For those interested I added a new song to the channel: vectorman
Nothing fancy, just felt like trying out how far you can take it with using only a single sine wave as basic wave form. And the chosen song in questions uses a lot of sines to begin with, (using the expanded channel 3 on YM2612 => 4 x 1 operator vectorman-day1) so it did not turn out too bad I think.
I wonder if this technique would allow "porting" SNES songs to SCC... or doesn't that make any sense?
You can say that in essence the SNES audio system is a more advanced version of the SCC. Due to sample memory restriction of 64kb, for SNES instrument samples are often reduced to one single period like SCC. So this type of samples could convert quite well. But bigger samples like percussion & drums, or multi tonal like chords (not that you want these as there often bad for pitching), will be very difficult and even tend to say impossible. I am still not convinced ISR sample method can cope with this.
Really nice track .
Did you use your original technique to produce the waves, or do you sample from an FM sound or a subtractive synth wave with cutoff filter?
It is much more simpler than that. Like I mentioned before for this song I limited myself to a single sine (a bit silly maybe) and then use some wave effects like stretch, multiplier, mix and gain to modify this sine wave on the fly in the track itself. Result are very basic (and static wave forms). I never used this feature much until now, but I think it will have its uses in some cases.
Basically I needed a track to try out for the re-player song conversion tool and ultimately the (z80) re-player, which I have been working on lately...
nice job! Do you like the game itself or was it just a random nice song that you picked?
Great work! BTW ave you seen the last encoder ?
It is on github on the Trilotrackr repository
I have added the possibility to specify the pitch from command line
All is very draft but it should work for encoding instruments playing a known note
the CPU usage meter seems to have some awfully high spikes... what can you tell about the CPU usage playing these files?
Great work! BTW ave you seen the last encoder ?
Yes, I did. I tried very briefly converting a sample or two still not with much success... for now I will focus my energy first on other things before I revisit this.
CPU usage meter seems to have some awfully high spikes...
The CPU meter is just for show ... in essence it is a simple idle time counter. When switching to another pattern or doing other things that take longer than 1 int to complete the counter is not properly reset and gives a spike. Also redrawing the updated wave forms (visualization part) is a large chunk on CPU time because of waiting on VDP (about one block on the meter per channel). In earlier versions drawing was hooked onto the interrupt as well. I had to unhook it because it was causing slowdown is some instances, now the wave drawing automatically drops to 30 fps if it cannot keep up anymore or CPU has more important editing thing to do.
The CPU usage for this song is actually fairly low as there are only minimal wave updates as the effects are limited to only one event per step, compared to a stack of wave effects every int in the other songs.
Do you think this technique would be usable in games, where the CPU cost may not be too high to leave room for the game itself? So, how much CPU time do you think is needed to play music with this technique?