Homebrew RetroArch - A new multi-system emulator

[MassiveRican]

@nintygaming... I love u bro! Lol. Seriously, you saved me a lot of time by providing these links and nowadays that's truly one of the best things u can do for anyone. plus u gave me tons of valuable info which I can apply to enhance my gaming experience. U sir deserve a medal lol.

This thread has literally been a crash course for me in video and it seems now its about to turn into audio. Thank you RetroArch, nintygaming and all who provided insightful info over the last few pages!

 

[mudlord]

Easy:

1) Get LOSSLESS recordings. MP3 completely invalidates results as the encoder WILL introduce artifacts. Make them roughly the same length and of the same portion of the game of said system.
2) Get a ABX tool like foobar2000 with foo_abx or something.
3) Run the tests several times to see if you can really hear a difference, or if its just the placebo effect (ie: bullshitting)
4) Send us the logs of the ABX tests.

 

[the_randomizer]

Easy:

1) Get LOSSLESS recordings. MP3 completely invalidates results as the encoder WILL introduce artifacts. Make them roughly the same length and of the same portion of the game of said system.
2) Get a ABX tool like foobar2000 with foo_abx or something.
3) Run the tests several times to see if you can really hear a difference, or if its just the placebo effect (ie: bullshitting)
4) Send us the logs of the ABX tests.

Lossless, as in wave file. Gotcha
I have foobar2000, so that's covered.
The plugin I have, but I'll the other what to do

Thanks for clarifying that, I was really confused, but now we're getting somewhere.

 

[tacodaemon]

I can certainly ABX the lossless files myself, but what is the point when the volume varies so greatly? It's pointless when you can tell the difference by volume alone. Shall I normalize the volume or will that invalidate the test results as well? Did you listen to the files that were uploaded?

I'll find an example that doesn't require gameplay that will create differences among the files and normalize the volumes and test, Let me know if you would like different test parameters.

Edit: Here are the original WAV files from the previous test of FCE and Nestopia in .9.8.2 with unmodified settings from fresh install. https://anonfiles.com/file/6814a866a106219ee60ce078c5233e89

 

[the_randomizer]

I can certainly ABX the lossless files myself, but what is the point when the volume varies so greatly? It's pointless when you can tell the difference by volume alone. Shall I normalize the volume or will that invalidate the test results as well? Did you listen to the files that were uploaded?

I'll find an example that doesn't require gameplay that will create differences among the files and normalize the volumes and test, Let me know if you would like different test parameters.

Oops, ninja'd, disregard PM I sent. I don't think it would invalidate the test as even with the distorted volumes, the audio is still tinny, so using the normalize plugin in Audacity will help. ABX testing is the only way we can convince the authors of RetroArch, or so Libretroretroarch stated. A good sound to test is the coin sound effects, the mushroom shrink/grow, warp pipes, etc. A good level to test would be anything that plays the underground music theme. You've got this, I know you can. Normalizing the volume will help a lot when comparing the results.

 

[mudlord]

the emulator is definitely doing something weird when sampling the audio in RetroArch. I know for a fact that issue doesn't happen in the Windows version.

Which version, etc

 

[the_randomizer]

Which version, etc

0.98.1 for the Wii, the emulator core is the port of Nestopia 1.43. On the Windows version, the issue is nonexistent, but I spoke with tacodaemon via PM and he will assemble all audio test results tomorrow from the ABX testing.

Ugh, I hope the evidence is sufficient enough. If the goal for RetroArch is accuracy, I would assume that the audio aspect would, you know, sound a little closer to real hardware, but I digress.

 

[LibretroRetroArc]

0.98.1 for the Wii, the emulator core is the port of Nestopia 1.43.

The Windows version of NEStopia that sounds correctly according to you - specifically. There is now NEStopia Undead as well (which is what NEStopia is based on).

 

[mudlord]

If the goal for RetroArch is accuracy

DING DING DING, the magic buzzword HAD to pop up!

When did maister and co ever say the retroarch framework is for accuracy? it runs things apart from emulators you know......

