Clipping or Distortion on Vocal High Notes of Ripped Audio

Try this. Amplify (not Normalize) to -7dB instead of the default value of zero. That will put you much closer to the old sound standard for Music CDs and it might be a much better match to your other music.


Hi Koz,

Thanks for both of your suggestions. I will burn Kathleen Battle and also try your other “Amplify” suggestion. If Battle sounds bad on that stereo, then I’ll know that it’s the culprit. I used “0” normalization, initially, due to an Audacity You Tube video that I had seen. I thought that it was the best setting for my audio…but…I’m a novice at this so I’ll take your advice and try a new approach.

Thanks for getting back to me so quickly.

You may just have encountered a situation where the default values in the tools aren’t appropriate. Or your car is broken.

At the dawn of recorded history, the volume levels of the digital services were chosen to avoid damage at all costs. Everybody had scars from the very narrow range of loudnesses that were possible in analog. For example, standard digital broadcast audio can double in volume three times before overload damage occurs – both US and Europe. Lots of room for explosions and gunshots. Audio on a Music CD was chosen around those values and I still have Music CDs recorded very early that conformed to those standards.

Then the inevitable loudness wars kicked in and we played “my show is louder than yours.” I have one commercially recorded Electronica Dance CD that lives just every so slightly below overload. The whole disk. In Audacity, the blue waves look like a solid block.

You can only get away with that in post production. There’s no way to record a live performance and automatically have it punched and loud like a commercially produced disk. We get regular posters who want to do that. “How come my performance isn’t as loud as the one from Warner Digital Audio.” Worse, “How can I get there.”

Hire Warner Digital Audio.

So if you’re doing classic musical performances and you know your car has troubles with energetic notes, don’t put any music up there. 3dB change is just barely perceptible so you can do that instead of Amplifying to 0. Nowhere is it written that you have to punch the music so hard your ears bleed.

Unless you want them to. We have tools that can do that, too. The Battle piece has been changed slightly to purposely put multiple peak notes near but not quite hitting maximum.


Hi Koz,

I played your Battle last evening and found that it sounded wonderful on my car stereo. It sounded just as lovely as both of the CD’s that I have of hers, her Carnegie Hall debut and a recording she made with Wynton Marsalis of Baroque music. I turned up the volume as loud as it would go and there was no distortion whatsoever…so…on to plan B. I’ll redo the 16 songs using your suggestions. I’ll initially do 2 or 3 and then pop them into the ol car stereo to see if there is a difference.

I have a Christmas performance next week and have been listening to a lot of my Christmas CD’s including Christmas at Carnegie Hall with Battle and my favorite mezzo, Frederica von Stade along with Andre Previn. Believe me, there has been no distortion at all with any of my CD’s on that CD player, as I mentioned before. I know a good recording when I hear one…at least …ones that have been professionally done.

I’ll start on the new project tomorrow using your suggestions and let you know what the outcome is. I’ll leave the Input Volume at midpoint and use the Amplify option at -7db, just as you suggested…no normalizing.

Thanks for all of your help.

I played your Battle last evening and found that it sounded wonderful on my car stereo.

That’s not good news. That means your performances may not sound any better just by lowering the volume slightly. I was so hoping Kathleen was going to sound terrible and that would give us a very clear pathway for your success.

Now, it’s back in foggy territory again. Why would some performances sound awful and only in your car?

This is where I/We need to stop thinking analytically (which hasn’t worked out so well) and go to right-brain association.

I need stronger coffee for that…


Free Association…

Sometimes we can solve these things by making them worse. Either direction, better or worse works. I would die to get my hands on one of those DVDs… I wonder if there is something other than volume messing up…

Do a test. Find one of your sample performances with significant distortion and do an Amplify to about -3dB or so (no chance of overload), and then apply Effect > Low Pass Filter, Cutoff Frequency 15000Hz. Fifteen thousand. Leave the other two values alone. Does that one sound any better in the car? Did the problem go away?


When you take a breath, are the blue waves centered around the 0.0 point (measured on the left)?


