Issue with a hiss/whine in the background of some recordings

My zoom arrived today, and I haven’t had a chance to mess with it apart from a few test recordings. I am very impressed with it - I nearly pass ACX check with a raw recording.

Just for the fun of it, I’ll upload a sample sometime this weekend. It certainly sounds better than the Snowball and ATR, though this may just be my need for this to be the case shining through.

You asked about which problems I’m solving - there are three, and they’re all convenience oriented. 1) girlfriend gets her laptop back. 2) I no longer need to transfer the audio with a separate USB device. 3) the laptop’s fan seemed to still get picked up by the Snowball, however subtly.

Oh, and I’ve got a pop filter clipped on to the TV tray I’m using. So it’s a TV tray, a thick blanket draped over it, a stack of books, a toilet paper roll, and then the mic with pop filter. I’ll be speaking directly in to it.

I watched some videos on the device, and it’s recommended to merge the left and right stereo channels of its recordings in to a mono file to ensure less “space” in my voice - will ACX accept mono files? Additionally, are there other tweaks you’d recommend to the Zoom’s recordings, apart from the standard ACX macro?

Thanks so much as usual.

I nearly pass ACX check with a raw recording.

See? It’s a fairly competent recorder and it has no computer problems during the performance.

will ACX accept mono files?

They prefer mono files by a wide margin. We are a single person reading a book, not producing radio theater. All stereo does is take up twice the digital room and reduce the transmission speed by 2, all for no good reason.

There is one decision. How to make mono. I did it by adding Left and Right and divide by 2. Mix To Mono will do all that automatically. You should have few if any directional problems by doing that and the background noise will be slightly lower. If the musical tones seem OK, then that’s the way to go.

The microphone noise doesn’t halve. There’s a formula for it. Square root of the doobly-do… I gotta look it up.

Did you decide whether or not you’re going to connect the USB cable to transfer files, or pop the little chip out and buy a chip mounter for the computer. Both are valid, but my money is on the USB cable. If you screw up the chip socket, that’s the end of the recorder. Full Stop. If the USB system dies, you can still record and transfer with the chip.

There is a “down the road” decision. As you get rolling with your career, consider what you would do if the little Zoom goes into the bin. I don’t have two Zooms, but I do have two super good ways to capture my voice.


Additionally, are there other tweaks you’d recommend to the Zoom’s recordings, apart from the standard ACX macro?

No. ACX doesn’t like distracting production. If you sound like you, that’s good to go. The goal is listening to somebody telling a fascinating story over cups of tea.

They may object to tongue-ticking, lip smacking and gasping. That will salve your need to apply corrections. Those theatrical “errors” can be rough to fix.

Depending on how far along you are, you can submit a short test to ACX.

the standard ACX macro

A note on that. There is no standard ACX Macro. The Party Line is to apply the three effects, in order one at a time. Someone created a Macro that jammed all the effects in quickly one after the other, but sometimes the macro gets it wrong. Effect > Equalization doesn’t always work right under Macro control. It never got fixed.

While there is something on the timeline, Effect > Equalization and see what’s there. It should look like a dip-down-left which is the standard correction, but it may or may not say Low Rolloff for Speech in the Select Curve box.

Screen Shot 2019-11-01 at 17.04.53.png
So the correction and the name don’t match. I don’t know what it actually does when you apply it like that—or the Macro applies it like that. Worse, I don’t think it calls the right curve if you needed theatrical equalization for an effect and then Low Rolloff.

There is a way to check the work itself. Select some work > Analyze > Plot Spectrum. I expect almost everything to the left of 100Hz to be gone. That’s rumble and Low Rolloff is a rumble filter.


You can also submit a raw test here. Follow this recipe.

Go down the blue links. They’re pretty short. Everybody misses the silence at the beginning.


For now, here is a raw sample recorded using the Zoom on gain 7, just in my little walk-in closet. Let me know what you think.

My only cursory thought at the moment (this is based on a few repeated recordings) are that the letters S, F, and the occasional P (all plosives, no?) can come across harshly at times, and that it’s very sensitive to bumps, but that’s pretty reasonable, considering it isn’t mounted on anything except a toilet paper roll and some books.

The SS sounds are sibilance. Try another pass with the microphone in front of your cheek instead of your nose. Like that oblique position up the thread.

I got most of this essing to go away with the DeEsser tool.

This is Mastered and DeEssed. It easily passes ACX technical conformance and the theater is nearly perfect.

While that simple speech is nice, it was supposed to go for 18 seconds to give us many different words and sounds. The smaller/shorter the sample is, the more likely we are to miss something.

One of the complaints about the H1n is handling noises. Nobody will be hand-holding a interview with an H1n.

You can all but eliminate floor and table noises with the Book and Towel method.

