Dexed Librarian [TEASER]

[planetearth]

Someone did some research on all the DX patches available online. It turns out the vast majority of them are duplicates, saved under different names. (The fellow who did the research posted about it on Twitter, but I don't have a link handy.)

Curious... !

Because it is precisely the work that I did in 2014 (it needed 6 months) to finally compile this file that you can download in the middle of Dexed's webpage :

• https://asb2m10.github.io/dexed/#cartridge-manager

Download it... and read the file "Version and Readme.txt" which is at the root. It explains all the work that I did.

The first time I talked about that huge work was here (24 Aug 2014). Then I made several "beta" versions (0.x), then I released the version 1.0 here (05 Jan 2015).

And a guy now says that he did the work ? And he announces that on Twitter ? Problem: I don't have any account on Twitter nor on any social network...


It took me 6 months to make that monstrous work of patience and meticulousness (in French we say "a work of benedictine"), removing one by one all the dupe files having several filenames in the different collections and even within a same collection, repairing some files which were broken (and there are others which probably remain), leaving away all those too much broken and all those badly converted to TX816, etc., all that work using several scripts I did in VBA and an hexadecimal editor.

I would be curious to compare the content of "his file" with the content of mine. I wouldn't be surprised to find practically the same file with the same organization of directories, etc...

Many of us on KVR (and Gearslutz) know you did this work, @BlackWinny, and we appreciate your effort. It's disappointing--but not surprising--that some jerk is passing your work off as his.

I have a Twitter account, and I love to take apart people like this. So if anyone wants to give me this guy's address there, I'll be happy to see what he's trying to pull here.

Steve

[BlackWinny]

[planetearth]

And yes, the amount of presets which have been made for the Yamaha DX7 is so high that it needs a brand new type of memory management in the style of an internal database (with at least an index and perhaps even several indices) to reduce the huge memory hog.

Someone did some research on all the DX patches available online. It turns out the vast majority of them are duplicates, saved under different names. (The fellow who did the research posted about it on Twitter, but I don't have a link handy.)

It would be terrific, if this utility could help find and remove all the duplicates.

Maybe this is what he's talking about...?

"Dexed iOS FM Synthesizer Updated With 640,000 Unique Patches"
http://www.synthtopia.com/content/2019/ ... 9Hv8O1QLqk

This port of Dexed is for iOS, and costs $3.99. There's no indication the developer was going to release his "640,000" patches, however.

Steve

[BlackWinny]

Thank you Steve. Yes it is probably what he's talking about.

But as I don't have any iOS nor any OSX i'll not be able to compare his database with mine.

Bah ! That's life...

[McLilith]

No. That wasn't what I was talking about. The post I read mentioned nothing about having a product to offer. It was just some fellow that wrote some custom software to compare all his collected DX7 patches and automatically weed out the duplicates. I think, for him, it was largely just a fun programming exercise. I wish I could remember the percentage of patches that were actually unique. It was surprisingly small. Perhaps 10% to 20% unique patches, and the rest were dupes? Don't quote me on that. I just remember it was an impressive stat.

[BlackWinny]

Ok McLilith.

It would be nice if you could find again that post.

[McLilith]

I did look for it, but it might have been from a year ago or more. Anything that old on Twitter is almost impossible to find, especially since I don't even remember the user's name.

[planetearth]

I did look for it, but it might have been from a year ago or more. Anything that old on Twitter is almost impossible to find, especially since I don't even remember the user's name.

No worries. Hopefully, he's not trading on our friend BlackWinny's hard work, which is all I'm concerned about, really. If I do see him, I'll ask him about it. I'd be surprised if 80% are "duplicates", though. Many may have duplicate names, and many may sound almost exactly the same, but I think the real number of true dupes may be closer to 50%--which is still quite a lot. But since people have been developing sounds for 6-OP FM synths for decades, that's not too surprising.

That said, I'd still be interested to know where the guy who's selling the iPhone/iPad version of Dexed got most of the "640,000" patches he's providing. And many of those simply must be duplicates, wouldn't you think? For example, he may not considering two "E. Piano 1" patches to be "duplicates" if a couple of parameters are slightly different from each other, but if the patches sound the same, they're "duplicates" as far as the end-user is concerned.

And even at that, if some of the 640,000 patches he's providing came from the same professional developers like Grey Matter, Kid Nepro and others from back in the day, well...those are copyrighted. He can't "bundle" them with a product he's selling, if the inclusion of those patches somehow enhances the value of what he's selling.

Steve

[McLilith]

I can't imagine having 640,000 distinctly unique sounding and useful patches in any one synth. I think most of those would either be duplicates, or what I would call "junk" presets, included just to drive up the preset count for marketing purposes.

Also, just to audition all of them, if you only played each patch for 10 seconds, would take about 15 weeks!

(I'm assuming an 8-hour work day, 5 days a week, doing nothing but listening to presets.)

The whole deal just seems shady to me.

[BlackWinny]

Synthtopia - Dexed iOS FM Synthesizer Updated With 640,000 Unique Patches

(read the comments made since yesterday)

[BlackWinny]

I can't imagine having 640,000 distinctly unique sounding and useful patches in any one synth. I think most of those would either be duplicates, or what I would call "junk" presets, included just to drive up the preset count for marketing purposes.

Also, just to audition all of them, if you only played each patch for 10 seconds, would take about 15 weeks!

(I'm assuming an 8-hour work day, 5 days a week, doing nothing but listening to presets.)

The whole deal just seems shady to me.

Yes.


Let's imagine that the designers began to create presets for the DX7 on the 1st January 1983. Today is 31st July 2019. The difference between these two dates is 13,360 days. If I divide 640,000 by 13,360 I get around 48 presets per day. If I had hired a guy since the 1st January 1983 until today he would have made 48 presets each day, doing nothing else. And including the Sundays, the holidays, etc.



Let's make another count:

I even can't imagine the huge crowd of musicians or sound designers which would be needed to summarize to even MAKE all these 640,000 unique presets during the few years of the real fame of the 6-OP family directly compatible with the original DX7. The DX7 (including the version II) was sold only since 1983 up to 1989. Meaning: 6 years. If we reasonably imagine that on these 640,000 presets 1/4 of them have been created after 1989, it would remain 480,000 presets created during these 6 years, meaning... 80,000 presets per year!


That's totally impossible, even with... say... 80 sound designers it would mean that each one made 1,000 presets per year. Per year... during 6 years!

Totally insane.


Delirium tremens.

[zzz00m]

Conclusion: at least 90% of them must be dupes...

[McLilith]

I finally found the link for the person who took 1,853,728 DX7 presets, removed all the duplicate presets, and was left with only 31,380 unique presets. In other words, 98.3% of the individual presets in his collection were duplicates. My memory was fuzzy about the exact numbers when I first posted to this thread about it, but I knew it was an impressive statistic!

Also, I really don't get the impression this fellow is "ripping off" BlackWinny in any way. He just seems to be a computer scientist who used some software to whittle down his collection of DX7 presets and automatically remove the duplicates. To me, it just comes across as a fun computer programming exercise.

Also, if you have 2 hours to kill, you can listen to an audio clip that features the 31,380 unique presets. Be warned though, it's interesting in a way, but it's also a bit of an endurance test.

Here's the link:

https://soundcloud.com/bwhitman/the-313 ... of-the-dx7

[zzz00m]

I finally found the link for the person who took 1,853,728 DX7 presets, removed all the duplicate presets, and was left with only 31,380 unique presets. In other words, 98.3% of the individual presets in his collection were duplicates. My memory was fuzzy about the exact numbers when I first posted to this thread about it, but I knew it was an impressive statistic!

Also, I really don't get the impression this fellow is "ripping off" BlackWinny in any way. He just seems to be a computer scientist who used some software to whittle down his collection of DX7 presets and automatically remove the duplicates. To me, it just comes across as a fun computer programming exercise.

Also, if you have 2 hours to kill, you can listen to an audio clip that features the 31,380 unique presets. Be warned though, it's interesting in a way, but it's also a bit of an endurance test.

Here's the link:

https://soundcloud.com/bwhitman/the-313 ... of-the-dx7

If you actually listen to that SoundCloud link, it sounds like a short list of the same sounds looping over and over.

[BlackWinny]

Apparently he made a neat work too, indeed.

https://www.google.com/search?client=fi ... q=bwhitman

https://github.com/bwhitman/learnfm



And he works (or have worked) at the MIT: http://alumni.media.mit.edu/~bwhitman/