Thanks, always been curious about those. Would love it.
Obviously, there's been a big hiatus on the project. I was busy with University, then I was sick for the 6 weeks after (not COVID, apparently). I've only recently got time again, and University has started up. I'll still try and update the project document and start listening to people's recordings soon, though.
May be little late, but i have modded SC88 Pro with digital output. That may be point of interest for the records.
Super interested. Not too late, the project has had 5 months of delays with the year that I have had. Do you have any other SC devices? 😀
Also, did you do the mod yourself? If so, people would be interested in documentation.
DO mods would be a good thing to have for the classic modules, to make recording better.
Yes, it's my mod i did in 2016, last month i have published it with all details at github. https://github.com/vetal-esher/18bit-DIT
For Rolands , i have SC88 Pro and MT-32(old), and now i'm working on digital mod for MT-32, we all know that it is possible but without reverb since it is in chain after DAC.
edit: every module need schematic first, to see if i2s SDATA is L/R combined stream. The mod is theoretically compatible with all modules which generates 16-24bit LR combined i2s.
edit: as i digged deep into schematic, i found that reverb is also digital and it's all multiplexed in one stream which then demultiplexed and mixed in analog. This requires ton of logic, but i think there's a way to do all in fpga model. Lot of work.
Thankyou for publishing (and linking to) it! <3 Great work.
If I sent you modules, would you be able to convert them?
I can also send you MT-32 old and MT-32 new schematics, if you need them. And probably most SC ones.
It seems to be impossible at this time, since i'm in Russia and the postal services are unreliable due to great logistics problem (due to f..ing war started by mad creaker). You can order the pcb's at PCBway or JLCPCB, and AK4103 is found at Aliexpress. Gerbers for pcb production are on github. Then ask a friend or local radiomaster to solder it for you. I'll help with finding where to solder the i2s lines.
Thanks a lot. Sorry to hear about what's happening in your part of the world, hope you're OK and not too affected.
I finally have my recordings ready! Sorry for this late, huge submission:
https://1drv.ms/u/s!AsMEa3-pVaQOgQhZW2lDxqrRbtts?e=ogCHdj
First, any recordings with "MIDI" in the filename denote use of actual 5-pin MIDI input when USB virtual MIDI would have been possible. Since the 5-pin MIDI connector should be all I used, the addition to the name was to be sure it was obvious, on those devices that had a choice. That was done to avoid ground loop noise in most cases, and in the digital audio cases, just because it was already connected for the prior analog recordings. Everything without the "MIDI" naming only has the 5-pin MIDI input, so you shouldn't be under the false impression that I had used another method of connection.
The only recordings I have not done at 96kHz are 44100Hz for the MU2000 EX from its SPDIF toslink (because that's its required rate) to the UA-4FX with its optical input. Those files are named with "44100Hz" to make them obvious. On the topic of the MU2000 EX, I set it's UTIL->SYS->DigiOutGain to "+12". It doesn't clip with that and improves the dynamic range. I chose only to record it with the UTIL->SYS->Map set to "MU100Native", but doubt there's a difference for its GS sounds. I could be wrong about that, though. If you want two more sets of recordings (analog and digital), or another six songs, I will record them for you upon request.
The SC-8850 recordings all used the STUDIO-CAPTURE inputs 13 & 14 attenuated at -10dBu, since it's normally always connected there. I had tried other inputs on the STUDIO-CAPTURE, as well as on the UA-4FX because the SC-8850 MAP recording of Elise always showed significant 180-degree off-axis piano sounds between the right and left channels. That leads me to believe Roland purposefully used that to create the stereo piano effect, so I eventually gave up and sent you a recording from the STUDIO-CAPTURE inputs 13 & 14 attenuated at -10dBu for the SC-8850 MAP too. The SC-8850's other MAPs only had reinforcing waves for the Elise piano.
The SC-8820 recordings all used the STUDIO-CAPTURE inputs 13 & 14 attenuated at -10dBu.
The analog MU2000 EX recordings all used the STUDIO-CAPTURE inputs 15 & 16 attenuated at -10dBu, since it's always connected there.
The analog INTEGRA-7 recordings all used the STUDIO-CAPTURE inputs 5 & 6 with SENS set to 5.5dB, since it's always connected there.
