Axiom Home Page Forums General Discussion The Water Cooler Fun with Audacity - Dynamic Compression

So, having been pointed towards the "Audacity" audio editing software on Audioholics' AVRant Podcast, I've been having some fun.

Here are the waveforms for:

1. Leonard Bernstein conducting Gershwin's Rhapsody in Blue - nice wide dynamic range - it's an orchestral arrangement, after all.


2. Miles Davie, "So What" from Kind of Blue. Small group jazz, so much less natural dynamic range in the source. Lots of breathing room in this mastering of the recording (well, Coltrane's solo in the bottom channel is pushing it a little, perhaps, although there's no clipping visible when you zoom in on the wave peaks).


3. Fleetwood Mac, "Go Your Own Way".
The highest peaks correspond to the drum beats, and the song is mastered to just allow for the loudest peaks at the climax of the song.


4. Dixie Chicks, "Lubock or Leave It".
Dynamic compression Hell. When you zoom in on the peaks there is clipping out the wazoo, right there in the source material where there's nothing you can do about it.

Drown in the blue ocean.

Yuck.

I've been paying more and more attention to dynamic compression lately. I wish I didn't know about it.

Nice graphs.
Good demonstration.

Yeah, there should be a music review site with these graphs as standard.

Very good, Jim; a graphic illustration(literally)of some of the garbage being turned out in the pop area.

Nice demo, Thanks.

Is this easy to run with my HTPC, how do I hook it all up? I know Audacity does a whole lot more, but I would just be interested in seeing the compression of my CD collection.

Audacity is pretty darn easy to use (and also free) - just rip a file from CD, open in Audacity, and there you go - you'll get a graph like the ones above. You can zoom in, etc, for different views (to actually see the clipped waveforms on the Dixie Chicks, for example).

(It's also easy to make a file of test tones using Audacity, your choice of frequency and duration.)

So, you can't just play a track of a existing CD while the program is running? You have to rip the track first off the cd, and then open within Audacity?

What does the clipped waveform look like? Is it in the above pitures.?

Ok, I think I've figured it out. I'm ripping as lossless WAV. Ok, how are you capturing the jpg images above, I don't see where the program will allow for saving the graph?

I do have a nice screen capture tool called printscreen2000 I could use...

sounds forge is another cool piece of software. I played around with it on my step dad's computer... I think its like 300 bucks though

http://www.sonycreativesoftware.com/products/product.asp?pid=431

Printscreen will do the trick. I use Gadwin Printscreen, myself.

Ok, I just grabbed an old CD from Great White I had laying around. Here is the waveform and spectrum (energy) graphs for that song.

Based on the other graphs, that looks pretty clean.

And here is an example off of my kids Baby needs Mozart CD. Mozart's Divertimento String Quartet D Major by Salzburg Symphony No1.

Please believe me that I don't mean to sound cocky in saying this, but I've been working with audio editing software for years and have seen waveforms like this for just as long with recent recordings.

It pi**es me off, actually, that it's become so commonplace with recent mastering. It's gotten to the point where I'm surprised when a recent (popular) recording DOESN'T look like that. I still DO get surprised, however, to see how many discs go above "0" dB.

It's one of the biggest reasons why I'm re-exploring vinyl. Since those graphs are results of compression and normalizing in the final mastering stages, I'm hoping that some LPs are NOT mastered the same way....i.e., despite surface noise, etc...some LPs might actually offer superior playback because of the final steps of mastering for CD not having been done the same heavy-handed way.

Nice. What does that second set of graphs mean? Red = your ears are gonna blead at 106db on M80s??

I used audiacty to cut a couple of songs for my daughter. I believe it resamples when I writes a new track so you lose some quality, but thats not an issue for what I did.

The tracks? Floyd Pink Floyd - Ummagumma , Several Species of Small Furry Animals Gathered Together in a Cave and Grooving with a Pikt and Bike. She seems to have my taste for the eclectic.

Those tracks looked pretty good on the graphs, no compression, but I am almost afraid to look at some of my other stuff.

Fred

Yeah, looking at various tracks in my own collection confirmed just how prevalent the compression is.

I think the red graph shows the "energy" of the song. I really have no idea what I'm doing or what these graphs show me.

How can I tell from looking at these graphs if there is compression or clipping?

Mark. I didn't find out about dyanamic compression until after my brother gave all our vinyl. I suspect that with well mastered vinyl you could transcode to digital filtering out the snap crackle pop and still come out with a better sound than more recently mastered material.

Fred

Pity you wouldn't have the newer music...

Well, right now I am concentrating on rebuilding my collection of older music. Speaking of which, I just picked up Dire Straits Live at the BBC.

I'm gonna copy a couple of my CDs to disk and take a look at them using audacity.

