I have some old game music that's really good, but super lowQ.
They have characteristics like this:
Duration : 1:53.731 (2 507 776 samples)
Sample rate : 22050 Hz
Channels : 2
Bits per sample : 4
Bitrate : 176 kbps
Codec : Microsoft ADPCM
Encoding : lossy
<DECODED_BITSPERSAMPLE> : 16
So I wonder if it can be improved.
I'm not sure if it's permitted to attach an example.
I'm not sure if it's permitted to attach an example.
https://hydrogenaud.io/index.php/topic,3974.html
I've never tried it , but I've heard of people using "regular" noise reduction (with Audacity, etc.).
But of course, you have to save in a higher bit-depth than the original or the quantization noise will come back.
I'm not sure if it's permitted to attach an example.
https://hydrogenaud.io/index.php/topic,3974.html
That's too much of an unpleasant read for tonight. A simple yes/no would have sufficed. :)
I've never tried it , but I've heard of people using "regular" noise reduction (with Audacity, etc.).
But of course, you have to save in a higher bit-depth than the original or the quantization noise will come back.
I see. Maybe I'll make an attempt.
22.05 kHz sampling rate means that everything above about 10 kHz is gone.
That coupled with 4 bit ADPCM means lots of quantization noise.
Post a short (<30s) sample please.
>A simple yes/no would have sufficed.
No, it wouldn't have.
If you can't be bothered to read and follow our rules then you shouldn't be posting.
22.05 kHz sampling rate means that everything above about 10 kHz is gone.
That coupled with 4 bit ADPCM means lots of quantization noise.
Post a short (<30s) sample please.
Okay. Here's a sample.
>A simple yes/no would have sufficed.
No, it wouldn't have.
There's nothing in particular related to attachments though.
22.05 kHz sampling rate means that everything above about 10 kHz is gone.
That coupled with 4 bit ADPCM means lots of quantization noise.
Post a short (<30s) sample please.
Okay. Here's a sample.
The fact that it repeats the same 0.2 seconds of music dozens and dozens of times seems to be the most obvious fault. You're not going to have much luck trying to filter out that sort of thing.
You could try something like:
sox green04.wav output.wav lowpass 6k
The fact that it repeats the same 0.2 seconds of music dozens and dozens of times seems to be the most obvious fault. You're not going to have much luck trying to filter out that sort of thing.
What's repeated? Don't understand what you're talking about.
You could try something like:
sox green04.wav output.wav lowpass 6k
I've been experimenting with lowpass in audacity. It helps a bit. But takes much of the music with it.
mzso
Where was this music taken from?
mzso
Where was this music taken from?
Desperados: Wanted Dead or Alive
I know how this format sounds since I can save files in this format. I would say it is hopeless. A noise reduction filter requires a stable noise profile to work otherwise it will simply reduce both the noise and the signal, or create excessive pre and post echo, which damage percussive sound elements and produce unnatural fade of reverb and other artifacts which I don't know how to describe. The noise that ADPCM produces are not as stable as those found in tape hiss or low bit depth uncompressed PCM. The author of wavpack in this forum should be able to explain the nature of ADPCM in great detail.
Try to find the OST (if released) or go to websites like ocremix to see if someone made a close enough transcription of it, or visit vgmusic to find a MIDI version (if available) and listen to it with the soundfont you like. If you really love that music so much and cannot find an acceptable version, learn some music and learn to use a DAW and transcribe it yourself. I am not teasing you because I really did such things in the past.
The fact that it repeats the same 0.2 seconds of music dozens and dozens of times seems to be the most obvious fault. You're not going to have much luck trying to filter out that sort of thing.
What's repeated? Don't understand what you're talking about.
Everything after the end of the fade-in, or about 4 seconds in, is repeated again and again at a 5 Hz rate (in Cool Edit 2.1). Based on your comment I also tried it out with Audacity, and there it was just a 4 second long file. So, the repeat was an artifact of CEP 2.1.
My bad, sorry.
You could try something like:
sox green04.wav output.wav lowpass 6k
I've been experimenting with lowpass in audacity. It helps a bit. But takes much of the music with it.
I'm under the impression that the biggest fidelity gains can be obtained by playing with the methodology used to encode it.
My analysis of the file suggests that it was dithered (good), but that the dither was spectrally flat.
The next logical step would be the use of perceptually spectrally shaped dither (dither emphasized more at high frequencies) during the encoding, if the encoding is under your control.
If the encoding is out of your control, then rude approaches like spectral filtering (perhaps a variable low pass filter that increases its action at lower levels) is your best option.
