First time making a thread here, hi. I've been following Sonic content for a while and one of the most interesting things to me involves the musical development and history for most of the games of the series. One of the ones that have stood out to me the most has got to be the Japanese soundtrack for Sonic CD, more specifically the unused loops. I make this thread so many can clear up some confusion i'm having, such as the discovery for these loops and their extraction and restoration. Run through the history of how these loops came to light and start some discussion, i'm intrigued to hear what you all have to say.
In the early 00s, ICEknight discovered them when digging through the files of the 712 prototype. They're in the same format as the samples used in the past tracks, so he hacked it to play that sample when the level starts. Of course, they're not terribly high quality, they're 8-bit PCM and I don't remember what the sample rate was but it sure as hell wasn't redbook's 44.1khz. I suspect they're from an older iteration of the OST, as some, such as Metallic Madness present, are significantly different tempos. I do wonder if they were really meant to be the ends of the songs or if they were supposed to be intros, since the MM present sample sounds a lot like the opening of the past theme. There are a few ways one could attempt to clean up and enhance their quality. The lower sample rate means that higher frequencies are cut off, so those would have recreated somehow. A cheap trick is to pitch double a copy of the sample, highpass it where the old cutoff was, and EQ it until it resembles the redbook tracks in a spectrogram. Alternatively, portions very similar to the loops in the redbook tracks could be sourced for those higher frequencies. Not really sure what to do about the 8-bit quantization, however. It's mostly a problem at low amplitudes, as you have a lot less resolution to work with by that point. If it's strictly quantized and not dithered in any way, I suppose the waveform could be smoothed over with interpolation, but that would also kill off any intended higher frequencies still present. If there's dithering, it might even help to resample those quiet portions to half of the sample rate and then resample it back.
I find placements of the loops real funny when you try to traverse them. They play best for both cases, outro and intros, but they also work in certain cases like adding them at the end of the track but letting them loop back to a proposed loop time that isn't just the intro of the tracks. I've linked an example with Palmtree Panic to show you what I mean. file:///C:/Users/slebo/Downloads/Palmtree_example.mp3 Scratch the link i'll send it through as video. I pray for the day that Sonic Retro will let you reply using .mp3 formats. A while ago I actually made a proposal for Metallic Madness's Beta Loop that fixes the BPM to be consistent. Rendering it to sound like the Past's intro.
FYI, you can't link to a file which is on your hard drive, you need to host it somewhere online first.
Did the Mega CD ever support seeking CD audio tracks at a further point than its opening? I recall Soul Star had programming meant for synchronising CD audio tracks with enemy patterns according to Sarah Avory. If so that might explain why it never got used (deliberate seeking might've require significant amount of electromechanical timing)
Calling a precise location in minutes:seconds:frames (XX:XX:XX) probably isn't going to work too well. Each CD audio frame is about 588 samples or 1/75 a second and then there's read offsets of drives and pressing offsets of CDs that can really differ. As you add and remove data you'll be readjusting this point a lot. It just isn't worth it and that's before you factor in seek times varying between drives. You're better off using files in the data track with samples you want to play by preloading them into memory if you want to be precise or using the sound chips for that stuff. Speaking of data tracks, it's the only portion of a CD where things have to be precise, so sync words (a string of Fs (hex)) and addresses (minutes seconds frames) as well the current mode are added into the start of each frame where there was previously just audio.
This is precisely why I've always been iffy on these being intended for loops, because they couldn't be precise enough. You'd still hear gaps or they'd cut off or overlap. I guess, perhaps, that could be the reason they weren't used in the first place, but I have other reasons for being slightly doubtful. Some of the clips would throw off the time signature of the songs, there would be an extra measure in the final loop in some songs. But I'm an idiot when it comes to audio, so I could be completely off. Some of the loops, to me, sound like they could have been end-of-act jingles. Remember, the You Say demo had no actual end of act song.
Are you referring to the openings of the tracks that are for some reason not released on the OST? I say "for some reason" because in particular the opening of Tidal Tempest here is referenced in the Past version of the track, which most people wouldn't know because it never gets played or released anywhere.
Good find on Devon's part, I think that basically settles how the PCM samples were meant to be used. Devon posted a video of how the code works (changed so it checks the correct status flag) here if anyone's interested. There's actually a released Sega CD game that does something similar (although not to cover up the track looping). The version of Flicky on Game no Kanzume Vol. 1 plays a PCM sample for the gameplay theme's intro, then starts playing the gameplay theme from CD. It works pretty well, so I think it's a shame they didn't finish the feature for Sonic CD.
This is a great excuse to why D.A. Garden (Originally Dubious Depths) never got an unused loop, since the track was still kept that far into development despite Dubious being jilted.
I have such a bad ear for music, that I never made this connection before. This is so cool. Oh wow, I'd love to see this expanded and built into a working prototype without the looping error, then tried on real hardware to see if what I said above about it probably not being actually seamless is true. That's a good example of what I was talking about, you can hear a gap after the loop. Was this on real hardware? EDIT: Ah, it was! Thanks so much for an actual example!
Another thing I noticed that's similar to this is that the 0.02 proto version of Palmtree Panic past's intro sounds alot like Palmtree Panic Present and Good Future's unused loops.
I feel like it would almost be easier to discover the original samples and instruments used in the redbook audio than to improve the existing loops.
