Strategies for reducing treble harshness on 1930s tracks

It's all about the equipment

Moderators: Mr Awesomer, JesseMiner, CafeSavoy

Message
Author
Haydn
Posts: 1277
Joined: Tue Nov 09, 2004 5:36 am
Location: London

Strategies for reducing treble harshness on 1930s tracks

#1 Post by Haydn » Thu Nov 13, 2008 5:43 am

I know this has been discussed a little before in a few threads including this one and this one. I've also found a few system add-ons which help - Volume Logic for the Mac and Breakaway for Windows.

I find it's an ongoing problem with 1930s music. A lot of stuff from, say, 1935 to 1941 features horn sections, and the trebles from these often sound very harsh. The music can sound bad and even hurt the ears. A simple answer when DJing is to turn down the treble, either for the whole track, or selectively when the horns are loud. But if you turn down the treble for the whole track, you lose the 'good' treble sounds like snare drum and higher frequencies of instruments like piano and guitar, and it's hard work to keep turning it up and down during a track. What's your strategy for dealing with those harsh trebles?

SoundInMotionDJ
Posts: 154
Joined: Wed May 30, 2007 8:27 pm

Re: Strategies for reducing treble harshness on 1930s tracks

#2 Post by SoundInMotionDJ » Thu Nov 13, 2008 7:25 am

Most 1930's music has about 40-50dB of dynamic range. That is not a lot of range to try to get all the content into. Further, there was no practical way to have a home based playback system actually reproduce much in the way of either low or high frequencies - so the mastering process tended to favor the mid frequencies.

When you put music like that on a system that can actually produce low and high frequencies, you get the "harshness" in the highs that you were discussing.

That said, it is not possible with a "three knob" EQ to do much about the problem. If you have a 31band or parametric EQ, you can experiment until you find the frequency range that needs to come down to remove the harshness.

Even with modern music, I do not like a "flat response." I setup a -6dB/oct slope on the EQ above 10kHz. You might try that as a place to start.

User avatar
Cyrano de Maniac
Posts: 97
Joined: Tue Mar 21, 2006 2:11 pm
Location: South Saint Paul, Minnesota
Contact:

#3 Post by Cyrano de Maniac » Thu Nov 13, 2008 12:33 pm

I think the first thing to do in general is to become familiar with how different frequencies are perceived, and the manner in which different instruments "use" these frequencies. An article such as http://www.digitalprosound.com/2002/03_ ... cerpt1.htm can be helpful in that regard.

If you have access to a parametric EQ (i.e on any halfway decent mixing board), one useful strategy is to substantially boost the gain of the midrange, and sweep the midrange frequency knob until you find the frequency that makes things sound really bad (e.g. makes those harsh horns sound even more blaring). Then leave the frequency set to that spot, and turn the gain knob down until the music sounds as good as you can get it.

With a graphic EQ it's tougher, as searching for the offending frequency (or frequencies) is more difficult than just sweeping a frequency knob back and forth. It does have the advantage, however, that you can address multiple offending frequencies more easily, should you need to do so. And if all you have is a little 5-band (or even 10-band like in iTunes) graphic equalizer, I almost wouldn't bother -- you might get lucky, but they simply don't have tight enough frequency bands to reliably notch out offending frequencies.

Brent

Haydn
Posts: 1277
Joined: Tue Nov 09, 2004 5:36 am
Location: London

Re: Strategies for reducing treble harshness on 1930s tracks

#4 Post by Haydn » Sun Dec 14, 2008 7:46 am

SoundInMotionDJ wrote:Most 1930's music has about 40-50dB of dynamic range. That is not a lot of range to try to get all the content into. Further, there was no practical way to have a home based playback system actually reproduce much in the way of either low or high frequencies - so the mastering process tended to favor the mid frequencies.

When you put music like that on a system that can actually produce low and high frequencies, you get the "harshness" in the highs that you were discussing.
When I hear 30s music played on vinyl or shellac (which, fortunately, you can quite often hear in London), I don't notice any problem with harsh trebles - the music just sounds great, even though there is extra surface noise. Somehow, those harsh trebles are there when the music is on CD, but not there when on vinyl/shellac.

Is there any way of emulating this softer 'non-harsh' vinyl/shellac sound without actually buying a vinyl/shellac copy of the music?

