View Single Post
Old 10 March 2017, 23:14   #6
Omnia fert aetas

ross's Avatar
Join Date: Mar 2017
Location: Crossing the Rubicon
Age: 47
Posts: 673
Originally Posted by Toni Wilen View Post
Switching off all interpolations and using 4 ch mode should result in non-processed output.
I'm not sure i've understood Your sentence..
How to switch-off interpolations?

Originally Posted by Toni Wilen View Post
Change should be instant but because volume in real hardware is PWM (volume value selects pulse width), results may not be exactly same if volume is continuously changed.
Yes, my idea/solution was to use some 'multiple' of PWM volume pulse cycle to reduce distortion.
With a period of 128 (maxvol*2), PAL freq. 27710, we have two complete and regular sequence; as a bonus is a frequency replicable with DMA to compare
If I sincronize copper for [startcycle:sample*2/volume|start+128:volume|start+256:sample*2/volume and so on] in teory we can have a working player...

Originally Posted by Toni Wilen View Post
sample * volume, then channel mixing (depends on selected channel mode) and interpolation, result is scaled (if not already) to 16 bit sample.
And.. YES: was exactly my formula in my scenario
This give me 4856 different sample value (>12bit!) with a single audio channel and the great property of being 'dense' on small value and 'relaxed' in high value. More dynamics and better for ear (and for compressor based on quantised table!).

Probably this is only a proof of concept, the copper is really difficult to use for synced video effect in this situation.. but if I find the time I try something.
I'm totally off-road?

Thanks Toni!
ross is offline  
Page generated in 0.05206 seconds with 9 queries