Rough pop/rock vocals

[tengiz]

You could try moving some absorbers from the side wall and on to the wall behind you if you don't have anything there already.

Back wall is already well covered. I can replace the side panels (4'x4' diffusers/absorbers) with just diffusers, which I would need to buy, and then I would need to do some damage to the walls as it's all glued .

But I will spend some time with the room as it is now. It already is much better with respect to the original issue that started this discussion.

 

[sigbergaudio]

Back wall is already well covered. I can replace the side panels (4'x4' diffusers/absorbers) with just diffusers, which I would need to buy, and then I would need to do some damage to the walls as it's all glued .

But I will spend some time with the room as it is now. It already is much better with respect to the original issue that started this discussion.

Probably no need to ruin your side wall stuff. Do you have anything on the floor (like a rug) or ceiling?

 

[Chris A]

...But I will spend some time with the room as it is now. It already is much better with respect to the original issue that started this discussion.

Now that you've spent a fair amount of time with tweaking your setup, it might help to see the amount of clipping on the track of interest (ABBA's Should I Laugh or Cry, from the YT version, converted to MP3):

No tweaking of your setup that I'm aware of will fix ~3 dB of clipping (the red vertical lines) across the track (among other issues not quite so readily apparent--like the apparent use of an Aphex Aural Exciter on the vocal stem that was mixed in--which significantly boosts harmonics and alters the phase response of the vocals).

If you want to hear the track declipped, bring it into Audacity. select the entire track, run Clip Fix from the Effects menu, then select Normalize from the Effects menu (Volume and Compression sub-menu). I got somewhere between 2.39-2.99 dB of clipping that is undone using these commands. That amount of clipping should sound bad on any reasonably hi-fi setup.

Chris

 

[Chris A]

BTW: this is a pretty common problem (clipping) on many popular music tracks.

As time goes on (from the 1990s-2010s--the approximate years of release of that version of the ABBA track), the mastering guys start to use other clipping tools that can't be undone using Clip Fix. To fix those later-mastered tracks, more sophisticated tools must be brought to the problem space--with less effectiveness.

Chris

 

[Chris A]

Here is an excerpt from the Dynamic Range Database that shows the versions of this album that have this clipping issue (1997-2005):

Chris

 

[tengiz]

Probably no need to ruin your side wall stuff. Do you have anything on the floor (like a rug) or ceiling?

I think I'm finally going to replace the seating and get something with an adjustable, removable, or foldable headrest—or something along those lines.

I have absorbers on the ceiling covering the first reflections. And just normal carpet on the floor.

 

[Chris A]

Here is an importable Audacity "filter curve" EQ preset file that I used to suppress some of the excessive female vocal harmonics that were impressed onto the track, and to recover deeper bass attenuated during mastering (attached below). You can run this EQ curve after clip-fix, then "Normalize" to reduce the overall gain to below clipping levels.

The track is barely listenable after these changes (Clip Fix, and Filter Curve EQ to suppress Aphex-type Aural Exciter damage).

Chris

 

[tengiz]

Now that you've spent a fair amount of time with tweaking your setup, it might help to see the amount of clipping on the track of interest (ABBA's Should I Laugh or Cry, from the YT version, converted to MP3):

View attachment 453243

No tweaking of your setup that I'm aware of will fix ~3 dB of clipping (the red vertical lines) across the track (among other issues not quite so readily apparent--like the apparent use of an Aphex Aural Exciter on the vocal stem that was mixed in--which significantly boosts harmonics and alters the phase response of the vocals).

If you want to hear the track declipped, bring it into Audacity. select the entire track, run Clip Fix from the Effects menu, then select Normalize from the Effects menu (Volume and Compression sub-menu). I got somewhere between 2.39-2.99 dB of clipping that is undone using these commands. That amount of clipping should sound bad on any reasonably hi-fi setup.

