Recording noise getting worse... not better

Hi forum.

Thank you in advance for maybe (possibly) helping me. I’m in a bit of a panic.

I’m an audiobook narrator, on my 14th book. I have used the same microphone (Blue Snowball), the same computer (MacBook) and same DAW (Audacity, obviously) since I began recording. I have improved my recording spaces from an open bedroom with virtually no safeguards in place, to an empty office in a commercial building, to an insulated closet alcove, to what I’m trying to work in now: a 4"x3" makeshift home booth insulated with blankets, foam with a homemade mudguard.

My sound quality and noise floor is getting WORSE and I can’t understand why. After moving into a space that is smaller and more insulated, I am devastated that I’m putting out poorer sounding audio. I don’t know how to fix it.

The steps I currently take are these:

  • Record the audio in my booth space.
  • Edit.
  • Master the audio, using a combination of the limiter, compressor, and amplification/normalization to bring up the volume to required specs.

This leaves me with an excessive amount of hiss which I cannot remove without severely damaging the rest of the audio. After I sufficiently get the hiss out via Noise Reduction, my voice ends up sounding like I’m inside a tunnel. I’m absolutely astonished that this is happening, and am in a panic because I just began a project, and have at least 2 more lined up. I don’t feel like I can continue putting out audio with these poor results.

I completely get that I’m using a lower quality microphone, and should upgrade. But I don’t understand how the quality is getting worse if everything I’m using has stayed the same, and my recording space should (in theory) be better than before. All my research suggests that hiss is not an indicator that a microphone is failing. Am I wrong? Is my mic the problem?

I have attached some audio. When I normalize to -3.2 in order to bring the levels between -18 and -23, I get the intense hiss in the second sample.

Thanks in advance for any info you can provide.
mastered audacity sample.aup (1.23 KB)
audacity sample.aup (1.38 KB)

I’m surprised that you got this far without knowing that AUP isn’t a sound file. It’s a project manager text file. You can export a ten second clip as WAV (Microsoft) and post that. Try to get some performance and a second or two of silence in the same clip.

Anyway. One thing you can do is make a recording back at the office exactly the way you used to. Is it much noisier than it used to be? There aren’t many variations with a Snowball. Do you have the same switch settings? It has a -10 switch for loud noises and that should not be used for voice. You should be using position 1.

Koz

My fault. In my haste, I forgot to export. WAV files are attached here.

Unfortunately, I no longer have access to the office space where I was previously recording. So I can’t go back.

The mic switch has been in position 1 since I bought it. I haven’t changed it for any of the projects I’ve recorded.

All we needed is the raw recording. It’s damaged. The voice volume is way too low. The background noise level (hisss) is about -62 which is about right for a snowball, but your voice is -40. It should be -6, many times louder.

If you haven’t changed your announcing style or your physical distance from the microphone, then something broke.

Now it’s very difficult. You have contradictory symptoms and technical diagnosis across multiple time zones.

You can get symptoms like that if you announce into the back of the microphone at the 1 setting. You should be speaking into the BLUE nameplate.

Reach around to the rear of the microphone and switch it between the settings. Record and announce as you go. “This is setting one.” “This is setting two.” “This is setting three.” You should not be wearing headphones as you do this.

Switches and connections are the first choices for defects. Are you still using the same USB connection on the computer? Same computer?

Koz

Yes, same computer and same USB cables and connections as always. The only things that have changed are my operating systems and the Audacity version I’m using. My OS X is Yosemite 10.10.5, and Audacity is version 2.1.1, as of very recently.

Your comment made me check some Audacity settings. My recording volume setting was at .50. I have adjusted it to 1.00 and recorded a new sample. The sample is made using mic setting 1. I have samples of the mic settings 2 and 3, but unsure if those are of any interest since I shouldn’t be using them anyway.

Perhaps my volume setting got moved during a software change… I’m unsure. Either way, it seems to intensify the hiss. Does this sound to you like something is wrong from a tech standpoint? Or is it more likely that the audio I’m generating is normal, and my previous mastering techniques are no longer appropriate and I’m simply applying the wrong tools?

And yes, I’m definitely speaking into the front of the microphone, with a pop screen. The mic is roughly 8 inches from my face.

Thank you for everything so far.

My recording volume setting was at .50. I have adjusted it to 1.00 and recorded a new sample.
… it seems to intensify the hiss.

Which is actually making this worse. Typically, Audacity and Macs don’t change the volume of USB microphones. You plug a USB device in and the volume controls go all the way up and turn gray [inactive].

