Something about the Super NT that seems to be overlooked (in reviews) is the thing actually has better audio than the original SNES. Luke over at LinusTechTips figured out what Analogue have implemented and goes into detail here
Question; is this something new or have emulators been doing it all along? Is this an enhancement that is present in any of Retorarchs cores?
Thanks.
I can’t find any real information on how it handles audio and how that differs from an original SNES, and his explanation is a little … vague. I also think it’s an odd claim that processing the audio differently from original hardware would be closer to what the composers intended, since those composers tested their music on the original hardware. They knew what was coming out of it and presumably tailored their music to the hardware.
Anyway, higan/bsnes and an SNES modded for digital audio have the same output bit for bit, so if kevtris puts some secret sauce on the audio, it’s going to be different from what higan (and thus RetroArch running higan) puts out.
I think honestly this is mostly just advertising and buzzwords if I were to be very honest with you.
And I don’t want to mean offense to any ‘tech’ Youtuber, I’m sure a lot of people watch that stuff and it can be very entertaining, but I kinda doubt they have the frame of reference or tech knowhow to be really able to discern fact from fiction (or marketing, in this case).
It would be interesting if someone who owns a Super NT could do a side-by-side comparison. Until there is actual data all we have is hearsay and theory. I’d love to hear how Aquatic Ambience or Corridors of Time sounds on that things in comparison to actual hardware (or Higan)
What’s the odds of it just being a high quality resampler?
And honestly, RetroArch has an impeccable resampler, Sinc-based resampler (there are others available too but they don’t have the same quality levels). And you can configure the quality of the resampler too since 1.7.1 to be as high or low as you want. Lower settings might give you lower latency though.
Honestly, though, I don’t find the entire comparison to be fair here since these guys have to offer up pretty much no sourcecode, or any kind of proof, of anything they are doing, yet they have lots of open source software to look at and base their work on. So I don’t really like all these hyped-up claims of superiority, same with the latency bragging points where they were trying to claim emulation was inherently inferior when even in that case, RetroArch has next-frame response times and under decent conditions there is no real perceptible difference.
In short, I’d say that while I am not opposed to Analogue NT or their products, and people might derive enjoyment from them and even buy them for legitimate purposes, I do not want to go down this kind of ‘tit-for-tat’ bragging war with them or really entertain any claims of superiority. I do not like this competition-oriented mode that certain youtubers (and in effect their company) seems to want to lead us down.
So I’d say videos like this are non-constructive and unhelpful, especially when they are narrated by people that frankly are not engineers and don’t know really anything meaningful about technology in that case other than what some company CEO regurgitates to them. They might be photogenic and they might be able to present themselves in front of a camera, but that does not mean they know the first thing about writing an audio resampler, or any kind of coding in fact.
There is so much hype around the Super NT because it uses hardware to emulate instead of software and people think it’s infallible. No lag, no bugs, better than real hardware. People who way that their is no lag are lying. The only way to hook up a Super NT is HDMI and that means it has to be hooked up to a modern display which means there will be input latency. How much? That depends on your display. No bugs? Well just about every reviewer including Digital Foundry encountered bugs in games already. They’ve been squashed, but those bugs were obvious and happened early in those games so they were easy to catch. What about ones not so obvious and don’t come around until later in a game and perhaps they may be game breaking. Nobody knows for sure. They certainly haven’t had the countless amount of hours of testing under their belt like Byuu and every single user of BSNES has. As far as audio sampling goes, I’m not sure how you up sample sound that has already been sampled at a particular frequency. 32kHz is still going to sound like 32kHz after you up sample it to 48kHz. As any audiophile should know you can’t take lower quality and up sample it to a higher quality without faking it through some sort of interpolation. Hell my Yamaha reciever can do that to MP3s. Is it better? That is up to the ear of the beholder to decide.
All that being said, I can tell you what the Super NT cannot do. Ever wanted to remove the crippling slowdown in games like Gradius III? The SNES9X core can do that. I made a You Tube video showing the night and day difference. Yes it’s not fool proof and I have seen some wicked bugs ensue in some games as a result of the overclocking hack in SNES9X but it works wonders for certain games like Gradius III and Biometal. Gradius III is a completely different game without all the slowdown.
I’m not saying the Super NT is junk, but it’s amazing that so many people look down on software emulation unless it’s stuffed inside a cheap plastic box that costs 60 to 70 bucks. Then there are the purists who turn up their nose to software emulation and wanna blow a wad of cash on a Super NT that’s so accurate that it can’t even play lightgun games! Oh wait, Retroarch can do that and you don’t even need a CRT to do it! I’m a real purist. Go real hardware or don’t. I don’t want anything in between like a Retron or Retrobit or a costly FPGA that claims to be the second coming of Christ. I have original hardware hooked to a 35 inch Sony Trinitron. I have a computer hooked to my HDTV upstairs. I love my real hardware, and I love Retroarch. I actually play Retroarch way more often LOL. I played all the way through Jumping Flash 2 on Beetle PSX. I dare anyone to tell me how inferior it is to my actual Playstation. Sorry, I didn’t mean to rant, but it just sort of happened.
