Removing distortion

One of the contributors to my podcast sent me an interview he recorded in Skype. I’m surprised he didn’t notice it, but every time his interviewee speaks, there’s distortion. It sounded to me as if he had placed a mic in front of a rattly speaker, but he says he recorded with the Skype recoder plugin.

I found a video online that recommended using Clip Fix (under Effects) to fix distortion, but it didn’t make much difference. Is there something else I can do in Audacity to smooth out the sound? It’s hurting my ears to listen to this interview.

It’s hurting my ears to listen to this interview.

And would probably hurt our ears if we could hear it. We need to experience it to guess at a solution. Post a WAV clip of the work on the forum. If it’s stereo (two blue waves) post 10 seconds. Mono, 20 seconds. We can’t do anything with a two second blast of trash.

Scroll down from a forum text panel > Upload attachment > Browse.

https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1

Koz

Exactly. I’m afraid nobody’s going to be able to get through it.

Here’s a clip.

This is after extensive surgery.

This is odd. Analysis of the original proves there are no tones higher than 7500 Hz. None. The musical tonal population of the clip goes out to about 7500 and drops as a rock. That’s significant because AM radio goes away at 5000. So the best this clip will ever do is slightly better than your local AM radio station. And it’s distorted, I’m guessing from simple overload.

So that’s all I can do. I eliminated all the tones higher than about 2500 (worse than a telephone) and that got rid of most of the harsh crispness. It is actually better because I can understand what’s being said.

Whoever shot this should find a different recording technique and pay attention to the volume.

Koz

Skype recoder plugin.

Can you blow that out a little? Which software exactly on what kind of machine? Links? Has it ever worked or is this the first pass?

That didn’t come from Skype, did it? Recording Skype is a major battle. They change their technology on occasion and even pay-to-play Windows software like Pamela occasionally loses it.

You can record Skype with perfect reliability. All it takes is two computers, or one Skype computer and a sound recorder. Basically, that’s how all the grownups do it. That’s how the networks manage to get good Skype interviews day after day after day.

I did it with two older Macs.

Pando does it with a much larger setup, but if you strip off all the stuff you don’t really need, like their scheduling and research computers, it’s down to two computers and a sound mixer, and most of the mixer isn’t being used.

What’s key is recording Skype on a different machine than the one running Skype. However you get there, that’s critical.

I know people are going to say, “My cousin in Tallahassee gets this to work reliably every time on one machine.”

Or they do until Skype changes their algorithm, or worse, their setup is a celebrity unicorn. There is only one of them in the country. We had a unicorn post on the forum once. Nobody else could get it to work the way he had it. It doesn’t count unless everybody can do it.

Koz

Koz, that does sound better. What did you use? The Low Pass Filter? I tried that just now but it didn’t seem to have much effect.

As I’ve already explained, it did come from Skype, but not recorded by me, and actually recorded on a different continent from where I am. I’m told it was recorded from Skype using the recorder plugin, but I’ve never had such bad results using that, and I’ve recorded hundreds of podcast episodes from Skype.

Multiple machines, not doable on my budget, but as I say, I’ve never gotten this kind of result on ones that I’ve recorded.

I need to go back over the tools to remember what I did (different machine).

I’ve never gotten this kind of result on ones that I’ve recorded.

How? I’ve wondered why Skype doesn’t have a button you push to record the conversation. Everybody on earth wants that button.

We’re assuming the performer’s microphone was working OK? I bet it wasn’t. It sounds like what happens when somebody plugs a sound mixer, not a microphone, into the MIC-IN of a laptop. Very great mismatch.

Because of digital and encoding delays, nobody has sidetone any more like older telephones did. You never get to hear what you sound like. Your first clue that your own voice is a distorted mess is someone will comment they can’t understand you and you need to call back.

I’m surprised he didn’t notice it,

Are you the Producer? Do you decide what goes into the show? You need to have a nice chat with the recordist. They can’t submit voices that sound like five miles of bad road.

I’m looking for the settings.

Koz

Yes, I’m the producer, although often I’m a participant as well. This same contributor has done other interviews that didn’t sound bad.

He says that the interviewee was talking on Skype via his phone, and that that was probably the source of the problem. Probably not a good mic on his phone.

First I removed low pitched rumble with Effect > Equalization: Low rolloff for speech, Length about 5000.

Then Effect > Low Pass Filter: 2500, 24dB. That’s a pretty stiff filter.

Koz

The clip has significant trash below 100Hz. Tones that low are not needed for speech and it’s not unusual for people to use the 100Hz filter on sound mixers. I can still hear wind noises in the background even with that.

Koz

OK, that works! This at least won’t split people’s eardrums. Will have to set some rules with the team about aborting interviews for bad sound quality. He said he noticed the bad sound quality after he started recording… It can be frustrating to have to try to reschedule, but pointless to record the interview if it’s going to be too tough to listen to.