Beyond magic computer behavior, your symptoms are typical of the microphone doing something wrong. Your voice and the hiss both got louder, right?

Do that test and do it live. While you’re recording in position 1, reach around and switch to 2 and then to 3 announcing it as you go. Stop Audacity, export the file and post it. It should easily fit in ten or eleven seconds for forum posting. Try to keep your voice constant as you go.

The switch could be affecting the recording volume…and it could be dirty, corroded or broken. Pure age and/or humid environments can do that. You don’t have to have bad housekeeping to get problems like this.

It’s possible the voice volume, instead of just doing what it’s supposed to do, goes nuts as you change it. That will be a broken switch.

Assume my Voice of Doom Voice: What are you going to do if we confirm the microphone is broken? We can’t fix hardware from multiple time zones away. You should be looking ahead to that possibility. The poster who killed her recording system was able to keep her announcing/presenting contracts going with a borrowed computer and her old microphone and environment.

Koz

I’m tearing your test posting apart. That’s not all rain-on-a-summer-afternoon hiss back there. You have a lot of room noises. Do you have vent fans or air conditioners? Refrigerators? I applied Steve’s rumble voice filter and I’m still left with traffic noises…something. It’s not loud enough to identify, but it sounds like you’re trying to record next to a freeway. This in addition to the snowball normal hisssss sound.

Both of those sounds together are killing noise reduction. I can’t make the clip pass even with very firm noise corrections.

What kind of Mac do you have? The earlier machines had audible fan noises and had troubles with live sound production.

Suddenly having four walls around you may be a nasty noise surprise compared to the office environment.

Are you using the snowball on its little stand? Can you add the Koz Patented Noise Reducer? Any bath towel. Any heavy book. John Grisham works well, too.

Koz

Did you know you could get the alphabet in fewer characters just by reversing the animals?

Quick brown dogs jump over the lazy fox.

32 characters. People have posted painful, odd, quirky sentences that do better, but this one is pretty easy to remember.

Koz

What are you going to do if we confirm the microphone is broken?

I’ll do some research, and get a new one right away. I’m less panicked with that prospect than just continuing to fumble around with my space. A broken mic has a clear solution: get a new mic. Unidentifiable room noise when I’m away from a road and all computer fans is what made me panicky, because it doesn’t have a clear answer or timeline.

Details that are neither here nor there: I have two authors vying for me to work on a total of 5 books, and I’m in the middle of one of them (the only one with an actual deadline at this time). If I tell that author that I have tech problems, I think she’ll understand. All other books don’t have formal contracts yet, but the both authors are not inclined to dump me since I’m the “voice” of each of their multi-books series’. Changing narrators mid-series gets bad reactions from readers/listeners. So, they’re both willing to work with my schedule, and have said as much. I just didn’t want work to halt, as both authors are paying me directly, so I got scared that my audio was getting worse for a reason I couldn’t pin down. I guess the worst result of a broken mic scenario is that I have 7 chapters of garbage that will need to be re-done.

You have a lot of room noises. Do you have vent fans or air conditioners? Refrigerators? … it sounds like you’re trying to record next to a freeway.

Well, damn.

This is the part that astonishes me most. I’m recording in an insulated pop-up closet (3" x 4", with blankets, foam and a mudguard that I made). It’s currently in my living room, which has no windows that face the road. There’s only two skylights, as it’s an attic space. Residential area, non urban. This room receives very limited street noise, save for the occasional loud muffler or motorcycle, which I hold for if I hear them. I can confidently say that the samples I’ve uploaded were recorded when no such vehicles were in earshot.

No fans or AC, or vents. No central air. This room has a gas heater, which isn’t running during recording. Only the pilot light is lit. All other computers in the apartment were turned off, so they’re not humming.

What kind of Mac do you have? The earlier machines had audible fan noises and had troubles with live sound production.

MacBook circa 2009. For this exact reason, it stays outside of the recording space. But also, my laptop fan hasn’t kicked on during recording anyway (short chapters, cool room temps, etc). There is a refrigerator about 15 feet away from me… but I’ve made a point to record the samples when it doesn’t run. Still, I unplugged it for this latest test file, attached here. If this provides significant improvement, I’ll be shocked, considering how inadequate my previous spaces were which gave me better results.

Yes, I’m using the 3-footed stand. It was sitting on a tabletop with a blanket. For this sample, I’ve added a book for it to sit on top of. So now it’s blanket → book → mic stand → mic. A John Grisham was not available.