It may need to be re-emphasized that spectrally filtering out quantization noise is
not how these things are usually handled.
I'm now trying to contact the composer himself, since I remember I liked this music too.
Downloaded your file and fiddled with it somehow.
There's nothing in particular related to attachments though.
-> 9.
Users found in violation of this rule will be subject to immediate punishment. Properly cited short music clips (no longer than necessary and under 30 seconds) for codec testing purposes are allowed.
Discussions about illegal filesharing sites/forums and ripping groups are not wanted here and can lead to a warning or worse.
Everything after the end of the fade-in, or about 4 seconds in, is repeated again and again at a 5 Hz rate (in Cool Edit 2.1). Based on your comment I also tried it out with Audacity, and there it was just a 4 second long file. So, the repeat was an artifact of CEP 2.1.
My bad, sorry.
I think the file might have gotten corrupted on your side. It's actually a 30s sample.
I'm now trying to contact the composer himself, since I remember I liked this music too.
To what end?
Downloaded your file and fiddled with it somehow.
It's a bit better. What'd you do with it?
Downloaded your file and fiddled with it somehow.
It's a bit better. What'd you do with it?
It seems that the MS ADPCM format of your file is a bit different from mine. Also your 30s sample doesn't have a lot of high frequency so that I can reduce the noise a bit by using Adobe Audition 1.5's hiss reduction without capturing a noise profile. After that I added a bit of bright reverb to compensate the damage of reverb continuity caused by the hiss reduction effect.
But you should not be too optimistic about this processing method since it doesn't work with more complex audio contents.
Just hope the composer can make a positive reply.
To what end?
I wrote to a company which is supposedly linked to Giovanni Vindigni (the composer). They process requests within 24-48 hours.
http://www.two-notes.com/en/artists/giovanni-vindigni-2/
To what end?
I wrote to a company which is supposedly linked to Giovanni Vindigni (the composer). They process requests within 24-48 hours.
http://www.two-notes.com/en/artists/giovanni-vindigni-2/
I see, but you didn't say the reason. Do you want to acquire a higher quality master of this or other music from him?
How about this:
http://wayback.archive.org/web/20160331194525/http://www.vindigni.de/kontakt/index.html
or this:
http://wayback.archive.org/web/20131208103549/http://vindigni.de/kontakt/index.html
To what end?
I wrote to a company which is supposedly linked to Giovanni Vindigni (the composer). They process requests within 24-48 hours.
http://www.two-notes.com/en/artists/giovanni-vindigni-2/
I see, but you didn't say the reason. Do you want to acquire a higher quality master of this or other music from him?
How about this:
http://wayback.archive.org/web/20160331194525/http://www.vindigni.de/kontakt/index.html
or this:
http://wayback.archive.org/web/20131208103549/http://vindigni.de/kontakt/index.html
Yeah, I wanted to get higher quality soundtrack from that game. Thanks for the links!
Yeah, I wanted to get higher quality soundtrack from that game. Thanks for the links!
Hi! Did you succeed in contacting him?
I just remembered this old thread.
I wonder... Is there something like an "AI upscale" sort of thing nowadays that can improve the sound of low-fi audio talked about here?
(I would expect AI to work well on something like this with acoustic music, since there aren't an infinite variety of instruments. The question is whether anyone worked on something like this.)
I just remembered this old thread.
I wonder... Is there something like an "AI upscale" sort of thing nowadays that can improve the sound of low-fi audio talked about here?
(I would expect AI to work well on something like this with acoustic music, since there aren't an infinite variety of instruments. The question is whether anyone worked on something like this.)
Another thing which would be interesting in this topic is AI assisted audio compression. Instead of coding actual high bandwidth audio data, represent it as parametric data as much as possible, and let the AI "dream" the audio back.
Voice codecs seems to already benefit from this approach, eg. Google's WaveNET, Lyra codecs.
I just remembered this old thread.
I wonder... Is there something like an "AI upscale" sort of thing nowadays that can improve the sound of low-fi audio talked about here?
(I would expect AI to work well on something like this with acoustic music, since there aren't an infinite variety of instruments. The question is whether anyone worked on something like this.)
Another thing which would be interesting in this topic is AI assisted audio compression. Instead of coding actual high bandwidth audio data, represent it as parametric data as much as possible, and let the AI "dream" the audio back.
Voice codecs seems to already benefit from this approach, eg. Google's WaveNET, Lyra codecs.
Is bandwidth still that costly to be worth dealing with such? For 100-150 kbps you can get near perfect stereo audio.