Last weekend, I made a day and hit a bunch of pawn shops, specifically looking for keyboards.  I guess collecting CDs was getting too hard, and collecting CD players was too infrequent, so keyboards is now what I collect.  I had two a couple weeks ago, now I have five.  I’ll figure out something.

So, to keep that story short, I made a deal at the first place I went and the rest of the day was pretty crappy.  For some odd reason, my phone could not keep a GPS signal, so my trip was cut short.  No idea what was up with the GPS, but I hope it’s not a regular happening.

So anyway, at this first shop, they had a keyboard out front – a Roland D70, which is a 76 key synth from 1990.  Not bad.  The original price was $720 and it was marked down to $450.  ehhh, not that great.  But there was a sign nearby that said anything on that table, make an offer. Hmmm.  I did a quick price check and the D70 sells for about $500.  Ok, let’s at least check it out.

I ask to try it and the first thing I see is that the MIDI thru jack is ripped out.  I have no idea how something like that happens.  Not a deal breaker because I wouldn’t need that port, but it is a negotiation point.  It powers up and I start testing the keys.  To my amazement, some of the keys don’t work.  And when I say they don’t work, I don’t mean they don’t make a sound when you press them, I’m saying you could not physically push them down.  Five keys had that problem, all black keys.

The store got a little busy right then, so I had an extended period to consider what level of effort a repair would be.  Absolute worst case, find a dead donor board and swap the keybed.  The electronics seemed fine, the issue was only mechanical.  Time to barter.

Know this about me:  I don’t haggle.  I like to be a people-pleaser.  I don’t like the discomfort of potentially insulting the person who is offering me a service or product.  So I continually tell myself, I don’t need this board.  There’s no reason not to walk away.  I set my price at $250.  Now remember, this is a pawn shop, so it’s likely the person pawning it only got like $100 at most for it, so my price is still giving them a profit.  However, that price was lower than their lowest time-based discount price was.

The salesman came back and I explained the problem and the level of effort I’d have to take to see if it’s even repairable and told him I could only offer $250.  He immediately said, no, I already have $400 into it.  Well, that was quick.  So I was like, ok, I can’t take it.  But I can tell you there’s a repair shop nearby that could fix it for you if you want to get its full value.  I know a pawn shop doesn’t want to sell good stuff; they want to turn over product with as little hassle as possible.  Fuck, they don’t even clean things they put on the floor.  Such a simple thing to increase the value, and they don’t.

He changes his tune quickly, "let me ask the manager."  Yup, just like a car dealership.  However, unlike a dealership, he came back and said, manager says he’ll take $250.  I didn’t have to fight wave after wave of bosses to get the deal.  So, deal done and back home to see what’s up with this thing.

To get to the point, this keyboard had what I’d heard about in repair videos but had never seen yet: the Roland Red Glue.  This glue, in keyboards from 1985-1990, would melt in heat and humidity, loosening the key weights and running into the internals.  The five broken black keys all had their weights fall and that is why they could not be pushed down.  It’s quite funny to me to think that this board might have originally worked well, but because pawn shops suck, they might have left the A/C off at night to save a couple bucks and caused this problem all on their own.  I love that sort of justice.  But anyway, the problem was mine now.

I did some research and found that the usual fix is to soak the keys in drain cleaner to dissolve the red glue and then reattach the weights with epoxy glue.  The recommendation was to use a drain cleaner with lye.  Well, guess what, you can’t get that anymore.  Why?  Goddamn methheads.  Lye is used in making meth.  So I picked up some other cleaner that had some of the chemicals that are associated with lye and hoped that would do the job.

The keys soaked overnight and while it did appear to dissolve all the running glue and freed the weights from the five black keys that had their weights fall, the other key weights were still rock solid.  Solid enough that I broke the plastic on a key trying to pry the weight out.

So the next day, I spent a lot of time scrubbing the keys, to get whatever glue was left (which had turned black) and planned my next move.  In one video, the person said the glue was susceptible to high pH and tested his soaking solution before starting to show it was high pH.  Well, I can make a high pH solution with some pool chemicals.  So I took the broken key as my new sacrificial tester, bought some alkalinity increaser from the pool store, and soaked the key in a solution with a pH off the charts of my pool test strips.  After a few hours, the key weight was as firmly attached as ever.

So at this point, I think I’m just going to fix the five black key weights, buy a replacement key for the one I broke, and put it all back together.  Some part of me says it’s leaving a job unfinished, but another part of me is saying, those other key weights are on there.  And I’m not leaving my keyboards in a non-climate controlled environment anyway, so they should remain solid.  I dissolved all the excess glue that had seeped out, so the only glue left is what is behind the weight.

