant tone sample on my local server and imported the URL as PVR channel so that I could more easily test the resampling.

Kodi PVR resampling audio fluctuation (The effect is not as pronounced in the video due to phone recording and several format conversions. But it can still easily be spotted.)

Whenever rr changes are > 0.001, it is clearly audible. And that happens quite a lot, not just in the beginning. It can also clearly be spotted on conventional TV, especially for music/melodies. Audio clearly gets distorted, sounding quite funny at times.

My conclusion from this is that the resampling is hardly feasible for Android TV AudioTrack with its latency.

Hope @fritsch won't again kill this conversation with his NLP skills.