You can even make 16KHz 4-bit ADPCM recordings sound like a CD-quality track.
For example, you can use an exciter, which interpolates the low-to-mid range frequencies into the high range spectrum, or simply resample it with no interpolation, and manually filter out the aliased frequency bands.
Personally I believe kode54 is right. As someone who obesses over having as highest quality as possible it really bugs me to side with kode, but I'd rather have the original low-quality version rather than knowing it's been processed (because who's to say their version is 100% accurate?)
That's not just me defending my objectively brilliant remasters at all, nope, not at all ðŸ˜
Honestly the only part that bugs me about remastering is when it is labeled as such with no mention of it being a fan-made thing.
A great demonstration of this is with a game like Goldeneye 007, where the USF files play at about 22 kHz, yet if you play the game on a real Nintendo 64 it sounds better and has frequencies well into the ultrasonic range. It doesn't sound as good as if the music were natively at a higher sampling rate, of course, but it's still an improvement.
I find that if you use the MultiResampler DSP in foobar and set it to cubic interpolation, you get a good approximation of N64, Wii (I assume GameCube as well but haven't specifically tested), and 3DS resampling. For PS2 games with 22 or 24 kHz streams, they come out a little more muffled on real hardware than with any of MultiResampler's interpolation options because the PS2 uses an interpolation method not offered.
Note that the Wii and 3DS appear to use cubic interpolation to get everything to 32 kHz for their software mixing stages, and then evidently output with sinc interpolation applied, which allows alias frequencies until about 18 kHz for both systems. This is true even for Wii games played on a Wii U and captured digitally from the HDMI output, so it has to be done in software rather than with a filter in the audio output circuit. But when you're listening to ripped files, you might as well just set MultiResampler to do one pass with cubic interpolation to your target output rate.
Lanczos resampling tends to sound (and look) the best, but if you want those high frequencies back, you'll need to give up all interpolation altogether.
Most people don't understand much about the fourier domain, so I won't bore people with the details.
Lanczos interpolation is a form of sinc interpolation and can tend to make things like drum and cymbal samples on those systems too muffled. If you want high frequencies that sound closer to intended without distortion, cubic interpolation is a good idea. Linear interpolation is only subtly different as well.
@MoldyPond: Indeed. But I have a pedantic note about that. kode54 added that output option to foobar's USF decoder after I discussed this stuff with him a while back. He said at the time it applies the same cubic interpolation scheme that is used in the N64's mixing stage when dealing with samples, but I found that the spectogram isn't quite as similar to that of a real N64 when compared to using MultiResampler's cubic interpolation. ¯\_(ツ)_/¯
You are right though about it still not being right at 48kHz. Before Diddy Kong Racing's USF was fixed a few months ago I had a new line-in recording and listening back to it, it still sounds slightly better quality. Not quite as clean and raw as the USF, but more... right.
Still though, having the ability to export a 50 song soundtrack as WAV files in 2 minutes as opposed to having to record every song individually is something I will never complain about :D
The best method, in my opinion, is to not interpolate it, and then to manually filter all the aliasing artefacts out.
It's a huge effort, but nothing will ever come close to sounding as good as that.
Good luck with that though :(
Guess what I did!
One of my objectively brilliant remasters!
Here's an example, Ancient Lake recorded from the sound test menu of DKR, then painstakingly remastered:
Link
@MoldyPond: It would be really awesome if somebody could somehow figure out exactly what a real Nintendo 64 does so it could be implemented. I'll say though that I abandoned my idea of recording Goldeneye's music from hardware when I A/B tested it versus the USF upsampled with MultiResampler and found a difference so tiny I decided I'd be doing all that work for truly no actual gain. With cubic interpolation, it really is extremely similar to what the real thing sounds like.
@simonkwii: GBA and DS music can use instrument samples with different sampling rates, making individual instruments alias at different frequencies, so there's not really a one-size-fits-all-instruments point to start filtering. But if you're filtering out all the aliasing artifacts, isn't that functionally the same as sinc interpolation?
