For the former, you can use the PSG logger and use psg2mid, for the latter, someone should do some work... (It's in our feature requeste tracker, I think.) But seriously: it takes quite some work, also on architecture to separate channels of individual chips, so it may take a while before this is done.

Quote:

Wolf, not everyone can write down notes while listening to music...

Yes. I, for example, almost not have 'ear for music' (I don't know, how translate this term to english) - I can't determine, which note is playing, only can somehow determine intervals, and compare it with notes, which I take on guitar or synth. But in many modern sound editors frequency analysers exists. With this feature and zooming of recorded wave it came piece of cake to write down exact notes, slides, volumes, fast arpeggios and licks, and so on (also you can use midi sequencer with audio files support, and place recorder channel with you midi track to play them together in sync, to compare it and fix differences). So it can be solution for people like me.

manuel: just a silly idea. I don't know how exactly the triggering of a soundcard works, but I estimate that there's a freq register, a volume register, a channelnumber and some more stuff (FM registers 'n stuff), at *this* point you have seperate channels or busses, and that FM algo mixes them together into one operator iirc. What about ditching the one single FM-Pac (or any chip) and select 6 or 9 FM-PACs in the extentions list (provided you have 6 or 9 FM-PACs to pick)! Then you divide these channel registers (which you have!) between each FM-Pac device, you use one channel per device, and you log each device.

Quote:

Wolf, not everyone can write down notes while listening to music...

Yes, everybody can... If you use KSS and diskwrite every channel to a wav file, you open a wav in a sampleeditor... You zoom in far enough to actually see the tones... Then you mark that one tone you can't hear good enough and play it looped... In the meanwhile you strike every near enough tone in moonblaster's wavemenu to see if they match

I used this to do some fast arpeggio kinda thingies in Konami remakes and it really does the trick... You can make a complete 1:1 copy of each track that way (set aside the soundcolors)...

Quote:

at *this* point you have seperate channels or busses, and that FM algo mixes them together into one operator iirc.

The other way around. The FM algo mixes operator outputs to produce one FM channel...

Quote:

What about ditching the one single FM-Pac (or any chip) and select 6 or 9 FM-PACs in the extentions list (provided you have 6 or 9 FM-PACs to pick)! Then you divide these channel registers (which you have!) between each FM-Pac device, you use one channel per device, and you log each device.

That won't help really.
First, in the end, openMSX generates just one stereo output track, with all devices mixed. So splitting one FM-PAC with 6 (or 9) channels into 6 (or 9) 1-channel FM-PACs will have no effect at all.
Second, there is a point in the emulation code where the outputs of all channels of a device are mixed together to produce the final output wave for this device: that would be the point to generate the separate output files... Can't be too hard, actually

I'm not meaning to sound rude to anyone, but training your ears is not too hard either and will get you much more in return on the long run as a musician. Maybe for some fast difficult licks, you could just time-stretch that part to make it easier but other than that I see no reason why not to train your ears.

Neither I or J-War have difficulties finding out what notes are used (J-War is a band leader in RNFF-land, and I'm a TV/film-composer (HKU/KMT/MTt-trained), while it doesn't say everything, at least it says we're no n00bs or something ). I've never seen something that plays the Dragonslayer6 intro channel by channel (e.g. with mutes), yet I was perfectly able to find out all notes. While I didn't recheck each note in the end, I'm pretty sure that *if* there are errors then it's less than 0.10% and not a relevant error anyway. Also -but it might be before you activily joined MRC- J-War once relased promos (short fragments) of about all MG2 tunes.. which was years ago, and I'm not sure whether he used something like KSS for that!

The problem is the time it takes. This whole debate is to make it easier/faster, not to make an impossible thing possible (because it isn't impossible). Remaking one song is no problem, but J-War and I (and perhaps others too) eventually wish to remake a whole game. Do you know how many tunes there are in the Xak trilogy alone? Also, Illusion City, Shalom, SD-Snatcher, Metal Gear 2 etc. are all games with an enormous amount of tunes, and they're not even simple ones (endboss SD-Snatcher anyone? ) So a bit of automation would really come in handy, channel mutes and automatic diskwriting, or perhaps even output to a midifile, would save months of boring/dull/uninspiring work.

erikd, generally you right. But side effect of 'training ears' we already have: more than 50% midi covers of VG music in internet contains very noticeable errors. Sometime even hard to recognize, what was covered. So, I think, better train ears before work on cover, which plan to be published, then train in that process.

btw, aside from being easy for us, there are also people around who could really *need* seperate channels because their own hearing skills won't be enough. So, the importance of a feature like this (seperate channel diskwriting) is important, I think .

Mind you, I'm not implying that some people in particular are lesser musicians or anything. I know some impressive stuff has been produced!
Without trying to sound arrogant, I can't see myself saving any time listening and analyzing separated channels instead of a mix (unless the mix is very unclear, maybe). I suppose a lot of my 'musical hearing' depends on hearing harmonies as opposed to separate lines only.
Of course I can't have anything against a feature to output separated channels if it helps other people, if you're willing to implement it. My point was really that in my experience it helps much more to train one's ears, since you won't have to depend on extra features like that.

Listening to harmonies could be a way to reproduce a tune (it was often done like that in the active mainscene years), but would you get each and every note accurately replicated then?? As far as I know J-War and myself, we don't really go for sound-somewhat-alikes but for 1:1 copies. Because often enough people won't truly remember one certain note somewhere, but they will notice it when it's gone. So in our particular case we'll be studying/copying individal lines, not global harmonies 'n stuff.

I see your point. I'm not saying I'm hearing *only* harmonies, but it's quicker for me to analyze them from the whole thing instead of separated channels but of course your milage may vary.
And then again, I'm not really into making 1:1 copies so there's probably a difference there too.
So what if the original's voicing of a harmony can be improved (always a matter of taste of course), would you go for the improved voicing or always stick with the original? For example I can imagine the original composer went through some hoops to squeeze his ideas in the limited amount of available voices.

I try to stick to the fundamentals of a tune. And if something 's added then they function as decoration without changing the fundamentals.. for example in Dragonslayer 6 I stuck to the original notes, but in the end the project I was multi-tracking had over 30 (audio-)tracks with many layers 'n little details. DS6 was clearly a tune based on 9 channels, Metal Gear 1 then again was based on 3 channels. As far as I'm concerned this MG1 tune is again quite close to the original, not extremely different, just more meat and more whipped cream. ^_^

Ok, I can see your point. I haven't done any computer music remakes, so I guess I can't judge what features are useful and which aren't. The last couple years I've mostly done just my own music, playing live instruments, a bit of sequencing and working with singers and rappers. I suppose that's a whole different way of working

Those tracks sound truly awesome, btw

More in the download database