The INTEGRA-7 USB Digital Audio recordings (named with "USB-DA") were done at 96kHz, though MIDI was played through its real MIDI input (due to already having been setup for the cleanest possible analog recording). I did an additional three songs, in which I hoped to show you what my patched QMIDIRoute with the INTEGRA-7 SuperNATURAL remapping qmr file will do, although they aren't what your after as they're basically realtime-edited versions of your files. These extras are USB Digital Audio recordings of the INTEGRA-7 with "QMIDIRoute" in their filenames. EDIT: Here's a link to the INTEGRA-7 MIDI files created by sequencing QMIDIRoute's output for your examination: https://1drv.ms/u/s!AsMEa3-pVaQOgQmOh0BadHMGJXtw?e=VAJncL
The SC-33 recordings all used the STUDIO-CAPTURE inputs 13 & 14 attenuated at -10dBu.
The SC-55mkII recordings all used the STUDIO-CAPTURE inputs 13 & 14 attenuated at -10dBu.
The PMA-5 recordings all used the STUDIO-CAPTURE inputs 13 & 14 attenuated at -10dBu. For them, I also turned on "MIDI Update" on the PMA-5's UI, just in case something wouldn't be set otherwise. For the "Batteries" recordings I used six Panasonic Eneloop AA batteries, and for the recordings without the name "Batteries", I used the BOSS R3W009-200 PSU it came with.
The SCP-55 recordings all used the STUDIO-CAPTURE inputs 1 & 2 with SENS set to 20dB. At first, in pure DOS I attempted GSPlay, but it had timing discrepancies, so I eventually installed it on a windows laptop. That allowed me to use my consistently-timed Linux MIDI player (audacious) by connecting the MCB-3's MIDI IN to the SCP-55 Synth virtually with midiOX. The MIDI IN is probably only intended for sequencing otherwise. The MCB-3's Phones output from DOS (with no mixer software) is quieter than all other modules turned up to max, and with windows 98 it's still quieter, but I was able to adjust volumes. A difficulty with the volume set too high in windows is obvious internal clipping. To work around that, I set "Playback" volume for "Synthesizer" to about 2/3 up with main "Volume" at max, and the "MIDI" screen's "MIDI Volume, "Chorus", and "Reverb" each at max. The other MCB-3 audio inputs were muted. It didn't make much difference, but I recorded using the laptop's charger connected (named "Laptop-PSU"), and then without that and only the laptop's battery power (named "Laptop-Battery").
The CM-500 recordings all used the STUDIO-CAPTURE inputs 1 & 2 with SENS set to 10.5dB, and the CM-500 was set to Mode "C", for GS ("CM-300") of course.
Any interest in recording the sounds of a CM-32LN before I mod it?
RiverBoa wrote on 2022-09-17, 20:13:Any interest in recording the sounds of a CM-32LN before I mod it?
What kind of mod do you mean?
esher wrote on 2022-09-17, 20:20:RiverBoa wrote on 2022-09-17, 20:13:Any interest in recording the sounds of a CM-32LN before I mod it?
What kind of mod do you mean?
There's a fix for the Vibrato issue being too fast for the CM-500 and CM-32LN that reprograms the control rom and replaces an oscillator. So it changes the sound to what's expected from the CM-32L.
RiverBoa wrote on 2022-09-17, 20:24:esher wrote on 2022-09-17, 20:20:RiverBoa wrote on 2022-09-17, 20:13:Any interest in recording the sounds of a CM-32LN before I mod it?
What kind of mod do you mean?
There's a fix for the Vibrato issue being too fast for the CM-500 and CM-32LN that reprograms the control rom and replaces an oscillator. So it changes the sound to what's expected from the CM-32L.
I think in any way it's worth to be recorded before and after, with proper MIDI file where vibrato changes can be clearly heard.
RiverBoa wrote on 2022-09-17, 20:24:esher wrote on 2022-09-17, 20:20:RiverBoa wrote on 2022-09-17, 20:13:Any interest in recording the sounds of a CM-32LN before I mod it?
What kind of mod do you mean?
There's a fix for the Vibrato issue being too fast for the CM-500 and CM-32LN that reprograms the control rom and replaces an oscillator. So it changes the sound to what's expected from the CM-32L.
I am planning on modifying my CM-500 for the correct vibrato, but I'm tempted to put the oscillator on a switch, and keep the original, unpatched ROM handily inside, just in case I ever want to revert it for posterity. I agree that future explorations could be hampered if we all mod them without making it reversible. In either case, and you probably know this, but the CM-500 and CM-32LN's vibrato issue only affects the LA synthesis, so it's not really related to the Sound Canvas Comparison Project.