The replacement key arrived quicker than I expected.  And immediately upon opening, I check it.  Red Glue.  Ok, off into a drain clearer bath overnight.  The next day, I scrubbed the dissolved glue off and installed the key.  I had been waiting for this and had installed every other key in advance.  I put everything back together and powered the board up.

No sound.  Further, the display was completely garbled.  Actually, this is ok.  I’ve heard that you might need to do a memory erase and then go through a tedious process of transmitting data to the device to restore it.  So I download the data file, get the utility program and try.  And try again, and again.  We’re not seeing any success here.  I read a couple more posts about the process and someone comments on turning off a memory protect switch.  Oh yeah, that would help.  A few more tries.  More reading.  Eventually what worked for me was navigating to a system menu, enabling sysex receive, and setting the device id to 17 (why I don’t know).  But then I had sound.

Unfortunately, some of the keys weren’t responsive.  I was too optimistic about their integrity and I probably shouldn’t have been.  So, everything comes apart again and I disassemble the keybed completely again.  Now I have to clean the membrane contacts and pads with alcohol, which wasn’t really as bad as I thought it might me.  I put on some music and went at it.  Two albums later, I was done and reassembling everything.

I had sound and now I had keys that work.  Some of the black keys were more sensitive than others, but that’s livable.  I can tweak any obnoxiously incorrect velocity in the sequencer if I need to.  And that actually wraps up the repair and restore of this device.

In the meantime, I’m buying another keyboard next weekend, which might need some attention, and also the keyboard that started this buying frenzy, the Equinox, needs some care.  The pitch wheel is wonky and I think some of the faders are dirty and spamming the bus.  And I know the battery is low on that, too.  So, no shortage of future projects, let’s hope they don’t become as involved as this one.

Almost exactly 3 years ago, I was mulling over how to apply star ratings to albums so I could sort of make some sense of my music collection.  I actually never went through with it, but now I’m considering tackling something even bigger – applying ratings to the songs.

The big motivator here is building playlists.  When I first started with Plex, I had a vision of kind of a radio station feel to the whole thing, and appropriately, I made playlists that sounded like radio stations, or more like channels on XM.  And that worked pretty well for a while.  At some point, I can’t remember what happened, but I lost all my playlists and had to recreate them.  Ugh.

I remade some of my more used playlists and then I started reconsidering the others and broke them up into decades and sometimes by genre.  The problem there was sometimes I didn’t want to listen to all one decade of music.  So I made one massive playlist of all the singles in all decades.  And that one has been pretty much my go-to when I just need background music.  Realistically, I’m only playing 3 different playlists, but maybe with better metadata, that can change.

One big hurdle I’m facing is that there seems to be no way to get the ratings in and out of Plex from the files.  So whatever ratings I do, I would have to duplicate the effort in both Plex and the files.  And putting the ratings in the files has no benefit because Plex can’t import them (yet).  That sort of makes my Plex library more fragile, since there’s data in there I can’t just lose without a lot of effort lost as well.

Anyway, to disregard that problem for the moment, a bigger problem was how to efficiently get all that data?  Let’s step back even a little further, what exactly am I planning on with these ratings?

That point was something I dwelled on for a while.  I went around on it for a little while.  I considered using Plex collections, but those are only for albums, not songs.  Songs have Plex tags, and I looked into using them.  I thought maybe tags like "Single", "Top40", "Top10", and "#1" might be good.  But my inner software architect was displeased.  Initially, you would assume that "Top10" would also mean the track was "Top40" and also "Single" because of the inclusive nature.  So you’d only need one tag per song.  But that’s going to make the filters (queries) really messy because you have to put that logic into the filter.  If you want a list of "Singles" you have to also include "Top40, "Top10", etc.  The alterative is to use all appropriate tags where needed, so a #1 song would have all four tags on it.  That’s not pleasant either.  ugh.

So going back to the thoughts I had in my earlier post, what if I just made the star rating mean whatever I want it to mean.  So I quickly wrote down a scale:

1 star – Single
2 stars – Top 100
3 stars – Top 40
4 stars – Top 10
5 stars – #1

I think that’s usable.  And before I change my mind on it, how about that efficiency concern, now?  I had almost 2000 albums to go through and determine which songs were released as singles and what their position was on the charts.  And which chart, at that?