Hi Koz,

I will do what you suggest and let you know. Now, I’m tired so forgive me if I sound brain dead, and I don’t drink coffee unfortunately, but I rip a song from the DVD audio again and then follow your most recent instructions. Correct?

You should know, that not all the songs have evident distortion when played on the car stereo. Some, more complex arias might be too “busy” between what the pianist is doing and what I’m singing, and I may not be able to hear any problem.

One song, was edited in Audacity directly from one of the audio cassette tapes that I had made from the VCR tapes that were created in the marketing department I mentioned in a previous email. It sounds okay, but it’s not as “clean” as the other songs taken from the DVD audio. I had to do this because when I saw the DVD ripped waveform in Audacity for that particular song, for some bizarre reason, there was a large gap at the end of the song right in the middle of my last long and beautiful note…the money note. I was aghast! Being a novice, I tried to “fix” it but could never make it sound just right so, I was forced to use the audio taped version…which I was very grateful to have. What I’m getting at, too, is that there is no distortion at all in that particular song even though it’s “murkier” sounding than the other songs because it’s taken from a tape. It’s still good enough to use and happens to be the most beautiful song on the CD. My favorite, anyway.

So, my question to you is, when people have video taped versions of movies converted into DVD, is there something that the lab might do that is different to unwittingly cause this “high strangeness” to happen…that you are aware of?

I will commence with this latest project tomorrow morning! Grateful for the help, but much too tired to start tonight.


Have a good evening.

I’m suggesting that 15000 filter as a guess that there may be some sound so high pitched that humans can’t hear it, but that drive your car crazy. It’s remotely possible to get damage like that when transfering between video and audio standards which are slightly different.

This filter will remove that possibility. Also the slight reduction in volume (-3dB) will take care of any possibility of overload distortion. If neither of those works, I have a more radical possibility, but we’re not going to make Christmas.

A friend of mine on the other side of the city was telling me about her adventures as a Dramatic Soprano. Her husband peaked over his newspaper and said, “That means loud.”


In the posted sample there is a very high frequency whistle at about 15800 Hz
This will not be removed by a low pass filter at 15000 but can be very effectively removed with a notch filter.
Here is a suitable notch filter - I’ve customised it so that the default settings are pre-set to 15800 Hz
modified-notch.ny (821 Bytes)
Instructions for installing plug-ins are here:

I’m not sure that I agree. There definitely seems to be some distortion on that top E, though it is not particularly severe.
My suspicion is that some part of the original recording equipment was running close to overload and has introduced a little distortion, but the major problem is that the car CD player is resonating at just the wrong frequencies and making it sound much worse than it really is. If I’m correct then there’s probably not much you can do about it, but the problem will probably not be evident to anyone else unless they sit in your car to listen to it.

I’d suggest that with the tests that Koz has described, you also run the notch filter on the audio, or do two versions, with and without the notch filter.
I’d also like to suggest another test - With the track that you posted, apply the “Change Speed” effect with a value of -19 (minus 19)
This will sound “wrong” as it will make the music slower and lower pitched, but it will tell us a lot if that “cures” the distortion problem.

The distortion in the singer’s voice seems to be modulated by the piano …

In the posted sample there is a very high frequency whistle at about 15800 Hz

I was hoping for something further up. Like beyond 17.8KHz. Beyond that, the player has to guess at the data since 44100 is indeterminate up there, but video services is not.

There definitely seems to be some distortion on that top E…

That last note? I’ll take your word for the pitch. Yes, I sorta heard it, too, but I wasn’t looking for something barely perceptible in a quiet room. I was looking for something that would scare people in a moving car which is the complaint.


The distortion in the singer’s voice seems to be modulated by the piano: when the piano note fades out the singers voice becomes distorted.

Perfectly correct. But it’s only distorted on the Music CD and only in the car. Play the CD anywhere else and it’s fine – or at least acceptable.

We need to remember that this isn’t a top studio recording; this is as good as you can do by going to a digital video tape and through multiple bounces to a Music CD, which, for some reason, is magic. I would have expected that to work correctly. We do similar tricks at work.

I wouldn’t be surprised if the digital camcorder had some AGC on it, but the result appears to be acceptable until the final step.