Additionally, I would argue that while linear interpolation, and cubic for that matter, have audible aliasing artifacts, that's the point when the samples are so bandlimited in the first place. The interpolation is used to synthesize treble frequencies that aren't there to improve the sound while saving on memory or CPU usage. When a cymbal sample is only 8 kHz or something, I think the composer or sound designer is counting on aliasing to create the treble associated with a cymbal crash. But of course you can do whatever you think sounds best; that's what I'm doing too and I don't think we have to stick to what the actual hardware does. Nintendo doesn't either. The EQ applied to the audio for GBA games in the Game Boy Player isn't at all like what a real GBA sounds like, but I think it's quite good. Also, their releases of Famicom music never sound like they were recorded from a standard Famicom's only output method: RF.
I don't mean to try to contradict you on your Diddy Kong Racing work, but that track has about 3 dB more dynamic range if you turn the music volume slider down in the game. It has to be all the way down to about 25% for it not to clip! I also think the reverb at the end could be dialed back or even eliminated entirely and it would sound fine. But I do think EQing the treble to boost it is a good idea and makes a big difference.
That being said I've actually asked about that in the USF request thread. I let JoshW know about the issues with the old USF set of DKR and he fixed and reuploaded it, so now it's perfect except really really loud and you can easily hear the clipping if you don't upsample it. When creating USFs is there a volume setting of any sort?
You can even make 16KHz 4-bit ADPCM recordings sound like a CD-quality track.
For example, you can use an exciter, which interpolates the low-to-mid range frequencies into the high range spectrum, or simply resample it with no interpolation, and manually filter out the aliased frequency bands.
Purely snake oil, nothing more.
LOL nice meme
edited 8:51 PM EDT May 4, 2018
The Ancient Lake "remastered" track is not of higher quality either. One can easily prove this by examining its spectrogram. Everything above 11000 Hz is mirrored. This is the aliasing effect. What we hear as high frequencies are ringing artifacts of the interpolation method used (cubic?). Using Whittaker–Shannon interpolation should avoid those artifacts and give you the same result as the console output, just at higher sample rate.
The only chance to get higher quality tunes is to re-synthesize them. Only possible for games that use sequenced audio ofc. Since the instrument samples are compressed themselves, the gain may be marginal. Doing this for DKR's Ancient Lake this looks / hears the following (using fluidsynth with 7th order sinc interpolation for all instrument samples):
https://drive.google.com/open?id=14XMTot2eI61DB4QiGk-D70Rs1TiW8CIG
Please.
But I can't say quality can't be improved when, despite sound quality being an almost entirely subjective thing so there isn't any one correct answer, I think most people would agree that Diddy Kong Racing sounds better cubic interpolated (which is the way it sounds on real hardware because that's what real hardware does, contra your statement in the last sentence of your second paragraph) than with Whittaker–Shannon interpolation and lacking any frequencies above 11 kHz. Yes, they're aliasing artifacts, but that doesn't mean that they're unpleasant or that it sounds worse.
I like to add some distortion to my electric guitar sometimes, which creates overtones that aren't present in the pre-processed signal to achieve a desired sound; that's functionally equivalent to a Nintendo 64, GameCube, Wii, or PlayStation 2 upsampling with interpolation to create treble frequencies that wouldn't otherwise exist.
Evidently console designers have decided that resampling with interpolation methods (other than sinc) does improve sound quality because they've made the conscious decision to do just that.
Like I said in a previous post, sure it's not as good as natively using the higher sampling rate in the first place, so re-synthesizing should sound better as a general rule. But Star Fox 64 and Ocarina of Time's soundtrack albums seem to have been recorded pre-conversion to N64 and I really kinda prefer how they sound in the game.
I think the main problem with the arguments here is that the title asks to "improve" low-quality audio (which in and of itself is completely subjective). Just because you've taken a low-quality source and processed it to sound "better" doesn't mean it actually is. Data lost is data lost, no exceptions, but that also doesn't mean you can't make it sound better than the low-quality version.
I'd say the better argument here is Purist Audio Vs. Enhanced Audio.
In response to the original question, yes there are (many) ways to improve low-quality audio, but it raises the extra question of "Knowing that it's been processed externally by a random person with no affiliation to the original source material and this edit is not official in any way whatsoever, is that a concern to you at all?"
edited 5:52 PM EDT May 11, 2018
However I don't see evidence that console designers consciously use resampling to improve sound quality. I only see that software synthesizers need resampling at instrument sample level. Why they haven't chosen sinc? Because it's too expensive. So they've chosen something cheaper, at the cost of arising artifacts... whether intended or not idk.