So obviously there’s a bunch of compromises that need to be made in this process.  The first was determining what stars mean.  The next will be starting with one source of data.  I considered I could use Wikipedia and look up each album to get the singles and chart positions, but that is woefully underpopulated, so I can’t use that for my primary source.  Billboard does this stuff for a living, I could try them.

As it turned out, someone had made a downloadable dataset of all the songs and chart positions on Billboard’s Hot 100 charts up to 2020 (far more than I needed).  After a quick download and import into a SQL database for easy querying, I felt I was closer.  While I was going to miss out on any 1-star entries, the dataset of the Hot 100 would cover 2-5 star entries, and I could backfill later.

At this particular point, I’m not able to do any automation of the rating import, because I can’t figure out how Plex stores the rating in their database.  I manually changed some things and didn’t see any data changes in the database, so initially, it’s going to be manual entry.  And then I can start building playlists based on singles and chart position, maybe mixing genre and release year into it.  Hopefully that gets me somewhere pretty good.

In my previous post, I talked about music and "remastering" some of my old music.  Where I left off is that I was trying to redo some old keyboard pieces that used the Yamaha SW1000XG.  I bought a Yamaha MU80 as a replacement and that didn’t have the same sounds, so I bought an MU100.  To my surprise, again, not the same sounds.  So while I lick my $400 wounds and decide how I want to go from here, I made progressions on another musical concept.

I had written some guitar tunes a long while back, before I became a more aware and less offensive person.  As fortune would have it at that time, my voice could not cope with the style of singing required for the songs, so all I had recorded was the music.  There are some guide vocals in some songs, which are cringey to say the least.  It’s for the best they stay muted.  But anyway, the recording of the instrument parts left a bit to be desired as well, so I set myself to it to clean those up.

The first issue, which is just like the Yamaha issue, is trying to find the effects that I used when recording the tracks.  After many failed attempts to match up the guitar effects plugins, I gave up and chose new effect patches for the tracks.  They don’t sound the same as the originals, but no one’s heard the originals, so whatever.

The next step was cleanup.  In the original recording, there was a major problem with bleedthough in the mixer I owned, so a lot of tracks have a background noise of the click track.  Through a lot of clever editing and some aggressive fadeouts, I was able to hide any noticeable clicks.  As I made those edits, I determined how to best organize the project for mixdown.  This led to a solution of having the midi drum track span the full length of the song, including pre-silence and fadeout.  That way I could set the locators (which determine what part to mixdown) to the selected drum track and be good to go.

The step after that was mixing, burning, and testing the tracks in CD players: home and car, plus through computer speakers.  I have a spindle of 100 CDRs that I never thought I’d use.  I’m going to use them now.  As I did my tests, I adjusted track times, in cases where the mix cut off too quickly or in some cases, didn’t leave enough lead space for a CD player to audibly start the track immediately.  That was weird: that even if you want a track to start absolutely immediately, you still need a small bit of silence at the beginning otherwise it sort of quickly fades in.

And that was actually a problem, because I had two tracks that segued into one another – I couldn’t have a silent gap between them.  This issue was compounded by the software I was using to write the CDs.  Coming up with a resolution involved another step and more software.  To solve the gapless issue, I had to create a CUE sheet, which would identify the exact placement of the track boundaries on the disc.  And instead of burning multiple audio files, you burn one file that contains the whole CD audio.  The CUE file points to sections in that one audio file.

So now I have to create a single file of the entire album’s audio.  And this forced me to do the proper step of CD mastering.  In this step you work with all the mixed tracks together at once and make them sound cohesive.  And at the same time, you work out the timing of the tracks and the gaps between them.  It was something I was aware of in my listening tests – that some tracks needed volume adjustments – and the mastering process gave me that opportunity to balance everything out.  It’s something I expect to do in future projects.

So I’m up to test disc #6 now, which contains the level-matched tracks and also the gapless track changes where needed thanks to the CUE file.  When I burnt the CD using a new utility that utilized CUE files, I noticed some mentions of CD-TEXT being written, which allows CD players to pick up and display the track title.  I haven’t been able to see that in any players I’ve tried yet, but that’s another target to hit for future test versions.

It was a while ago I made a post with a lot of reservations.  It was regarding hauling out my music stuff and getting back into music.  And my reservations at the time were that I wasn’t going to get very far with my initiative because I’d been through the process many times in the past and each time ended up packing everything up and putting it away with nothing to show for the effort.