Chris

Wow! This makes my eyes bleed

The reason I specifically chose Should I Laugh or Cry is because, for the first 40 seconds or so, the track isn’t loud—just some guitar, synth, vocals, and effects. In the image you captured, the first instance of clipping appears to be around 00:45, when Frida hits that first loud sustained note. But before that, there’s no visible clipping, right?

Still, I hear that “rusty metal” quality in her voice starting with the very first note, around 00:20. So it doesn’t seem to be caused by obvious clipping.

That said, I’m totally with you—especially when you go back to earlier ABBA recordings, even from before the loudness war era. There are definitely a lot of artifacts in the vocals that sound like either direct clipping or effects that to me sound like they mimic that kind of distortion.

 

[sigbergaudio]

I think I'm finally going to replace the seating and get something with an adjustable, removable, or foldable headrest—or something along those lines.

I have absorbers on the ceiling covering the first reflections. And just normal carpet on the floor.

Overall it sounds like your room is more than adequately treated, so we're probably back to a combination of speakers and possibly your tinnitus.

 

[Chris A]

I found that the clipping was about 25-50% of the problem, while the synthesized female voice harmonics from what looks like an Aphex Aural Exciter constitute the remainder of the problem.

You can pull the 3200 and 4600 Hz notch valley's down a bit more from where I set them to make the resulting overall track more listenable.

This is a pretty damaged track--as they go (IME). Fortunately, the higher frequency female vocal Aural Exciter frequencies seem to be pretty well contained in those two bands, so the overall effect of attenuating those two narrow bands does not damage the overall listening experience. Here is the resulting cumulative spectral density map for the track after these operation, showing the ~1/f downward slope that's characteristic of "pink-noise type" music spectra (-17 dB/decade slope).:

Chris

 

[tengiz]

Here is an importable Audacity "filter curve" EQ preset file that I used to suppress some of the excessive female vocal harmonics that were impressed onto the track, and to recover deeper bass attenuated during mastering (attached below). You can run this EQ curve after clip-fix, then "Normalize" to reduce the overall gain to below clipping levels.

View attachment 453302

The track is barely listenable after these changes (Clip Fix, and Filter Curve EQ to suppress Aphex-type Aural Exciter damage).

Chris

I just tried this with the version available through Amazon Music downloads. I'm using Audacity for the first time, so I might not be doing it exactly as you suggested—but at first glance, when I open it in Audacity, it looks clean. I mean, there doesn’t seem to be much clipping happening here.

I also listened to the track before and after applying the EQ filter you shared. Alas, my ears aren't what they used to be—I can't say I hear any meaningful difference in the metallic/robotic quality of the voice.

Is it possible in Audacity to switch between versions of the track while playing it? Maybe then it will be more apparent. But I can't find a way to do it.

 

[Mart68]

I just tried this with the version available through Amazon Music downloads. I'm using Audacity for the first time, so I might not be doing it exactly as you suggested—but at first glance, when I open it in Audacity, it looks clean. I mean, there doesn’t seem to be much clipping happening here.



I also listened to the track before and after applying the EQ filter you shared. Alas, my ears aren't what they used to be—I can't say I hear any meaningful difference in the metallic/robotic quality of the voice.

I don't think you're going to hear any difference because that's just how they've processed her vocal in that opening 'salvo.'

Clearly you are not playing the later mastering with the heavy compression but I think the fact is - she's meant to sound like that.

 

[tengiz]

I don't think you're going to hear any difference because that's just how they've processed her vocal in that opening 'salvo.'

Clearly you are not playing the later mastering with the heavy compression but I think the fact is - she's meant to sound like that.

Yes, it seems so. And at least some of her solo work—like the track below—has the same vocal quality. She seems to like that sound.

 

[Chris A]

I just tried this with the version available through Amazon Music downloads. I'm using Audacity for the first time, so I might not be doing it exactly as you suggested—but at first glance, when I open it in Audacity, it looks clean. I mean, there doesn’t seem to be much clipping happening here.