Fred

Here's a "zoomed in" look at a small section of Lubbock or Leave it, amounting to a few hundreths of a second. Apparent clipping is marked with arrows.

I did find an article when google searching that talked about anthing that gets chopped off flat above 1.0 or below -1.0 is an example of clipping. The dB waveform from the dropdown is pretty cool as well.

I'll have to check some of my old metal cd's from the 80's.

You guys beat me to it, I have been working on looking at all my cds to see which ones are clipped or just plain too loud and then I was going to post my results. I am also going to compare any remastered cds against the originals I happen to have.

That'll be a good comparison, Jason. I was wondering about that this morning--the remasters vs. the originals.

OK folks, let the fun begin. I started with Dire Straits Live at the BBC (havn't even listened to it yet)track 1, Down to the Waterline.

When I opened it in Audacity it looked pretty crunched towards the end of the song.


Once you zoom in though it looks pretty good. Looks like a tiny bit of clipping but nothing really.



From what I have read here, anything done by Knopfler is damn good.

Not bad

Next up, a classic. Led Zeppilin 1 Dazed and Confused. Interestingly, the full view of the song looks less crunched than the Straits song. Once you zoom in though...



OK, not really bat at all, but you can see there is more stuff right at the edge and some clipping.

So far so good. Theres something to be said for liking old stuff that was burned in the 80s and 90s.

Next up...

... Santana, Supernatural. Its really my only recent CD.

And looking at the full song...



We have a winner folks. Dosn't look any better zoomed in.



I must say, I am kind of dissapointed with Santana. I always though of him as a musicians musician.

I was impressed with the Dire Straits track. Once you zoom in, you can see a fair amount of dynamic range in the recording, even though it is fairly loud.

As an interesting side note, in the late 70s and early 80s when I recorded a lot of stuff tape to tape, the rule of thumb was to set the volume so that those nifty digital meters were just below red (no clipping) even though none of us knew what clipping was.

By the way Jason, I remember reading an article where the author gave an example of a Zeppelin song with the original, remastered in the 90s and remastered post 2000. You could see the progressive increase in volume and dynamic compression. The last one didn't look quite as bad as the Dixie Chicks or Santana, but it was close.

I tried some Ozzy stuff and reversed the track to listen backwards, didn't hear any satanic messages.

This is fun stuff to see. I would bet that the albums that are bad are the ones that I like the music but just never want to listen to!

Jeb

As far as Santana, that really isn't surprising. Even before mastering, or recording, Santana is known for having a VERY compressed guitar sound. If you get a chance to watch him live, even if you are not a guitar player, you can notice that regardless if he barely touches the strings or really wacks them, the volume is the same (very little dynamics).

I think I will need to put a Jeff Beck re-master through this as his playing is on the other end as being very dynamic. Not sure how much it makes it through mastering though.

Dare someone to try St. Anger.

Seeing as you're the only one who can actually stand to listen to it... go for it!

I installed this software and looked at a couple of flac files during my lunch break. Pretty neat stuff.

I looked at some Bock and Chopin (classical and piano) and it was very clean.

I looked at dave matthews band "stand up" album. It looked pretty good. Not quite as good as the classical songs but still very good.

I looked at an mp3 I donwloaded from a local club DJ and WOW. There was no background color at all.. just a huge blob of blue allthe way across. (It made santana look great!)

Bock?

Did anyone else just hear a chicken?

That was good!

Errr... Bach

Glad I could provide some amusement.





Don't you see the resemblance?

So you're talking about Yo-han Sabazchen Bock?

exactly

Welcome to the last phase of the loudness wars. I've posted a lot of great articles on the subject. This has been happening for decades in some form or another. But the recent trend is blaming MP3s for the noisy compression. Not true!

Anyway, here's some reading for you all:

http://www.turnmeup.org/
http://en.wikipedia.org/wiki/Loudness_war
http://www.rollingstone.com/news/story/17777619/the_death_of_high_fidelity
http://www.guardian.co.uk/technology/2007/jan/18/pop.music

Zoë

I mite add, I was using the foh-net-ik spelling so that
no-bud-ee wood get kuhn-fyoozed

I was still confused.

Doubly so, apparently.

I prefer to be triple confused.

There is obviously a limit on those graphs were things start clipping. What sets that limit? Is it the dynamic range of the CD, or the original recording gear?

Gremlins.

Thanks, JP!

Turns out, those darn gremlins are responsible for a lot of things. . .

Like my unfinished charts and disorganized office.

Couldn't be the time I spend on the boards. Nah. Gotta be gremlins.

is the way they compress and limit the tracks now i guess is that they want everything loader for the radio i recently had some mastering classes and i guess the producer now a days want everything atleast 2 to 3 db higher

Yup, it's all for radio play. You can't be outdone by another cd as if it is louder than yours then it sounds better on the radio, supposedly. I guess people don't like turning up the volume.