The instruments agreed with me. Microphone noise tends to have a signature because it’s caused by random molecular motion, yadda, yadda. The point is, it’s graceful and predictable when measured.

You have some of that, but you also have concentrations of noises typical of something physical such as fans, engines, motors, compressors, etc. We regularly scare people when we analyze their sound and tell them they live in Britain or a European country just by the kind of noise they have.

The John Grisham Filter works by having a heavy, stable platform (the book) isolated by soft, fluffy gush (the bath towel) all sitting on a table or desk which may or may not be making noises (but now isn’t significant). I would use more than one layer of bath towel or blanket. If you push down on the book, it should move.

authors are not inclined to dump me since I’m the “voice” of each of their multi-books series’.

Totally understandable. ACX is very clear that consistency is a Really Big Deal. They were talking about chapter to chapter, but it’s certainly worth considering book to book. I would not know how to behave if my Sarah Vowell novels were read by somebody else. I greatly enjoy her presentation.

As I type this, I’m rifling through my mental file cabinet: What else could cause this effect? Typically, once the microphone turns the voice into digital transmission, that’s the end of the noise problems. OK, granted, you can have “holes” in the presentation where the digital stream just stops through some error, but digital errors are extremely unlikely to produce subtle or gentle noises.

Record some where else, just temporarily. If the noises reduce or even change, we’ll have an answer.

Koz

Being professionally obsessive, I would just plug in my second Snowball and cable. And yes, I do have two computers if needed. My older one has screen problems which do not affect sound.

Koz

You may have hit it. I’m reading back thorough the last couple of postings.

“This is position one” in your last clip has clear white noise (and nothing else) and can be made to pass ACX testing with moderate corrections. So the motor noise appears to be coming up through the furniture from the apartment below.

Throw another couple of layers of towel between the book and desk and record another ten second forum test. Hold your breath one or two seconds, announce normal for about 8 seconds, Export WAV.

Koz

So the motor noise appears to be coming up through the furniture from the apartment below.

Bastards.

It’s surprising, but only just, since I am doing this in an old house. And my newly constructed blanket-closet-room doesn’t currently have any safeguards for the floor. I can also fix that, though, for the future.

Attached is the opening sentence from a chapter I tried to record today. Same space, same conditions, but two layers of towel added underneath the book. I also put the recording volume back to .50, since it seems that doesn’t actively contribute towards the issue. Is that right, or did I read your previous post incorrectly?

I have not yet dragged my computer and mic somewhere else to give a test sample for a completely different space. I’ll think something up and execute as soon as possible.

Thanks again!

I did have high hopes. So the good news is no more Pennsylvania Rail Road and Golden State Freeway in the sound. It’s just pure hiss from microphone electronics.

Noise through direct conduction and touching the floor or desk isn’t hard to fix. Attached 1 is me fixing it with plumbing supplies and United States Postal Service rubber bands. You fixed it through Harcourt-Brace Publishing and Wamsutta bath supplies.

But we’re still in low volume territory. Really low volume. I think what tricked me was you were bending forward when you were producing your 1 - 2 - 3 recording and thus got louder. Now you’re back to normal announcing position and you faded back into the noise. I can’t rescue you without the voice starting to sound like you’re talking into a wine glass. Not recoverable.

Try a different computer. This isn’t a setting. If there’s something wrong with the USB connection, it could well affect the volume. The snowball is not an expensive microphone and it’s something of a slave to what the USB connection and cable are doing. Do you have a different cable? Even a shorter one. Announce closer to the computer just to see if your volume goes up. There is a technical limit to how long you can make a USB cable and it’s not all that long. Like 6 or 9 feet.

If both computers act the same way, then we have no choice but to try the new Snowball you have coming. You don’t need to move the studio. That was just to try and isolate the Golden State Freeway noise. We know what it is now.

You can diagnose this yourself. The Audacity instruments are pretty clear. Attached -2 your blue waves over something I shot in my quiet bedroom. That’s with no processing. That’s exactly what it looked like when I was speaking. Click on the graphics to reveal them.

Your presentation is just too low in volume and it’s very likely happening inside the microphone. Did you install custom Snowball software or drivers? I didn’t think the snowball had any. It’s a very simple microphone.

Koz
Screen Shot 2015-10-26 at 6.19.59 PM.png
Screen Shot 2015-10-26 at 6.01.46 PM.png

I would like to pull this out with a “Click Here and Everything Will Be OK,” but all your symptoms point to a hardware problem of some sort. Microphone, Cable, Computer.