Just to chime in with my experience with input lag. It is absolutely possible to get next frame response times with Retroarch. Genesis Plus GX get there (on my gaming PC) just from turning on GPU hard sync and setting max swapchain images to 1. Combine that with the nuked ym2612 sound core, the ability to disable to low pass filter and the new option for high quality audio resampling and I have no problem calling Retroarch the best way to play Mega Drive/Genesis games.
SNES latency though I have never gotten under 4 frames in any core. Now obviously 4 frames is basically imperceptible but if you are the kind of person that really wants no output lag then the NT does have value there. Its cheaper than buying a 1-Chip a scart cable and an OSSC anyway.
Fun fact. Not all games have have next frame input response. You need to consider the source material before you start counting frames. Everyone’s go to game when testing input latency on a SNES emulator (including me) is Super Mario World. Did you know that Mario does not jump on the next frame after the button press is registered by the console? Not only does he not jump on the very next frame, but he also doesn’t jump on the 2nd frame either. Mario jumps on the 3rd frame on actual hardware hooked up to a standard definition TV. Yeah that’s not a typo. The 3rd frame. Not the next, not the 2nd, the 3rd frame. When you then take into consideration the average of 4 frames that you get whenever you test input latency in Retro Arch, you’re actually only down a little over 1 frame, not 4. Then if you take into consideration that your display lags if you have an LCD monitor and especially an LCD TV, you could make up most if not even all that difference. Monitors and TVs will lag even at their native resolutions and game mode activated. How much it lags depends entirely on your display. Not enough people blame their displays. People swear by Frameisters because it makes 240p look so stunning on a LCD TV, but they never wanna talk too much about the added latency they add. Some admit to as much as 4 frames. Now they could be counting frames with the assumption that the action on screen happens on the very next frame after the console gets the signal from the controller. In that case the Frameister could only be adding a little over a frame. There’s a reason why the Sega Genesis was so popular among sports game lovers. The Genesis’s controls were faster, tighter, more responsive than the SNES versions. Why? Because the Genesis did have that next frame response and the SNES alot of times did not. Even when compared to the NES the SNES is still slower. I first thought that something was off when I went from playing Super Mario Bros. on real hardware which does have a next frame response to Super Mario World and while the controls are tight, they’re not as tight as Super Mario Bros. That’s when I pulled out my camera and found that Mario doesn’t jump on the next frame even on real hardware hooked to a standard def CRT.
To someone who plays real hardware on a LCD HDTV the Super NT will seem like a miracle in more ways than one. It will look so much better and the difference in the input latency will be so huge that it will seem like the Super NT isn’t lagging in the slightest. To those who bought a Frameister the Super NT will most likely be a lateral move. To those who use a PC with Retro Arch, the differences may not be worth the price of admission. I would LOVE to get my hands on one and do a head to head. I don’t see that happening unless my brother picks one up. I can afford one but I’d rather spend 200 bucks on something else that I know I really want. I don’t want to end up making an even bigger mistake than when I got a SNES Classic and found out Retro Arch was faster and more accurate. .
Years ago, ZSNES had an option to use Cubic or Sinc interpolation for audio instead of Gaussian interpolation of the actual hardware. I don’t think higan/bsnes has ever had that capability since it focuses on accuracy, but I’m not sure about Snes9x (the current version doesn’t seem to expose an interpolation option, though past versions might have). SPC players have also had the capability of changing the interpolation. So this is really nothing new as far as SNES emulation is concerned.
Using Cubic or Sinc does indeed make things sound cleaner than Gaussian, but it can make certain things sound wrong as well, since it would be different from the real hardware’s behavior.
I’ve gotten the chance to play around with one. It is honestly a lovely little thing. Super well made, hell even the PCB layout is pretty! Its clear a lot of love and attention went into it. Yes they did something with the sound. I only got to play Turtles in Time but it is super obvious on stuff like the voice clips. Much clearer than than the actual hardware. That guy struggling to say Prehistoric Turtlesaurus never sounded as good. What really got me was the buzzing of Baxter Stockman. I know the game damn well and on real hardware his buzzing has a mechanical artificial quality to it. On the NT God damn if it doesn’t sound like a buzzing fly! Also the image scaling is super impressive.
Whatever you think of software emulation vs FPGA (hardware emulation) the Super NT is a well made device.