Retro, The Sounds and The Music.

Discussion in 'Retro & Arcade' started by power, Jul 30, 2010.

  1. dirkmirk

    dirkmirk Member

    Joined:
    Apr 3, 2002
    Messages:
    5,403
    Location:
    Shoalhaven - Gods Country
    Something like that device housed In a simple case with power supply included would be nice, I've paid decent amount of money for a staking box(crypto) which is just a Pi3+ but its aesthetically pleasing and plug and play, something simple people will pay a premium for, if it's kitform and has nothing to house it in the appeal is not their for me...
     
    Last edited: Aug 4, 2020
  2. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    42,739
    Location:
    Brisbane
    That's not what I'm talking about at all. All of that hardware you're talking about in that post is different to the original.

    Nobody's making the Yamaha YMF chips inside sound cards. Nobody's making the SID chips inside the C64. If you want to buy these "new", you're buying small simulation devices, typically in the form of FPGAs, AT-Megas (Arduinos), ARM (RPis), etc.

    Nobody's making the bits and pieces needed to make a "new" MT32. Your only option from here is a simulation device. Doesn't matter if it's partial (i.e.: someone makes a smaller FPGA that plugs into some other original hardware), or total (MiSTer, etc).

    Yes, people love buying OLD hardware. And people love plugging NEW things into OLD hardware. But all of these new things are simulators. Every flash cart is an FPGA. Every replacement sound or video chip is an FPGA. Every upgraded CPU for your old Amiga is an FPGA. That's the point.

    The original comment was on cost. The complaint was the $200 outlay for said devices. I have looked at the Amiga/ST/C64 communities, and I have seen their devices, and they're crazy expensive.

    And that's fine. I get it and I get why. But the comment wasn't that these things don't exist. The comment was that they aren't cheap, and I explained why.

    Point stands - original hardware, like YMF chips, SID chips, M68K chips - these are all dead. All of the replacements are FPGA, and all of the replacements are expensive due to economies of scale. Our job as a community is to objectively compare them to the originals while we can, and improve them while we can.
     
    Last edited: Aug 4, 2020
  3. shredder

    shredder Member

    Joined:
    Dec 26, 2001
    Messages:
    13,969
    Location:
    New Zealand
    They sort of do, but this is niche stuff.
     
    Last edited: Aug 4, 2020
  4. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    42,739
    Location:
    Brisbane
    Quick check across all of these cards proves the point - the main core of these, the OPL chips, are stolen from existing cards, or worse, use inaccurate clones.

    See the video a few pages back as well (edit: reposted below), where "new" stock of chips are either clones from China, or old chips that are failing. These don't match the accuracy of the original chips/cards, so you're not even hearing the same thing.

    New card, old chip. The PCB fab isn't the important bit. The chip is. This is a real problem, and a short term solution.

    We *need* more MDFourier data on these, and we need to do it while the originals still live.



    All of these simulated/clone/fake chips are inaccurate compared to the originals. You end up paying through the nose to be no better than a "filthy casual". Food for thought.
     
    Last edited: Aug 4, 2020
  5. shredder

    shredder Member

    Joined:
    Dec 26, 2001
    Messages:
    13,969
    Location:
    New Zealand
    Good point aye. No new key chips made = countdown to extinction (hardware wise).

    And as I've said elsewhere, the carefully hoarded hardware we all have is not permanent at all. Give it regular use and most of it will be dead within a decade or two.
     
  6. oculi

    oculi Member

    Joined:
    Aug 18, 2004
    Messages:
    11,710
    They are still making analogue synthesizers but that may be off topic.

    I'm just glad there are some preservation efforts going on, a while back I used a PC based SID player that was just awful, plenty of people have been doing high quality recordings from real hardware for a while now which is nice.
     
  7. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    42,739
    Location:
    Brisbane
    Again, I'm talking about 1:1 replacement parts for old synths.

    New synths are very cool, and I love the sounds they make. I bought a "Pocket Operator" for my brother in law recently, and it's a spectacular little device.

    But the issue here is objective accuracy of old stuff. It's very clear with a little research just how bleak the future looks when people in 20-30 years time want to hear OPL or SID music *accurately*, and all they've got is an FPGA chip (whether it lives in a real C64 mobo or not), that some enthusiast ear-matched (or worse, memory-matched) and figured was "good enough".

    We've had cycle accuracy of CPUs for a while now, which has been a real boost for preservation. Audio is still miles behind, with even some of the best implementations not matching original hardware.

    I repeat ad nauseam - MDFourier is the answer. If it tells you the audio matches, it matches. If it says it's wrong, it's wrong. No "good enough", no "it's how I remember it". We need to be comparing old to new now, before old vanishes.
     
    Last edited: Aug 4, 2020
    oculi likes this.
  8. OP
    OP
    power

    power Member

    Joined:
    Apr 20, 2002
    Messages:
    64,837
    Location:
    brisbane
    i mean look i get it, there is a certain amount of augmentation and chips cannot be remade. that we can certainly agree on. Calling the old chips dead because they aren't made anymore is a bit of a reach.

    upgraded CPU that is an FPGA? No, only a few are like Witcher, and they are a little frowned upon (I mean what's the point?). People preferring accelerators made with proper Motorola CPU's on them like the TF's and ACA's.
     