I think that's the dynamic range of the CD format, but I'm not certain. As usual.

That's what I'd think, too, Ken. With all the technology available to us, you'd think we could improve on it.

It's actually a pretty big range, it's just that engineers are pushing up against it.

There is obviously a limit on those graphs were things start clipping. What sets that limit? Is it the dynamic range of the CD, or the original recording gear?

The limits on the graph (+1/-1) are the limits of the CD, but it's the fault of the mastering process, not the CD which has plenty of range to spare. The clipping could be in the recording; however, modern professional pop recordings won't have that problem (well, unless they do it on purpose for a "lo-fi" sound).

No, as Jim also pointed out, the dynamic range possible on CDs isn't the problem; the theoretical max range is about 98dB(6.02n + 1.76dB), and no recordings, even the most dynamic classical items, would exceed about 70dB(and even getting close to that leads to some bitter complaints about excessive dynamic range). The maximum level that digital recordings can achieve(termed 0 dBFS, for Full Scale)sets the limits on how high the recording level can be without clipping. Since louder sells in the pop arena, some recordings go up to and even beyond 0dBFS, resulting in the clipping that some of the charts show. But there's a difference between that and dynamic range; for example, it'd be theoretically possible to have a recording with zero dynamic range between the high and low points, but everything is recorded at higher than 0dBFS and is clipped.

A couple random memories which may be interesting in this discussion:

Many Zep fans claim the original cds from the 80s (or early 90s?) sound much better than even the most recent (remastered?) re-releases (2007). They (fans) claim it was because they (engineers) didn't use the original recordings in the most recent releases. But the louder is better for radio argument sounds more likely.

Also, back in the old LP days I read where Bob Ludwig (a famous mastering engineer or whatever they call it) said that groups would often want their sound louder - but he couldn't do that because the needle would jump right off the track of the record. I guess now they can do it.

Here is Boston- More than A Feeling, the remaster is on the bottom. Other than being louder it looks as though Tom prevented any clipping. Try as I might I could not identify any points that went flat.

I need to fire up my Audacity.

On a side note, if you remember about a month or so ago I was referencing listening volumes and my hearing condition. These graphs offer a visual for what I was explaining regarding what happens when mutliple conversations are going on around me. Compression

Well I have one disc with me right now, David Gilmour - On an Island. One of the better tracks is Pocket of Stones. Here is a graph of it, as well as a zoom of the peak we see.

Not to bad for rock music.

Thanks JohnK. And Jim, too!

I find this fascinating how bad recording techniques are the new norm. It's frustrating, also. The double-edged sword is that now I am much more aware of dynamically compressed recordings.

I can see how dynamic compression is misconstrued as a good thing. Living with roommates makes me aware of the volume level. Listening to music with a good dynamic range makes me sit there with the remote handy to turn down the peaks. There is no adjustment needed for dynamically compressed recordings because the music more or less stays at the same level.

I've got to say, the dynamic music has much more impact emotionally. It's like being on a rollercoaster with a killer loop or giant drop--you're going to remember that thrill. Take, on the other hand, a roller coaster that is nothing but loops and drops, sure it may be exciting at first, but it becomes too much after awhile. You need the lulls to enjoy the highs.

Now I know why some of my older CDs sound quiet--it's not that they're bad, just recorded well. A couple of weeks ago I put on Pink Floyd's "The Wall" and felt it was a good recording. I've read other message boards where people have complained about this particular CD, but, to me, it sounded clean and balanced with the loud parts where the loud parts should be.

I find this fascinating and it makes me want to run all my CDs through Audacity. But, then, I know if I do that I won't stop thinking how bad a CD is when I listen to it. Argh.

Sometimes ignorance is bliss.

That Gilmor looks pretty good Mark.

This thread is quite interesting. So, dynamic compression can be a function of the original recording as well as remastering. I wonder how heavy distortion on guitars effects dynamic range?

I finally listened to Dire Straits live after posting the other night and it sounded so clean to me compared to stuff like Led Zeppelin. I wonder how much of this has to do with techiques during the original recording? ... and how much is influenced by the waveforms I just looked at.

On a related note, is there any way of telling when a CD was produced by looking at the info on the slieve? I have been looking at copyright dates and noticing that it can vary a lot. For instance, on my last purchase I had a choice between two used copies of Wish You Were Here. One had a copyright date of 1971, the other of 1992.

If possible, I want to be able to pick up earlier productions of a CD because it is more likely to be less compressed.

Fred

Of course, there are also older recordings (see Joshua Tree) that are simply terrible, and are much improved on re-release.

I'm not talking about any of that new stuff.

Given that they weren't making CDs in 1971, it might be worth finding out what the real date of the pressing/mastering for CD--it may well be that these are the same CDs.

Oops, typo... meant 1975, though they weren't making cds then either.

They are definately not the same disks as the publishers were completely different. The one I picked up was produced in Japan.

I wonder about disks with copyright dates in the 80s and 90s from materials originally produced in the 70s.

Too much thinking...

How would guitar distortion affect this. Probably not intuitively

Distorted guitar music can be dynamic or compressed. A lot of that is a function of what is generating the distortion.

Over-riding the input of the amp (either "hot" pickups or a booster or compression pedal)
Pre amp distortion
Power amp distortion (even the rectifier type comes into play, so we could say portions of the power amp)
Speaker distortion

Can you tell I play guitar

As a for instance, you could play the same guitar into different amps of the same apparent distortion into the same speaker cabinets have have a pretty good difference in compression.

Say a Marshall JTM 45 versus a Soladano SLO

I will see about graphing the Jeff Beck tune I was thinking of (CWEAL)

Well, definitely some clipping in the remaster of Blow by Blow. Here is Jeff Beck Cause We Ended as Lovers

Mark, that's cool. I really appreciate your perspective. I was thinking about the same thing.

I think the difference is that the clipping we're seeing in the Audcacity analysis is caused by over-saturating the recording medium. I should think it would still be possible to record a "distorted" guitar without the recording causing additional distortion.

Very true Tom,
If you had a SA in a studio, you could see this. I have a couple friends who do. Trust me, its both science and art.

I mean, just changing the speaker, though same specs, can cause a change in compression in live amplified music. I tired hunting it up but couldn;t find it; but a studio friend of mine had a SA chart (what we are working with here) of the same guitar, amp but a different speaker, though it was same size, same RMS, same ohm, etc. It was pretty obvious the compression difference. (FYI: An Alnico Blue against a Jensen.)

Going to see how a modern day "Grammy" winning CD stacks up. Subject Jeff Beck Guitar Shop.

From Jeff Beck's Grammy Award winning Guitar Shop we bring you Where Were You. This is one of my fav tracks.
It provides I think for an interesting observation on what was voted best instrumental album of the year.

Jeff is known for dynamics (trust me, I have see him live multiple times and it shocks you that a musician other than one in the classical genre understands this and uses it so much. The last time I saw him, I walked out and thought the guitar lesson to me was - my god man, use DYNAMICS when you play.)

This track is odd though in the fact it is well known that the guitar is HIGHLY compressed before being amplified and recorded. (FYI: So much so that it can not be performed live in the same rendition).

Hmm:
Clipping is not there - so good mastering. We do see less definition between peaks/valleys which illustrates the compressed nature of the music.

Oh, and I guess you can tell I am a JB fan.

I finally took some time before heading off to bed to scan the 2 copies of Holst - The Planets, well only Jupiter, the Bringer of Jollity. It does show the difference in loudness very well. The top is the Andre Previn version apparently done as all digital(DDD) according to the jacket. The bottom is from Mehta's version which was Analog to Digital(ADD). Other than needing to turn up the volume on the Previn version I can't tell any difference.



I had my SPL meter out to check for any difference in Dynamic range while listening on the M80s and they both run in the mid 50 db range for the most part.

Thanks for the explanation Mark. I got to wondering after noticing that the Zeppelin song seemed more compressed then the Knopfler song I was looking at.

The upshot of all this for me is that most of the stuff I plan to re-acquire is probably going to be ok because it will be earlier productions of late 60s to early 80s music.

I think I will stay away from re-mastered productions if I can.

Fred

Very interesting, Jason; in addition to the overall higher level at which the Mehta disc was mastered, it can be noted how the two conductors, although working with the same score, occasionally applied different dynamic gradations. For example, about 4-5% in, Mehta applies a crescendo earlier than Previn does at the same point.

I had noticed the timings were a little off from each other and I guessed it was due to the different conductors. I really can't say I can hear much difference between the 2 other than the loudness factor, then again I have only listened to each one about 3 times.

Here is the 1994 remaster of Pink Floyd - Wish You Were Here from The Wall.



They stretched it to the limits, but I couldn't find anything going beyond the borders

I agree jakewash, that looks pretty good- but I was more surprised to hear this track was on The Wall album

Sorry, I had messed up my graphs, that's Hey You. I was queing up Wish you were Here at the time.

For a different take on looking at these waveforms, check out this, which I found after reading someone's (sorry, forget who) post on how Linkin Park songs all sound the same...

turns out they all LOOK the same, too

Linkin Parks songs all look the same

yes they do. While they are all over compressed, the 3rd one, 'Somewhere I Belong' looks just horrid!

Jason, I am just embarrassed that I couldn't tell it wasn't WYWH by the waveform