SoundInMotionDJ
Posts: 154
Joined: Wed May 30, 2007 8:27 pm

Re: Strategies for reducing treble harshness on 1930s tracks

#5 Post by SoundInMotionDJ » Mon Dec 15, 2008 7:44 am

Haydn wrote:When I hear 30s music played on vinyl or shellac (which, fortunately, you can quite often hear in London), I don't notice any problem with harsh trebles - the music just sounds great, even though there is extra surface noise. Somehow, those harsh trebles are there when the music is on CD, but not there when on vinyl/shellac.

Is there any way of emulating this softer 'non-harsh' vinyl/shellac sound without actually buying a vinyl/shellac copy of the music?
Yes - turn the volume down - or play on a system that does not really put out much above about 10kHz. My guess is that the transfer to CD included a "remastering" process that compressed the music and made everything louder.

If you look at the Equal Loudness Contour, notice that the overall curve "flattens" as the sound gets louder. This means that both bass and treble "sound louder" as the volume goes up, and that the "mids" (which includes the vocal range) can get lost in the mix.

Have you run an RTA on both recordings to look at the content? Audacity has a built in RTA that is good enough for these purposes. That might help you to better understand the differences between the vinyl and CD versions.

--Stan Graves

Haydn
Posts: 1277
Joined: Tue Nov 09, 2004 5:36 am
Location: London

Re: Strategies for reducing treble harshness on 1930s tracks

#6 Post by Haydn » Tue Dec 16, 2008 8:20 am

SoundInMotionDJ wrote:
Haydn wrote:Is there any way of emulating this softer 'non-harsh' vinyl/shellac sound without actually buying a vinyl/shellac copy of the music?
Yes - turn the volume down - or play on a system that does not really put out much above about 10kHz.
Thanks, but turning the volume or treble down aren't what I had in mind :wink:

SoundInMotionDJ wrote:My guess is that the transfer to CD included a "remastering" process that compressed the music and made everything louder.

I'm thinking that the remastering onto CD has a lot to with it. One day, I would like to sit down with an engineer who actually transfers 1930s music from original media to CD and ask them to explain this in detail. There's also something about 1930s records which makes this a particular problem with songs from that eral. I've never noticed shrill trebles with 1950s songs for example.

SoundInMotionDJ wrote:Have you run an RTA on both recordings to look at the content? Audacity has a built in RTA that is good enough for these purposes.

What's an RTA :?:

User avatar
Swifty
Posts: 448
Joined: Mon Nov 25, 2002 7:53 pm
Location: NY, NY
Contact:

#7 Post by Swifty » Tue Dec 16, 2008 9:16 am

Real Time Analyzer
"Dance like it hurts. Love like you need money. Work when people are watching."

Haydn
Posts: 1277
Joined: Tue Nov 09, 2004 5:36 am
Location: London

#8 Post by Haydn » Tue Dec 16, 2008 9:25 am

Swifty wrote:Real Time Analyzer
Thanks 8)
Haydn wrote:One day, I would like to sit down with an engineer who actually transfers 1930s music from original media to CD and ask them to explain this in detail.
By the way, has anyone here had this explained to them properly by an audio engineer?

SoundInMotionDJ
Posts: 154
Joined: Wed May 30, 2007 8:27 pm

Re: Strategies for reducing treble harshness on 1930s tracks

#9 Post by SoundInMotionDJ » Tue Dec 16, 2008 5:13 pm

Haydn wrote:
SoundInMotionDJ wrote:Have you run an RTA on both recordings to look at the content? Audacity has a built in RTA that is good enough for these purposes.

What's an RTA :?:
Here is what the RTA output of Audacity looks like (I forget what song I used to make this...but it was some modern pop song with lots of bass in the section that I profiled):

Image

This shows that there is a lot of low-frequency "rumble" below 22Hz.
There is almost nothing above 14kHz.
The peak of the low frequency is at 48Hz.

Haydn
Posts: 1277
Joined: Tue Nov 09, 2004 5:36 am
Location: London

Re: Strategies for reducing treble harshness on 1930s tracks

#10 Post by Haydn » Tue Dec 16, 2008 5:58 pm

SoundInMotionDJ wrote:
Haydn wrote:
SoundInMotionDJ wrote:Have you run an RTA on both recordings to look at the content? Audacity has a built in RTA that is good enough for these purposes.