  9. shredder

    shredder Member

    Joined:
    Dec 26, 2001
    Messages:
    13,969
    Location:
    New Zealand
    Given the work done so far for MiSTer, is not similar work FPGA'ing OPL and similar chips not "simply" a matter of moving sideways, and applying the techniques they've used to do those things in MiSTer?

    I guess I'd be surprised no one was doing it already. Given the hours of endeavour clearly going into sound card work in general.

    What is the fear here? That it won't be done? That some important things are slipping through the cracks?
     
    Last edited: Aug 4, 2020
  10. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    42,739
    Location:
    Brisbane
    I didn't call the chips dead. I called the manufacturing processes of these chips dead.

    Please understand what I'm saying instead of leaping to whatever conclusion you've reached here. I'll spell it out: nobody is making LA32, SID or Yamaha sound chips new. Every "new" chip on the market is a fake, and has accuracy issues. Every clone so far has accuracy issues (some are better than others, none are perfect). There is no future for retro audio right now that is based on newly created original hardware. That is "dead".

    All Cyclone 3 FPGAs: https://www.apollo-accelerators.com/

    These things are all over YouTube spruked by some of the biggest popular names there. I'm not sure what "frowned upon" means in that context, but they seem quite popular.

    Either way, Motorola CPUs still existing is the exception to the rule. In a similar way that some ARM and PowerPC processors are still made today that are compatible with older architectures. Compare any of those to the laundry list of CPUs and other chips that just don't exist any more. More to the point - these CPUs are not sound chips, and the discussion sparked from sound chips (specifically the question as to why nobody was making them cheaply). On that topic, we have a very real problem when it comes to the future of ACCURATE retro audio.

    Lots and lots of work is being done with MDFourier on consoles. MiSTer right now is the single most accurate FPGA implementation of 3 different families of Megadrive/Genesis audio hardware, and that's great. NES audio has been improving for a while and keeps getting better. Likewise PCEngine/TurboGrafx is catching up, and SNES is in development, so it will improve soon.

    To my knowledge, nobody is bothering with computer audio. Everything done to date is measured subjectively by ear, which isn't good enough (especially if it's older folks who are losing audio range, which isn't their fault, it's just a natural thing).

    The fear is that it's a process that will take years, maybe even decades, of gradual improvement. And by the time people start to care, the original hardware will be so few in number that we just can't do the work as we don't have the original thing available to the right people to do the comparisons.
     
    Last edited: Aug 4, 2020
  11. shredder

    shredder Member

    Joined:
    Dec 26, 2001
    Messages:
    13,969
    Location:
    New Zealand
    What should a person with a rare sound card be doing then, in the practical sense?

    I don't know what preservation means to me. Coming at it philosophically, I'm a person who doesn't even have children. So I'm not even dedicating my life to preserving my own species, in the way that most people are. But look, I'm totally a nerd/geek, always have been. The concept then, of working one's only life to preserve niche digital microcosms for the benefit of a minority of said species is... complex. As nerdgeek I laud those who do it, of course - how could one not??

    Is that view .. understandable at all?
     
    Last edited: Aug 4, 2020
  12. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    42,739
    Location:
    Brisbane
    Read the howtos for MDFourier, and work with the devs to generate the audio test sweeps.

    Once that's done, doing the testing. It involves running the MDFourier software to capture audio from your original sound card as well as clone hardware or emulators, and generating the graphs.

    Here's an example of the Sega Megadrive/Genesis. In all graphs, the three sound generators (FM "Frequency Modulation", PSG "Pulse Sound Generator", noise - random/white noise) are lines on the graphs. The closer these are to zero, the closer the sound of the compared devices are. In all cases, audio is compared to the original VA1 Megadrive/Genesis.

    1) Comparing a VA1 to a VA6 motherboard. Same Yamaha sound chip, ever so slightly different output due to analogue components on the boards around the chips, poor trace routing causing noise, etc. But these differences are minute.

    1.png

    Next, the Analogue Mega Sg. Developed by FPGA guru Kevtris, who despite being a very clever guy, dislikes open source. Analogue's designs are all proprietary and nobody outside of their company can improve them. Let's look at an audio comparison:

    3.png

    Miles off. Huge variances in the 2KHz-10KHz range that are obvious on the graph, and would be quite obvious even to the untrained ear. Better in the more critical 100Hz-1KHz range (human voices are in here), but still pretty terrible.

    Final comparison, MiSTer FPGA cores from a while back (earlier this year). Things have improved since, but these were still pretty good:


    2.png

    Larger variances at the very high end (pushing into 20KHz which is the upper bounds of what humans can hear). But even then +/- 3dBFS at those frequencies puts them into the almost indistinguishable category. The critical 100Hz-1KHz is very good. More to the point, the data is objective and nobody can disagree with it. It's good, but it can improve, so people are improving it. No ego, no opinion.