The songs that stick out in my mind are already documented in youtube videos, but youtube audio quality isn't as good as the 96kHz lossless recordings we're attempting here, so I'd say yes, by all means! The songs off the top of my head are Monkey Island 2's intro, and Space Quest 3's intro, but there must be others. 😀
jaffa225man wrote on 2022-09-17, 23:01:RiverBoa wrote on 2022-09-17, 20:24:esher wrote on 2022-09-17, 20:20:What kind of mod do you mean?
There's a fix for the Vibrato issue being too fast for the CM-500 and CM-32LN that reprograms the control rom and replaces an oscillator. So it changes the sound to what's expected from the CM-32L.
I am planning on modifying my CM-500 for the correct vibrato, but I'm tempted to put the oscillator on a switch, and keep the original, unpatched ROM handily inside, just in case I ever want to revert it for posterity. I agree that future explorations could be hampered if we all mod them without making it reversible. In either case, and you probably know this, but the CM-500 and CM-32LN's vibrato issue only affects the LA synthesis, so it's not really related to the Sound Canvas Comparison Project.
The songs that stick out in my mind are already documented in youtube videos, but youtube audio quality isn't as good as the 96kHz lossless recordings we're attempting here, so I'd say yes, by all means! The songs off the top of my head are Monkey Island 2's intro, and Space Quest 3's intro, but there must be others. 😀
This is worth another master topic. For the LA-synthe chip, i'm working on Digital Output mod which can be applied to all Roland LA synths (i think). There's no so much details at this moment, but in 2-3 months i think i will get some results from prototype (i have no much free time these days).
Awesome info. While not related to the topic it's still very much of interest!!
Yes, please record your CM-32LN, I'd love to hear how it sounds with some classic tracks like SQ3 intro (despite and also because of the vibrato), but really anything. Would be good to hear it do some CM SFX games as well, like TFX, Larry 5, etc.
Hello, new guy here 😀
Does anyone know what soundset/rom is in the Roland FG-1000, it has a soundcanavas soundset i think.
oldinternet wrote on 2022-10-05, 05:27:Hello, new guy here 😀
Does anyone know what soundset/rom is in the Roland FG-1000, it has a soundcanavas soundset i think.
I didn't know it existed until you posted, but according to its manual, page 29 says it's compatible with "Roland's new GS format", and page 36 says Control Change 0/MSB is recognized which is used to set variations.
That sounds good, but the Instrument Table beginning on page 34 shows a only a tiny subset, of non-zero (variation) CC0 sounds, than even the SC-33 has (which itself is missing the SC-55's MT-32 variations on CC0 values of 127). So it probably has mostly the same sounds that it lists, but not a complete set by any means.
It also has an extra sound effect I don't think anything else I've seen has: on PC# 125, and variation 6, there's one called "Buzzer" that I would be very interested to hear! Even more interesting are all the internal songs, of which only one or two appear to have been included as demos with other sound modules.
Very weird. It appears to be a very, very early GS device, and as such isn't really "true" GS. Although it seems to have come out a few months after the SC-55, May 91 versus Feb 91, so I am at a loss as to how this could be billed as "GS" despite having 50 or so less GS instruments (61 GS variation tones in the SC-55 mk I, vs only 9 in the FG-1000). So for me looking back, it's basically a SC-7 map (GM only), plus 9 "bonus" GS tones.
Given it has 24-note polyphony, I'm guessing it uses the same parts as the SC-55 but has lower Wave ROM (or maybe the ROM is used for all the onboard music). Couldn't find a service manual and unfamiliar with any devices this would be equivalent to. It even looks kind of like a MT-32!
Incidentally, it also has a unique sound "Buzzer", as jaffa mentioned. It would appear the reason for this is to act like an alarm clock, see page 32 of the User Manual where a "Buzzer" can be used as an alarm sound or "Song".
One other unexplained aspect goes to the "Option Card" slot, which cryptically gets no mention in the manual until the Specifications, where it just refers you to a dealer to ask about what's available! Given Google shows nothing, either the cards were never made, were only released in Japan, or are purely memory card types that allow saving or add song data (see also the AR-1000, which is a similar but different device from the same "family", uses SRAM cards, another possibility).
It's not MT, as it's not based on LA through GM map. I's definitely early version of SC/GS, alpha version 😀