What's an RTA :?:
Here is what the RTA output of Audacity looks like (I forget what song I used to make this...but it was some modern pop song with lots of bass in the section that I profiled):

Image

This shows that there is a lot of low-frequency "rumble" below 22Hz.
There is almost nothing above 14kHz.
The peak of the low frequency is at 48Hz.
Looks interesting. If I could find a way of uploading graphics from my computer to this site, I could upload the analysis of some of the songs in my library.

Haydn
Posts: 1277
Joined: Tue Nov 09, 2004 5:36 am
Location: London

Re: Strategies for reducing treble harshness on 1930s tracks

#11 Post by Haydn » Mon Apr 06, 2009 12:15 pm

SoundInMotionDJ wrote:Most 1930's music has about 40-50dB of dynamic range. That is not a lot of range to try to get all the content into. Further, there was no practical way to have a home based playback system actually reproduce much in the way of either low or high frequencies - so the mastering process tended to favor the mid frequencies.

When you put music like that on a system that can actually produce low and high frequencies, you get the "harshness" in the highs that you were discussing.
What about radio transcription discs recorded in the 1930s? Like this one recorded in 1936 and 1937 -

http://www.allmusic.com/cg/amg.dll?p=am ... fuxq8hldke

As on nearly all transcription recordings, the sound is great and there is none of the treble harshness that affects a lot of other recordings from the time.

User avatar
Cyrano de Maniac
Posts: 97
Joined: Tue Mar 21, 2006 2:11 pm
Location: South Saint Paul, Minnesota
Contact:

Re: Strategies for reducing treble harshness on 1930s tracks

#12 Post by Cyrano de Maniac » Tue Apr 07, 2009 8:39 am

SoundInMotionDJ wrote:
Haydn wrote:
SoundInMotionDJ wrote:Have you run an RTA on both recordings to look at the content? Audacity has a built in RTA that is good enough for these purposes.

What's an RTA :?:
Here is what the RTA output of Audacity looks like (I forget what song I used to make this...but it was some modern pop song with lots of bass in the section that I profiled):

Image

This shows that there is a lot of low-frequency "rumble" below 22Hz.
There is almost nothing above 14kHz.
The peak of the low frequency is at 48Hz.
I take it this spectrogram was run against the entire song? Unfortunately that won't likely show you what you're really interested in, as it's sort of the average across every moment of the song. The 48Hz peak might be from a kick drum or bass or just a recording technology artifact.

If you think about it, generally the shrillness comes in short bursts from the horns, or less often from voices or other instruments. I'd suggest rerunning the spectrogram against only a tiny clip of the song, maybe even just a few notes, where you experience the harshness. That should give you a better idea at which frequencies the problem lies.

Even then, you run into the problem that loudness as perceived by human ears is not constant across the audio spectrum. That's evident in part by the 48Hz peak you see -- you didn't even notice it before as the treble harshness stood out. Generally you can put a lot of energy into the lows and it's not perceived as particularly loud to the ears, though at something like a rock concert you might feel it pounding in your chest. The higher frequencies require a lot less energy to be perceived as loud. The spectrogram shows an energy distribution without taking into account human perception.

Anyway, point was that you'll want to isolate a few shrill notes and re-run the spectrogram. That'll give you a better idea of where the problem frequencies are.

Surreal
Posts: 402
Joined: Fri Aug 25, 2006 3:31 pm
Location: Winnipeg, Manitoba, Canada

#13 Post by Surreal » Tue Apr 07, 2009 11:12 am

Audacity only grabs teh first 23.8 seconds (at least that's what mine tells me when I try to select anything larger).

Do any of those autocorrection functions account for perceived hearing?

Toon Town Dave
Posts: 661
Joined: Wed Nov 20, 2002 2:52 pm
Location: Saskatoon, Canada

#14 Post by Toon Town Dave » Tue Apr 07, 2009 5:02 pm

This might be a more useful tool for analyzing an audio signal:
http://www.zelscope.com/index.html

Haydn
Posts: 1277
Joined: Tue Nov 09, 2004 5:36 am
Location: London

#15 Post by Haydn » Wed Apr 08, 2009 12:49 am

Before we get too technical, does anyone actually use a Real Time Analyzer or other tool to improve the sound quality of 1930s material - and if so, how?

Locked