    This data doesn't exist for PC and microcomputer audio right now. You bought a new OPL3 card from someone on Vogons. It sounds "good to you". But how good? How accurate? Does it match your original cards of choice? How do you even know?

    We need data, and we need it before more hardware ends up in landfill or private collections.
     
    Last edited: Aug 4, 2020
  13. shredder

    shredder Member

    Joined:
    Dec 26, 2001
    Messages:
    13,969
    Location:
    New Zealand
    First time seeing dbFS. 1minutegoog says it's a amplitude scale based on an known upper bounds, or something. I'll have to digest that.

    Where does the noise line come from? Background common to both?

    If anything I'd be interested in doing my OPL4, because that's the rarer one.
     
    Last edited: Aug 4, 2020
  14. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    42,739
    Location:
    Brisbane
    Noise is a sound channel on many sound chips. As in, you actually want to generate noise sometimes.

    Good explanation here for the Nintendo NES, but it applies to many other devices and sound cards.

     
  15. Grant

    Grant Member

    Joined:
    Jan 23, 2002
    Messages:
    1,506
    Location:
    Wollongong
    +1 MODs (you should listen to MODs while talking about audio hardware preservation).


    Does anyone know the current state and/or history of OPL documentation (the chip family, not any board implementations)? I assume MAME has them covered to a high quality standard, but is it based on vendor documentation (specs), experimental reverse engineering, decapping real chips and looking at the physical implementation, or (unlikely) actual released silicon implementation docs?

    I'd hope that Yamaha still has this stuff in a vault somewhere, but whether they know where it is or whether they'd ever be willing to release it, no one can say. Japanese companies have a reputation of being extremely conservative, but the people that work at Yamaha today are fans who grew up with the early synths.
     
  16. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    42,739
    Location:
    Brisbane
    I have only read hearsay, but that suggested MAME wasn't very accurate.

    I don't know if there are other tools people are using to compare these things with outside of MDFourier. Any discussion group I land on just has endless MP3s posted and asks people to judge for themselves. Lossy-compressed audio listened to over headphones doesn't make for objective measurements, however.

    You can see for yourself how deep MDFourier goes to compare things ( shredder this also explains why they use dBFS):
    http://junkerhq.net/MDFourier/files/mdfourier.pdf

    But even without it, perhaps comparing audio waveforms in various audio editors could be a manual version of this? But I haven't seen anyone doing that. So again, every discussion is punctuated by a bunch of folk on forums telling everyone else what they think is good enough with no actual evidence.
     
    Last edited: Aug 4, 2020
  17. OP
    OP
    power

    power Member

    Joined:
    Apr 20, 2002
    Messages:
    64,837
    Location:
    brisbane
    did anyone watch this, i was a little disappointed but occasionally surprised by the choices.

    I nearly unsubscribed when Try started ragging on Turrican - Nintendo fanboy's can be the worst. They really lock themselves out with some of their console-centric content and snobbery toward the euro scene.

     
    Last edited: Aug 4, 2020
    elvis likes this.
  18. Sphinx2000

    Sphinx2000 Member

    Joined:
    Sep 16, 2001
    Messages:
    9,497
    Location:
    Brisbane
    Sif rag on Turrican 1/2 OST, some of the best out there.
    I have had much respect for the Euro scene since the demo/tracker scene days of Jonne Valtonen (Purple Motion) and many others.
     
    Last edited: Aug 4, 2020
  19. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    42,739
    Location:
    Brisbane
    It's really hard to imagine just how much the NES dominated the US at the time. Even as a filthy console baby, Sega dominated my childhood (and the entire town I lived in) across both the 8 and 16 bit eras, and when you check out any US podcast/youtube-channel/whatever covering that era, you just see how utterly myopic they are to everything of that era.

    Even as far as I've heard "Euro" used as a criticism of great games (i.e.: "It's a fun title, but it just feels a little too Euro"). Like, how is that even a comment on how good or bad a title is?

    I guess that's part of the reason we all follow a bunch of UK/EU folks to try and balance that silliness out.
     
  20. OP
    OP
    power

    power Member

    Joined:
    Apr 20, 2002
    Messages:
    64,837
    Location:
    brisbane
    skip to about 1:09 Coury brings up Turrican 3 and then Try can't contain himself, I think he's missed out on a lot of gaming history doing this. It's times like this when I appreciate how open John Linneman is when he openly states he doesn't have a history with 8 and 16 bit computers but he appreciates them and has a desire to actually jump in and sits down with people like Audi Sorlie and discusses that side of video gaming.

    I honestly think that some console kids are too used to using a gamepad, the true early days had a stick and a button - that was it. They just can't deal with the control scheme. To me it's very natural to play a side scrolling game with a stick, up for jump and button for fire.

    In the context of music, the european houses were doing great things in a different way to Japan with different HW. The other thing that can't be understated is the music that came out of the demoscene, nothing like a good tracker and a few kb's of animation. ;)
     
    Last edited: Aug 4, 2020

Share This Page

Advertisement: