Ok, maybe I should document this a bit now since I'm feeling like it... also maybe it would be a good idea to eventually incorporate this to the wiki.
So, the pitch sweeps in ch1 & 2. They work the same way as NES pitch sweeps. I love old audio chips, they had all these weird illogical quirks and sometimes downright bugs, which later turned out people could use to their advantage to create more creative and interesting sounds. Thing with the NES pitch sweeps is, that they don't keep track of the sweep position in any way. If you write to the pitch register, then the pitch is set to that pitch, and the pitch sweep just keeps on going from whatever pitch you set it to, even if it's lower than the previous pitch and you were going up. The pitch sweep basically just increments or decrements the pitch at an interval, no matter what the current pitch is. I've preserved this functionality in the NES instrument, which means that you can combine a pitch automation/ note detuning with the emulated pitch sweep for interesting effects. You can get the sounds very much like the ones used in Zelda 2 when you walk in to the random encounters, or the punch sounds from Punch Out... this weird jagged pitch sweep sound where the pitch curve gets modulated by the hardware's own pitch sweep... sounds very nostalgic and awesome. Another thing to note is the noise channel (channel 4). Early consoles like the NES used a very simple form of pseudo-random noise generation, which is called a LFSR, or a Linear Feedback Shift Register. It's basically just a very simple piece of binary logic that generates repeating bitstreams by a couple of very simple binary operations (XOR and SHL mostly). The repetition length of the bitstream it generates is dependent on the implementation details of the LFSR: which bits are xor:ed with which other bits, how many bits are in the register, etc. The NES LFSR had 15 bits, and 2 operation modes. Mode 0 produces a bitstream that is 32768 bits long, whereas mode 1 produces a much shorter repeating bitsream, only 93 bits. Thus, since the repeating pattern in Mode 1 is shorter, it sounds much more melodic (and kind of metallic). There are some funny quirks here too: if you switch from Mode 0 to Mode 1 in mid-note, the LFSR will not be re-seeded with the default value, but instead continues from the value it left in, which makes the 93-bit Mode 1 behave differently and produce a slightly different sound. This is consistent with NES behaviour, in a way... I feel I should also note, that some of the functionality in the NES instrument is a bit like "cheating" and not present in the original NES, but included for convenience: in the original NES, you couldn't adjust both the envelope and the volume of a channel at the same time. So if you have an envelope active, the vol knob should always be at 15, if you strive for authenticity. Also, the triangle channel (ch 3) had no volume control - but, there was a certain hardware glitch that could be abused to gain some amount of volume control for it, which involved using the (absent from this instrument) 5th channel, which was the sample-playback channel (yes, NES had primitive sampling capabilities!) I didn't include the 5th channel because it's a bit redundant and you can get basically the same thing by running a sample through a bitcrusher... Anyway, to make things simpler, I just included a volume knob for it regardless. It probably isn't accurate to how the ch3 volume adjustment worked in the NES, though. That's all for now - I'll write more when I finish the synth... ------------------------------------------------------------------------------ Time is money. Stop wasting it! Get your web API in 5 minutes. www.restlet.com/download http://p.sf.net/sfu/restlet _______________________________________________ LMMS-devel mailing list LMMS-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lmms-devel