There is a cartoon (has to be New Yorker) where two scientists cover a blackboard with figures and calculations. In the middle of the board is printed: [Magic Happens].

That’s what we have.


I still think the piano is causing the distortion but I’ve changed my mind about AGC being the cause.

The distortion in the singing corresponds with bass piano notes (and the changes are too snappy for AGC).
If you compare these waveforms … [watch and listen to playback in Audacity, pan 100% left]

One is rumble enhanced: a low pass with 100Hz cut off. When the enhanced rumble track peaks so does the distortion in the singers voice.

If the sound system in the car is bassy that would make the distortion sound worse, as it’s the bass notes which are causing it.

Filtering out a lot of the bass at this stage only helps a little: it doesn’t remove the distortion the bass notes have imparted to the singer …

Hi Guys!

I’ve just read all of your comments and I appreciate the input!

I’ve just listened to Trebor’s rumble enhanced clip and the sound I hear on my car CD player is just like that only not so pronounced, of course. I would describe it as a kind of “fluttering” sound during certain passages, and by the way, they don’t have to be loud or forceful passages either. I really don’t notice this on my Mac with headphones as I’ve mentioned before.

Let me describe the room I was singing in and this might help you too:

It was an oval rotunda with a high domed ceiling and marble floors and walls. There was no mic, of course. I had hired a professional sound guy from a performing arts center to video the performance but at the last minute, he had to pull out. A friend of mine just happened to bring his camcorder with tripod or I wouldn’t have had any record of the performance. I was so happy about that!

The camera was on the tripod off to my left and about 6 feet back, so what you are hearing is what is coming out of the left side of my mouth! LOL I had hoped that he would have set it up right down the middle of the room, but he didn’t. It was difficult, since the audience was gathered in a semi circle around me and the piano so I’m sure that was the reason why he didn’t set it up that way. As I said, I was just thrilled to have a record of the performance. I had to concentrate on my concert and not worry about the camera, etc.

Getting a late start today but will begin to redo a couple of the songs using your suggestions.

Thank you so much!

The rumble enhanced version was a visual demonstration of the association of singing distortion with bass piano notes: meant to be seen as it played on Audacity.

However this bass-cut remix is meant to be heard … [still distorted though]

Hi Trebor,

I just listened to the bass cut remix four times. I get your point and do see a difference. I’m beginning now to try to implement the suggestions that you’ve all been kind enough to make. Thanks!


Hi Guys,

Been trying to post the “Change Speed” to -19 clip for a bit now. I’ve cut it to 5 1/2 seconds cause each time I try to send it, I get an error message telling me that it’s too large. Hope it works this time. Haven’t tried it in my car yet but will do so shortly. Thought you might like to have it.

Thank you.

Drastically lowering the speed has cut the high frequencies and with them high frequency distortion.

But you’d be better off cutting back the high frequencies using the equaliser rather than slowing speed or pitch shifting …

The ‘after’ result of cutting the high frequencies (treble) is dulled,
but because the distortion was in the high frequencies, which have been cut, it becomes less obvious, (but it is still clearly there).
Audacity equalisation curve (Porgi Amor) #.png

The reason for the “change speed” suggestion is not as a “solution” to the problem. A 19% change in the speed makes little difference to the frequency range, but what it does do is to make a significant shift to where the peaks occur. My suspicion is that that the distortion is sounding a lot worse when played in the car than elsewhere due to resonant frequencies in the car sound system. By changing the speed we are in effect changing the musical key, so resonance should not occur on the same notes as before.

You shouldn’t :stuck_out_tongue: it’s roughly an E after the speed change. In the original clip it’s between a G# and an A (about 850 Hz).
I’m guessing that the reason that this note is between G# and A is due to a sightly inaccurate clock speed in the original recording equipment (or elsewhere in the recording chain) and as this is classical music, it may be worth correcting to speed to bring it into concert pitch. It could equally have been due to the piano being tuned slightly off concert, but even then I think it would do no harm to shift it to concert pitch as the amount of speed change required is only about -1.1%, (but that is still enough to be noticeable to many classically trained musicians). Hopefully the pitch/speed is out by the same amount throughout the recording and does not drift.