MoldyPond is right: Purist Audio lovers should just be happy with what a console natively outputs. Everybody else may subjectively "enhance" audio as he pleases.
@MoldyPond: To me, yes it is.
No interpolation method does anything to restore the missing frequencies from the sample/track. All it does is removing the aliasing artifacts from the harmonics generated when samples go unfiltered. Any interpolation method other than Linear is leaving aliasing artifacts on the track.
There is this common thing that people mistakenly call any non Linear method in audio interpolation as higher quality (Sinc, Catmull, Cosine etc). The only real high quality is Linear. We are not dealing with image files here. Any artificially generated audio harmonics from those algorithms result in a fake and metalic sound that sounds identical to what you hear from nearest neighbor (no interpolation). The frequencies that are gone, are gone. Exciters and other stuff do nothing proper to restore those frequencies, they just add more artifacts that result into what we associate as metalic/cheap electronics sound. You don't want any of that in your quality audio tracks.
Get your facts right.
edited 6:37 PM EDT May 12, 2018
Here's an alternative fact: No resampling (interpolation) method is 100% objectively perfect.
Tidbit example:
We can say that SNES samples were designed for the Gaussian interpolation of the system. Regardless of whether Mr. Sound Man consciously knew about the ass-end (for flavor; I like SNES) output of the system, because that was the target destination, Gaussian is correct in terms of accurate reproduction of SNES. If you want mathematically accurate, then Sinc. Both create artifacts AKA something there that wasn't there before.
https://memegenerator.net/img/instances/67464703/there-must-be-something-there-that-wasnt-there-beforenoise.jpg
edited 9:03 PM EDT May 12, 2018
What the fuck?
How much piss are you taking?
GBA and DS' native output sounds like utter horseshit, and I would rather shove a fork in my dick than listen to them without interpolation, how it was "originally intended".
Fuuuuuuuuck off.
edited 7:32 PM EDT May 12, 2018
@simonmkwii: Highly professional way of argumenting, well done.
It's shitlickers and nerds united, which just creates an utter fuckfest!
I was considering posting a reply to that call for resequencing with a report of "I would suggest quoting a high price per hour for all that work, but that would imply you would do it at all, much even for a price."
@bxaimc My god, the "Ways to improve 24 khz or lower sound files using audacity/audition" thread is a total nightmare!
It's basically just people flaming each other
bxaimc - Today at 12:18 AM
You expected something different?
I didn’t
everyone that responded in that thread has their own quirky ego
This thread has become a meme
@MoldyPond There are ways to render GBA audio in higher quality that GBA's DSP. One is to convert tracks to BASS midi and the other is this: Agbplay
Also for the record here's what happens with and without Linear interpolation. Any other than Linear interpolation will introduce a level of the artifacts shown on the right side of the Spectogram since they all rely on creating those fake metalic sounding harmonics:
Image
edited 12:49 PM EDT May 14, 2018
The cool kids use chat.
Talking past each other is fun. As was my sarcasm. Right kode54?
But yes, kode54 did get it right in post #2.
There's nothing objective about subjectively enhancing source material. There's nothing wrong with it, but it's up to personal taste, and shouldn't be advertised as something more right. There's already enough confusion without that (see this thread).
https://memegenerator.net/img/instances/76097381/worse-how-can-things-get-any-worse-look-around-you-helen-were-on-the-threshold-of-hell.jpg
By the way, I've gone all this time without saying yo bx. It's been a while, but I still remember you as a kickass dude. I noticed arbingordon also popped in here some pages back. I enjoyed chatting with both of you and hcs back half a decade ago or something, hope y'all are doing well.
Also, 24kHz isn't much lower than my threshold of hearing anyway. 12kHz vs 14.5kHz.
edited 4:27 AM EDT May 22, 2018
I forgot that thing was still being worked on. I've only been keeping up with Mame qsound emulation. I hope this doesn't mean I have to relog all my qsound rips since when I tested them in mame, notes were missed.
@simonmkwii
LOL
edited 2:44 PM EDT May 22, 2018
edited 2:53 PM EDT May 22, 2018