Using DSP to flatter lower quality recordings?

[SAC_2]

We all know how to use DSP to get the most accurate signal out of whatever setup we have. (Or do we?)

But accuracy can be bad if you listen to recordings of a wide range of mastering quality. Going from anything Steely Dan to Ahmad Jamal's The Awakening is flat out jarring. Is it possible to tune a DSP to narrow the bandwidth of sonic variation in a way that sacrifices accuracy but makes everything sound decent to good? Is this even possible to do with the parameters a DSP is limited to?

 

[DVDdoug]

DSP (Digital Signal Processing) can do LOTS of things... It can EQ (boost/cut bass or treble, etc.), it can dynamically compress, or add reverb or echo, etc. So it depends on what you want to do. I'm not familiar with Ahmad Jamal...

You can't always fix-up a bad recording or a bad production. In fact, MOST recording/production defects can't be fixed... That's why pros record in soundproof studios with good equipment, and a good setup, etc. And of course, you need a good performance and there are generally multiple takes in order to get a good one or to "comp" together parts of different takes.

 

[RayDunzl]

I have only one commercial CD where I might eventually adjust its "curve".

On the other hand, Audio Buddy has some ancient recordings of bands he was in recorded on various budget devices back in the day, and they can be improved with some frequency response, noise reduction, and (often) speed adjustments.

If you want very fine control, the ability to use an FIR filter is probably the way to go.

Example. One FIR filter:

Example is "room correction" but shows how FIR can pinpoint any number of frequencies for adjustmant.

I don't have the "power" to adjust the low frequencies via FIR, so that job is passed to the IIR, along with "broad stroke" adjustments, via biquad filter data:

Showing the effect of only one "biquad", the others bypassed:

Add those two corrections together, and that is what gets applied to the signal going toward the speakers.

Any questions?

Me too, but that's about my limit of knowledge in this area.

 

[Frgirard]

We all know how to use DSP to get the most accurate signal out of whatever setup we have. (Or do we?)

But accuracy can be bad if you listen to recordings of a wide range of mastering quality. Going from anything Steely Dan to Ahmad Jamal's The Awakening is flat out jarring. Is it possible to tune a DSP to narrow the bandwidth of sonic variation in a way that sacrifices accuracy but makes everything sound decent to good? Is this even possible to do with the parameters a DSP is limited to?

You cannot fixe the distortion due to the high compression.
You cannot change the dynamic range
All you can is to try to reduce the level of bass or other part of the bandwidth.
You cannot change the sound stage


But

The Joconde color is not good for you, do you go to the Louvre to paint the Joconde.

So boycotte the bad recording

 

[LTig]

You can change the soundstage. There are units which widen the stereo width, some of them can also narrow it.

 

[Frgirard]

You can change the soundstage. There are units which widen the stereo width, some of them can also narrow it.

https://www.masteringthemix.com/blo...ting-depth-and-height-in-the-mix?currency=EUR

 

[abdo123]

Might as well become a DJ at that point

 

[Grumpish]

Why not, sound engineers are using whole arsenals of DSP's to flatter mediocre performers on a daily basis. On the other hand, as they say.... garbage in garbage out.

 

[Frgirard]

the easiest way is not to buy.
when the consumer takes back power, he will have what he asks for: a job well done

 

[L5730]

You can sort of stuff everything through multiband compression and limiting and make it all homogenous, kind of simulating radio play. Does it sound 'better'? It sounds consistent I guess.
Check out JBBroadcast VST plugin, made by the guy behind ToneBoosters.
Also Reaper multiband compressor.

Orban or something rather brutal is what the radio stations use. Fast and slow compressors, limiting and all sorts. Still, it's been shown that better in = better out, even with that level of sonic mangling.

 

[DVDdoug]

I have (at least) one "pop" recording that has over-powering one-note bass so when I listen to it (rarely, and usually in the car) I turn-down the bass (Not actually digital signal processing).

And I've digitized a few older vinyl records that sound a little "dull" so I boosted the highs. In that case it was DSP (GoldWave or Audacity). Most older recordings released on CD don't have that problem but it was common on the original records. And of course I used "DSP" for click and pop reduction/removal, and other noise reduction.

 

[abdo123]

And I've digitized a few older vinyl records that sound a little "dull" so I boosted the highs.

this is quite strange to me since Bass is usually the first to go in a worn out record.