 

[the_randomizer]

The Windows version of NEStopia that sounds correctly according to you - specifically. There is now NEStopia Undead as well (which is what NEStopia is based on).

I'm confused at what you're trying to say. Compared to real hardware, yes, Nestopia is dead on, there are very little if any perceptible difference. The Wii port however, has volume distortion and tinny audio (much like when you reduce the sample frequency from 44.1 to 22 KHz). Genesis Plus GX 1.7.3 suffers from this as well when running from RetroArch, but not from the standalone emulator of the same version. So, I spoke with tacodaemon and he agreed run numerous ABX tests and will post them tomorrow, as suggested from an earlier post you made. Hopefully, the results will suffice and we can finally put this behind us.

My ears are trained and hypersensitive to the point of being able to perceive nuances and quirks when comparing real hardware to an emulator. Genesis Plus GX 1.7.3 sounds pretty damn close to a real Genesis/Megadrive when using the standalone emulator, but has distortion and tinniness when running in RetroArch 0.9.8.1. I should note that it didn't sound as bad in 0.9.7, but this one, the tinny sound really went to hell.

Summary

Nestopia 1.43/Undead in Windows - Very very close to real NES
FCEUmm - Very close to a real NES
Nestopia 1.43/RetroArch - Distorted, tinny audio, not as clear as FCEUmm
Genesis Plus GX 1.7.3 by itself - Near perfect YM2612 emulation, can't tell difference to a real system
Genesis Plus GX 1.7.3 with RetroArch - same issues as Nestopia on RetroArch, the sound is distorted and tinny.

What's funky is that two instances of the same emulator on the same system can sound so different.

Again, please, don't perceive this as me being confrontational, I'm merely pointing out something that is clearly an emulation bug that many feel needs to be brought to your attention. Rest assured, ABX tests will be here tomorrow.

DING DING DING, the magic buzzword HAD to pop up!

When did maister and co ever say the retroarch framework is for accuracy? it runs things apart from emulators you know......

Well, I was a moron for bringing that up. I've said enough to piss even more people off and wasted your time with what is perceived as a seemingly trivial issue, but I digress.
Nevertheless, the issue needs to be looked at, and while accuracy may not be #1 priority, said issues should be taken into consideration if you want people to enjoy the emulator even more than they already do. I can't see why it's hard to convince other people that the issue exists.

 

[LibretroRetroArc]

Ugh, I hope the evidence is sufficient enough. If the goal for RetroArch is accuracy, I would assume that the audio aspect would, you know, sound a little closer to real hardware, but I digress.

That's not RetroArch or libretro's goal. The goal is this -

www.libretro.org

The purpose of the project is to help ease the work of the emulator/game porter by giving him an API that allows him to target multiple platforms at once without having to redo any code. He doesn't have to worry about writing input/video/audio drivers - all of that is supplied to him by RetroArch. All he has to do is to have the emulator port hook into the libretro API and that's it - we take care of the rest.

And RetroArch is a direct implementation of that goal.

Regarding resamplers - IF it's the Hermite resampler that is making it sound 'tinny' then that is something you will have to live with - because the sinc resampler's LUT tables eat up 4+MB and we can't ever justify wasting that much amount of RAM just for a minuscule increase in audio quality that only a few people can even notice or be bothered with.

 

[the_randomizer]

That's not RetroArch or libretro's goal. The goal is this -

www.libretro.org



And RetroArch is a direct implementation of that goal.

Well, I was a moron for bringing that up. I've said enough to piss even more people off and wasted your time with what is perceived as a seemingly trivial issue, but I digress. Nevertheless, the issue needs to be looked at, and while accuracy may not be #1 priority, said issues should be taken into consideration if you want people to enjoy the emulator even more than they already do. I can't see why it's hard to convince other people that the issue exists. I don't know why it's so hard to explain to people that the issue is there and why it should be looked into. It the standalone Genesis Plus 1.7.3 gets the sound right, why the hell can't the RetroArch port sound just as crisp/clear? Again, ABX results will be here tomorrow.