I did think about this later and it could be your voice is the right volume, but the microphone noise has increased. Volume sliders in Audacity and the computer rarely increase volume. They all decrease it, but knowing that inexorably drags you back to asking why the noise is higher than normal. See: Microphone, Cable, Computer. It didn’t make any difference at all to the diagnostic process.

My first guess is the cable. USB cables much over about 3’ (1M) are not natural. USB was designed to connect your mouse, not long distance transmission to another room of the house.

If you have an inexpensive, long cable and you moved it a number of times (or possibly even once a day) it is going to fail. One of the failures is the inability to deliver enough high quality power to properly run the microphone. Two possible symptoms of low power are low volume and high noise.

Koz

All understandable.

My USB cable is the one that came with the mic, and appears to be roughly 3 -4 feet. It’s not pulled taught, as my laptop is directly outside the recording space (I’ve cut a hole in the canvas to feed the chord out). However, since I’ve owned this set-up, it has been moved many, many times. The mic and cable were re-stored in their original box whenever I was not recording.

No special drivers or software for the Snowball. It’s plug-n-play.

I think I could have mistakenly faked you out on the “samples 1 2 3”. To keep my hands towards the back of the mic for easy switching of settings, I was probably leaning several inches closer. Sorry for that.

I would sooner produce a sample on another computer… but now my issue is that the two other computers in this apartment are desktops with noisy fans. But, I suppose that wouldn’t matter much if we’re simply testing for overall levels. I’ll move some things around and decide which computer would be easiest to record on.

In the meantime, I haven’t ordered a replacement mic just yet. I think I’d like to upgrade from the Snowball, for something that may last longer. I was considering the $150-$200 range, so I’d like to do some additional research. If you have any suggestions, I’m open. If I get so “in my head” about the mic replacement that it takes a while, I believe a friend of mine may have a Yeti or Snowball I could borrow in the interim.

Beware the Yeti Curse.

“What’s that singing whiny sound behind my voice?”

That’s a combination of ratty USB power, cheap USB cable and inexpensive microphone with no power filtering. It’s not easily solvable in hardware or software.

You can gently suppress your hiss and it obediently fades into the background (given a small amount of hiss). In the same sense that gentle rain in the trees on a summer day will put you to sleep. Not so the Curse. That sound is like trying to ignore a baby screaming on a jet. It requires such intense filtering and patching in audio post production that submissions fail the ACX “Overprocessing” test.

Flynwill on the forum did manage to get rid of the sound in hardware by basically taking apart and redesigning the device electronics.

The ACX demo video has a one step up installation as an example of a working studio. They use a good quality analog microphone and a USB microphone amplifier and digitizer. But it wasn’t $100.

I have specifics on a different computer.


Change the cable. This combination used to work, right? The part most likely to wear out is the cable and a defective cable can cause your problems.

Koz

If you have a USB printer, it should be the same cable. USB A on one side and USB B on the other.

Koz

Having the Snowball fail, you can’t simply go and Google up a better microphone. You join the Universal, Eternal Quest for a Good Home Microphone (to speak in book titles).

It’s such a dreadful experience that I’m experimenting with stand-alone recorders that avoid using the computer altogether. Apparently, there are other people out there tired of the battle and doing similar experiments.


I went back to the notes.

Attach 1 is from the ACX lecture of how to set up a home studio.

http://www.acx.com/help/video-lessons-resources/200672590

It features the MBox Mini-2 microphone amplifier and digitizer and the Rode NT-1 analog microphone into a Mac. I have a note in to ACX about the MBox. It’s no longer available new.

The Mac is important because they don’t make noise.

In the training video, they do mention the Snowball as an alternative (attach 2).

Koz
Screen Shot 2015-10-28 at 12.51.24.png
Screen Shot 2015-10-28 at 13.20.35.png

Let us know how the Yeti experiment goes. You may be one of the fortunate people that plugs one in and starts recording work.

Meanwhile, I’m testing an analog solution. There is a cheaper version of the setup that appears in the ACX videos. It was designed and built after ACX recorded the video. The stand-alone USB audio adapter, a good mic and maybe even a microphone stand may come in around $250 or so.

It will be missing a lot of the shortcomings of a USB microphone. For one thing, you can place the digitizer and computer anywhere on your property. No more four foot cable length restriction.

As we go.

Koz