View attachment 453311

Then the version you have is one of the other versions that I showed in post #85, above. That's good. The most-replayed version off YT is the one I showed, and it's pretty badly clipped--to the point where it's audible.

This is why when you mention a particular track that you either like or that's got apparent problems, you really need to identify the exact place and/or version of the track you're using. The different releases of a single recording I've found can vary by as much as 10 dB (dynamic range--in terms of crest factor). Usually, the most offending recordings start to occur after a release date of 1991, when multi-band digital compressors hit the market. I use the DR Database to find the best (highest dynamic range) CDs/DVDs that are available, and avoid those versions that are heavily "remastered" (i.e., compressed and clipped to make them louder). The best CDs/DVDs are usually the oldest versions available (for music originally recorded before 1991). Then I "demaster" them to remove most of the old bass attenuation and other mastering EQ that shows up in the cum spectral density plot.

I also listened to the track before and after applying the EQ filter you shared. Alas, my ears aren't what they used to be—I can't say I hear any meaningful difference in the metallic/robotic quality of the voice.

Is it possible in Audacity to switch between versions of the track while playing it? Maybe then it will be more apparent. But I can't to to find a way to do it.

Here is one way (that I haven't tried): https://forum.audacityteam.org/t/a-b-comparison/44905/2

I simply play the resulting track, then hit "undo" and play again, then revert back to the resulting track using ctrl-Y to play the edited track again. This takes a second or so to hear the differences.

Chris

 

[MarkS]

Have you tried listening to the track over good headphones?

 

[tengiz]

Have you tried listening to the track over good headphones?

The best I have is old Shure SE420. It sounds less "metallic", but that sound quality is still there. But then again, it's so different from full size speakers, among other things, I cannot shake the perception that the sound stage is between the ears.

 

[ThatSoundsGood]

Wow! This makes my eyes bleed

The reason I specifically chose Should I Laugh or Cry is because, for the first 40 seconds or so, the track isn’t loud—just some guitar, synth, vocals, and effects. In the image you captured, the first instance of clipping appears to be around 00:45, when Frida hits that first loud sustained note. But before that, there’s no visible clipping, right?

Still, I hear that “rusty metal” quality in her voice starting with the very first note, around 00:20. So it doesn’t seem to be caused by obvious clipping.

That said, I’m totally with you—especially when you go back to earlier ABBA recordings, even from before the loudness war era. There are definitely a lot of artifacts in the vocals that sound like either direct clipping or effects that to me sound like they mimic that kind of distortion.

I think you're hearing the Eventide Harmonizer.

 

[ThatSoundsGood]

Now that you've spent a fair amount of time with tweaking your setup, it might help to see the amount of clipping on the track of interest (ABBA's Should I Laugh or Cry, from the YT version, converted to MP3):

View attachment 453243

No tweaking of your setup that I'm aware of will fix ~3 dB of clipping (the red vertical lines) across the track (among other issues not quite so readily apparent--like the apparent use of an Aphex Aural Exciter on the vocal stem that was mixed in--which significantly boosts harmonics and alters the phase response of the vocals).

If you want to hear the track declipped, bring it into Audacity. select the entire track, run Clip Fix from the Effects menu, then select Normalize from the Effects menu (Volume and Compression sub-menu). I got somewhere between 2.39-2.99 dB of clipping that is undone using these commands. That amount of clipping should sound bad on any reasonably hi-fi setup.

Chris

Where did you get this track to find out it was clipping? Can you tell me your process here? Is it from Spotify, Apple Music?

 

[MarkS]

The best [headphone] I have is old Shure SE420. It sounds less "metallic", but that sound quality is still there.

So it's on the recording. What else do you need to know?

 

[tengiz]

So it's on the recording. What else do you need to know?

At this point - I am done indeed. With the help of the forum: it basically took not letting Audyssey apply its filters above 0.5k for the main speakers.