Again, Nestopia and Genesis Plus GX in RetroArch have distortion/tinny audio issues, hence the ABX tests. Accuracy be damned, people just want to enjoy the emulators even more, but if these audio issues can't or won't be looked into, well, who knows.

 

[mudlord]

Won't be looked into? People just want PROOF that they exist in the first place!

I'm sure you wouldn't want to waste time on things when for all we know, it could be just the placebo effect in play.

 

[nintygaming]

That's not RetroArch or libretro's goal. The goal is this -

www.libretro.org



And RetroArch is a direct implementation of that goal.

Regarding resamplers - IF it's the Hermite resampler that is making it sound 'tinny' then that is something you will have to live with - because the sinc resampler's LUT tables eat up 4+MB and we can't ever justify wasting that much amount of RAM just for a minuscule increase in audio quality that only a few people can even notice or be bothered with.

Just out of curiosity, have you considered opening up a Google Code issue tracking page for this so that issues are reported in a neat, organized manner instead of being jumbled up in a mess on here?

I've been wondering why this hasn't been done yet. My apologies if I'm asking a question that has already been answered, but I have not found an answer.

 

[izzydeank]

I really wish people like the_randomizer would stop bugging the devs endlessly about stupid little crap that only they care about. I just get tired of reading through endless posts about trivial stuff that certain people will not give up about. They do this for free, but yet people act like they paid for something. IF YOU REALLY WANT IT THEN FIGURE OUT HOW TO DO IT YOURSELF. THANKS.

 

[the_randomizer]

Won't be looked into? People just want PROOF that they exist in the first place!

I'm sure you wouldn't want to waste time on things when for all we know, it could be just the placebo effect in play.

Right, but the tone some people convey in their posts gives the impression that the bugs aren't worth looking into and that it's an auditory hallucination, so yeah, ABX tests will be done, and I'm hoping the results will suffice. I don't know how much more of this I can handle, or how many more times I need to explain it in depth.

I really wish people like the_randomizer would stop bugging the devs endlessly about stupid little crap that only they care about. I just get tired of reading through endless posts about trivial stuff that certain people will not give up about. They do this for free, but yet people act like they paid for something. IF YOU REALLY WANT IT THEN FIGURE OUT HOW TO DO IT YOURSELF. THANKS.

Don't like my posts, either ignore them or block me.

This is my last post regarding the issue. I await results.

 

[mudlord]

Right, but the tone some people convey in their posts gives the impression that the bugs aren't worth looking into and that it's an auditory hallucination, so yeah, ABX tests will be done, and I'm hoping the results will suffice. I don't know how much more of this I can handle, or how many more times I need to explain it in depth.

Fair enough

 

[mudlord]

I really wish people like the_randomizer would stop bugging the devs endlessly about stupid little crap that only they care about. I just get tired of reading through endless posts about trivial stuff that certain people will not give up about. They do this for free, but yet people act like they paid for something. IF YOU REALLY WANT IT THEN FIGURE OUT HOW TO DO IT YOURSELF. THANKS.

Not really, the randomizer isnt bugging the dev. Its just ABX tests are valuable for audio related matters like signal quality. For obvious emulation bugs, they can be ignored. but when it comes to subjective things like audio quality, thats where they matter.

 

[the_randomizer]

Fair enough

Glad we're at an understanding. I absolutely love this emulator, I mean, having so many consoles in one is very convenient, and I wish the developers success as they continue to make it even more awesome

 

[LibretroRetroArc]

Just out of curiosity, have you considered opening up a Google Code issue tracking page for this so that issues are reported in a neat, organized manner instead of being jumbled up in a mess on here?

I've been wondering why this hasn't been done yet. My apologies if I'm asking a question that has already been answered, but I have not found an answer.

We won't use Google Code ever. It's slow, it's by Google (that is reason enough not to want to use it - they produce crap - whether it's a website or a Java OS layer), they have shit policies (killing projects overnight that aren't strictly GPL), and even their web client frontend stuff works horribly compared to Github.

There has been an Issues page for over two years (or even three years - RetroArch previously existed as SSNES for the better part of 1 and a half years) -

https://github.com/Themaister/RetroArch/issues?state=open