 

[Jimbob54]

the easiest way is not to buy.
when the consumer takes back power, he will have what he asks for: a job well done

Personally , I'd rather hear a badly recorded but exciting new artist/ album than any number of well recorded albums by established artists treading water. Of course, I would really have the exciting new artist produced well, but thats not the way of it.

Now , there is a special place in hell for established artists and labels that insist on releasing shoddily produced/ mastered records despite knowing it will sell millions. At least try and polish the turd.

 

[Tim Link]

Maybe some day machine learning software will be able to analyze a recording and completely reconstruct it to sound dramatically better. It could theoretically fix even compression, clipping, severe distortion, just about everything. I'm told that software is already being used for auto mastering work. It's cheap and taking work away from professionals. I'm also told it's not that good. I don't expect to be amazed anytime soon based on how unamazing Adobe's latest "Super Resolution" addition to Photoshop is. Really good software would recognize what the picture is and make up fake information that is highly realistic and convincing, and hopefully really good looking. The same would have to be done audibly to fix bad recordings.

 

[goat76]

I really think it's a bad idea to start doing your own personal "masterings" of individual songs and albums, it's probably better for the peace of mind to just find a good sweet spot/your global settings where most of your music collection sounds fairly well balanced, and just leave it there. Most of us will probably never stop altering the masters and lose the focus of the music listening, especially when we start noticing we probably want different settings from day to day, and even for different volume levels for the same album.

The most common problem with some records is that they can sound somewhat thin, but if the music is good you will be acclimated to the album's balance after the first song or two. If not, it's maybe not a very good album and you should probably swap it out for something better anyway.

 

[L5730]

.... Really good software would recognize what the picture is and make up fake information that is highly realistic and convincing, and hopefully really good looking. The same would have to be done audibly to fix bad recordings.

Adobe Photoshop's "Content Aware Fill" has done a darn good job filling in missing information for me. I was supplied with a couple of photos that didn't make for a great photo by themselves. One took the top left area, the other bottom right, overlapping some considerable amount in the middle, but not enough for a decent crop. Content aware fill made new rock formations in those two blank corners - yes new, they weren't cloned! The edge matching was even pretty darn solid too. Managed to salvage those two into one decent image. It's also useful if an image needs to be perspective tilted a little, fill in the triangles down the sides - though this is less noticeable if there are errors and one could just manually clone their way out of it fairly easily.

 

[dasdoing]

you could process all your music with something like this https://www.sonible.com/smarteq3/
some will sound worse, some better....but it would do what you want: they will be more alike

 

[SAC_2]

I really think it's a bad idea to start doing your own personal "masterings" of individual songs and albums, it's probably better for the peace of mind to just find a good sweet spot/your global settings where most of your music collection sounds fairly well balanced, and just leave it there.

This is what I'm looking for. I don't have the time or means to do that kind of individual mastering. Just looking to see what I can do after the "accuracy" tune of the DSP to tune to taste.

 

[Tim Link]

Adobe Photoshop's "Content Aware Fill" has done a darn good job filling in missing information for me. I was supplied with a couple of photos that didn't make for a great photo by themselves. One took the top left area, the other bottom right, overlapping some considerable amount in the middle, but not enough for a decent crop. Content aware fill made new rock formations in those two blank corners - yes new, they weren't cloned! The edge matching was even pretty darn solid too. Managed to salvage those two into one decent image. It's also useful if an image needs to be perspective tilted a little, fill in the triangles down the sides - though this is less noticeable if there are errors and one could just manually clone their way out of it fairly easily.

Yes, that does work very well. The upsampling is less impressive, although on some content it can seem to pull off a small miracle, based on some examples I've seen.

 

[escksu]

We all know how to use DSP to get the most accurate signal out of whatever setup we have. (Or do we?)

But accuracy can be bad if you listen to recordings of a wide range of mastering quality. Going from anything Steely Dan to Ahmad Jamal's The Awakening is flat out jarring. Is it possible to tune a DSP to narrow the bandwidth of sonic variation in a way that sacrifices accuracy but makes everything sound decent to good? Is this even possible to do with the parameters a DSP is limited to?

Yes, we can. But there is still a limit to what we can do and its still not going to entirely overcome the issues associated with a bad recording.