Michael Hartmann

I could imagine that this is again chip-dependent. So some chips will have a louder (and also more distorted) filter, and other chips are a bit more tame. The current...

After some experimentation, I found that adding this line: `Vsum = ( Vsum * 3276 ) >> 12;` before https://github.com/libsidplayfp/libsidplayfp/blob/9cc3421737837558d360105dcc9ace9eb7c500f9/src/builders/residfp-builder/residfp/Filter.h#L212 makes pretty much **ALL** 6581 tunes more balanced. Acid Jazz...

I've changed my code slightly to do this now: ``` constexpr auto filterGainOut = int ( 0.85f * 256 ); if ( lp ) Vmix += ( Vlp * filterGainOut...

Isn't the filter and mixer analog? There might be losses there. If you do everything 100% to spec, you will always be off from reality, as their manufacturing was so...

I don't even think this is a chip-to-chip issue. All 6581 tunes sounded more balanced and pleasant when reducing the filter output to 80-85%. Even for tunes whose chip profiles...

I like it better now. In my branch, I went as low as 0.85 because, for most songs, it sounds even more pleasant, but then a few digi tunes sound...

I found that with a lot of tunes, there are small skips. It's not related to the tunes themselves, but rather libsidplayfp. Sometimes, I can listen to a tune without...

I believe the re-sampling process from 1 MHz to the audio device's sample rate occasionally skips a buffer.

That would not explain it, as the same function is still used, but maybe with a different sample rate. So you never ever had the same skips when rendering to...

Because it depends on the output sample rate. Your realtime audio is probably at 48 kHz, whereas you render might be 44.1 kHz.