And, well, this is somewhat the same in that it has not been too productive.  I developed one idea I’d had for many years, but haven’t gotten enough to really make something concrete.  And while that was developing, I also worked on getting the recording station all set up.  I bought a new micro computer, monitor, and monitor stand.  I installed and set up my old Cubase software (which is way behind the times and yet more than I’ll ever need).  Although that’s all ready to go, I haven’t really started anything.

I knew I would have an uphill battle getting my physical abilities back since I hadn’t played in such a long time.  To my surprise, my capability came back faster than expected.  However, I plateaued quickly and my stamina was much diminished, so that was a little discouraging.

Instead of giving up, I decided to pivot a little bit and try to get some inspiration and relearn some engineering technique.  I have a lot of old music that exists in MP3 format.  It should be in FLAC format to be of the best quality.  Additionally, some of those songs need a little improvement.  One in particular has the beginning sort of cut off and I have no idea why I accepted that at the time.  Since I have the "source files" for the songs, I should "remaster" them in a sense and bring them up to a standard where I won’t need to worry about quality anymore.

What does that entail?  Well, I have to recreate the recording setup I had back when I recorded them.  This is not a trivial matter for me or for anyone who has ever attempted something like this.  While my case is relatively simple, imagine an actual professional musician trying to track down vintage synthesizers and recreating the patches that were used on each track.  It highlights the need for documentation in a studio.  I admit, I didn’t do hardly any – I never really gave it any thought.  So when I loaded up one of my old files and got a message about missing plugins, I essentially have to go hunting for vintage synthesizers.

After a certain length of time, there isn’t much hope for me to recreate some of the music as I would need thousands of dollars worth of older synths to do it, but a lot of my newer stuff used virtual synths and I still have that software.  I mean, most of it, I do.  Some I had to really go out and hunt for as it was discontinued.  I still don’t know if I have it all yet.  I’ve only worked on a couple songs.  Always keep backups of everything.

One of the bigger problems I faced is that I used a synth from the time that was on a sound card – the Yamaha SW1000XG.  I do still have that card, but I can’t install it in my new micro PC system.  I was able to find a virtual version of the same synth, called the SY50XG, but it had a serious problem where you couldn’t directly select the patches per channel.  You have to do patch changes through SysEx messages.  That’s not insurmountable, except for the fact that I don’t know the exact patch that I need.  That lack of documentation, you see.

So, money to the rescue, as usual.  The SW1000XG is supposedly a PC card version of the Yamaha MU80 synth module.  I was able to find one for under $150 on EBay, shipped from Japan.  When it arrives, I should hopefully have everything I need to recreate the old songs and remix them at full FLAC fidelity.  All I should have to do is change the port from what was the SW1000XG to the MU80 and the patch I had selected on the old synth should map right to the new one.

But even this overall process is a real pain.  My recording workstation is not comfortable.  I have the choice of standing or sitting on a wood stool.  The keyboard is a mini keyboard with embedded touchpad, like using a laptop.  And all this equipment is in my music room, so there’s no real space to stretch out.  I feel like I need to eliminate my guest bedroom and make that a studio room, but I don’t want to do something drastic like that yet.

Over the long weekend, I worked on the project on and off in something like 30-minute increments.  Most of it was installing missing software synths and testing them out.  The recording PC is not network connected, so if I needed anything, I would have to walk back and forth between that and my regular PC in another room, transferring files on a USB drive (they used to call that "sneakernet" in the days before widespread computer networks).  So that process was annoying and exhausting in itself.

But I guess the big positive takeaway is that I haven’t given up yet.

Follow-up edit:

It turns out the MU80 is not the same thing as an SW1000XG.  After receiving the device and integrating it with my setup, I tested it out on a track I knew to use a lot of Yamaha sounds.  Very specifically, the drum kit I needed didn’t exist on the MU80.  Research, which I should have done before purchasing, would have given me the information I needed.  One web site gushing about the SW1000XG having 1200 sounds and 46 drum kits, then a Wikipedia article for the Yamaha MU series listing the different models and their capabilities gave me the full story.

The SW1000XG came out in 1998.  The MU80 came out in 1994 and had 729 sounds and 21 drum kits.  The MU100 came out in 1997 and had 1267 sounds and 46 drum kits.  And, you know, even if I was dumb enough to ignore the timeline, I should have given some credit to the model naming scheme.

The end result is I have to buy a Yamaha MU100, meaning I now have an extra sound module that is of little use to me.  Luckily, they aren’t that much more expensive than the MU80, but still, double the cost kind of sucks.  I suppose I can sell the MU80 and recoup some of that cost.