Thanks!

He said he noticed the bad sound quality after he started recording…

So depending on what broke, you would be setting yourself up for a duplicate failure.

I still don’t think I have all the pieces of this. If the performer is on Skype, why was it being recorded multiple continents away? Time Zone issues?

Koz

I’m not sure who you’re referring to as the “performer”…

For this podcast, there are several people who record interviews or book discussions, and they happen to live in different countries. I’m an American in Japan, but there are a couple people in Australia who are involved, and a couple in North America. In this case, it was a guy in Australia interviewing someone in the US.

The performer is the person whose voice is on the clip. To resolve sound issues it’s best to know where it’s been.

So the performer is simply talking into his cellphone? Someone has a license (required last I checked) to cross-connect his cellphone call to Skype. Skype delivers the connection to the recordist who connects to Skype and uses special purpose software to record it.

Did I get it?

It is a little rough to believe the cellphone was at fault because they have mature voice processing, echo handling and environment processing. The performance appears to be a normal person having a normal conversation in a normal voice—in a windy location.

Skype likewise has been doing this for a while, but they got their reputation for reliability by brutally taking over the computer running it. We know that it’s not difficult to record the local microphone; that’s a service of the computer, but it can be very tricky to record the far side. That’s a service of Skype and they don’t always like to share.

It’s the far side that failed.

I think I mentioned even the highly regarded Pamela paid software got snookered a while back when Skype made an unannounced change and put several podcasts into the dirt. So this may seem to be a unique distortion one-off, but it’s perfectly compatible with past practices.

You should check your systems to make sure they’re still working like you think they’re supposed to. Nothing like connecting to a desirable interview and have it fail. I would call the Skype connection and make sure the recordist can capture it.

Koz

Not sure if you noticed, but there are two people in the recording – the interviewee, whose voice was messed up by whatever the problem was, and the interviewer, who sounds fine.

I know another podcaster who requires his team to get the interviewees to make a recording of themselves on their end and send the interviewer the MP3 of themselves. I’ve done this sometimes with other members of my team, but it seems a little, I don’t know, presumptuous, to ask that of interviewees. Many of whom may not understand how to do it.

Would be interested in knowing what others do.

interviewee, whose voice was messed up by whatever the problem was, and the interviewer, who sounds fine.

Perfectly correct. The local microphone is recorded directly. The far side comes from Skype.

I know another podcaster who requires his team to get the interviewees to make a recording of themselves on their end and send the interviewer the MP3 of themselves.

Also correct, although as you noted it’s uncomfortable to ask an Important Person to do that, beyond they’re not knowing how. I think in your case with third party interviews, you’re stuck with capturing it.

One recent forum poster with a top quality, separately recorded, multi-participant podcast had a rescue job when one of his regular performers messed up their local recording. It is like marching cats.


The two computer technique doesn’t have to be two computers. It can be a Skype computer and a separate sound recorder. You listen to the transmission on headphones and split it off to a recorder. No special software or drivers and Skype thinks it’s running the show. We know it’s not and we’re not telling.

I happened to have a second, much older laptop available to record a test. Denise and I are on opposite US coasts. We’re both on headphones. She’s talking into her laptop—full stop. I’m the one with the extra mixer, etc. It should be remembered the goal was to produce/test a finished podcast in real time including music. Your task would be much simpler.

I took out most of the “Can you hear me OK” and housekeeping. But it is a test. It has fluffs and errors.

She sounds like she’s sitting on the sofa behind me, doesn’t she? She’s not. She’s 3000 miles that way [points east]

It could be said we produced the perfect podcast. We spent the whole thing discussing when to have the next podcast.

Koz

I don’t want to talk you into anything you don’t want to do, but the goal is to remove the sound from the Skype computer as fast and cleanly as you can.

Make sure you check your recording software as close to the performance time that you can.

The desperation method is pick up the phone and record that. I did it with a special microphone. It goes in my ear and the phone goes on top. Record it in Audacity. It uses bone conduction tricks etc.

http://kozco.com/tech/audacity/clips/TP7SoundTest1.mp3

http://kozco.com/tech/audacity/clips/TP7SoundTest2.mp3

My laptop doesn’t have a microphone connection, so that’s what that white thing is.

Now that you’re all excited, I didn’t try it yet with my cellphone. This is all happening on a landline that nobody has any more.

The cellphone test is on my to-do list.

Koz

Here’s a more modern one from when a water main exploded in front of my house.

These are completely unedited except for cutting down to size. There is a natural 6dB change in volume between the two sides that’s easy to fix in Audacity. You should not be talking over the guest anyway.

And we are talking about an emergency recording when your better quality transmission fails.

Koz

I didn’t try it yet with my cellphone.

It works with the cellphone, too. I was afraid of the cell radio energy getting into the show, but no. I tried it on a little stand-alone recorder and plugged into the laptop. Both work fine.

Koz