Each one rejects different sounds. Combine them and nearly no sound can make it up through the desk.

toilet paper roll

I used three toilet paper rolls as I needed a little more of a boost than the hand towel roll in the illustration. Neither one makes any noise by itself if you strike one with a pencil.

If you decide to go with it as is, this is the De-Esser tool.

And these are the settings I used—after mastering. DeEsser is an absolute tool. It will not automatically go look for problems.

Screen Shot 2019-11-02 at 12.51.38.png


Here’s our second raw take, this time pointing toward my cheek, and hopefully of a suitable length.

It’s fine, but your performance volume went into the mud. This is a sample of your first post connected with your second.

You can adjust this yourself. The blue waves will tell you what you’re doing. The first post was nearly perfect. This is a screen capture from the first one followed by this latest one.

Screen Shot 2019-11-03 at 12.35.03.png
See the blue tips on the first clip almost meet the 0.5 marks up and down? That’s the goal. The volume meters on the front of the H1n will tell you , too, but those are a little rough to set because you can’t announce and see what you’re doing at the same time.

Moving the microphone didn’t give all that much improvement in sound quality, and this one fails noise.

So it’s official. Record like you did the first time, correct fluffs and mistakes as needed, Master and apply DeEsser.

…French Legal System…

That’s from ‘Crash Course’ with John Green, right?


I’m just now waking up. You didn’t add the two microphones or tracks together, right? You picked one and deleted the other one. That may mess up the moving microphone trick. That can have odd results if you only use one of the two microphones. So either add the microphones or don’t use the sideways placement trick.


Moving the microphone didn’t give all that much improvement in sound quality, and this one fails noise.

It was pointing at my cheek, but it was definitely more faint, obviously.

That’s from ‘Crash Course’ with John Green, right?

Nope, just from a rough and ready audition I thought I’d throw together today, mainly for the purposes of showing you a halfway decent sample. I did master it and send it off to a rights holder, but I don’t tend to get nonfiction roles, usually.

I’m just now waking up. You didn’t add the two microphones or tracks together, right? You picked one and deleted the other one. That may mess up the moving microphone trick. That can have odd results if you only use one of the two microphones. So either add the microphones or don’t use the sideways placement trick.

Simply split stereo to mono and deleted a track. That’s the maneuver, no?

I’ll get dee-esser and carry on, then. I don’t like to get too optimistic with my precarious recording situation, but this is by far the best I’ve had anything sounding.

An aside, but, you mentioned the H4N recorder earlier. “Professional” reviewers on YouTube seem iffy on the H1N for voice work, (probably re: specifically voice acting, not audiobooks) but, rights holders on ACX don’t seem to mind the quality I’m turning out. Would the H4N provide a marked increase in quality, in your opinion? (I know you’re not a sales rep, so expound as much or as little as you’d like on this) I watched a couple reviews that more or less claimed the H1N and H4N were pretty close to one another in terms of actual audio quality output from their built-in mics.

Thanks again, Koz. Your attention to detail and concern for the lowly are commendable traits!

Simply split stereo to mono and deleted a track. That’s the maneuver, no?

That’s one of them. The other one is Select the stereo track > Tracks > Mix > Mix Stereo down to Mono.

The two microphones on top of the H1n aren’t strictly Left and Right. I believe they’re matrixed to Mid-Side inside the unit and then remixed to Left and Right on the way out. So if you move the H1n to one side and pick the wrong microphone, you could get what you got. Higher noise and low volume.

Try it again and this time Mix to Mono. The rest of the performance was perfect.

I don’t like to get too optimistic with my precarious recording situation

Don’t tell anybody. If people are writing you checks, particlarly if you passed ACX, nobody cares how you did it except the recording community who use it as an opportunity to offer opinions.

There was a Zoom that was noticeably better than the others. The original Zoom H2 could crank out terrific sound and shows one after the other.

Of an Indiana summer’s evening. Thanks to Kelly, Sam an Donny for helping out.
Recorded on a Canon Rebel T3i and a Zoom H2.

Audio recorded on a Zoom H2 handy recorder, video on a Sony Bloggie touch, edited with Acid music studio and Vegas Pro 10.

This one’s special. You can listen to Josh switch between the camcorder sound and the H2.

That’s the original H2, not the H2n. The original H2 that’s going for original equipment prices previously owned but in good shape on eBay. I would not buy one. You have to go through the learning curve all over again.

I can sense Marketing and Promotion at Zoom. I would not be shocked if the H2 and the H4 used the same microphones and just gussied up everything else.


I wonder if anybody on YouTube posted a Zoom-Off.


I think this is one of Josh’s originals. It’s got a lot of echo and reverb, but still. It’s him and an H2. Currently at over nine million views.

I do have an H4, no “n”. I think it’s reached end of life. It has system noises the last time I tried to use it. It’s very old.

As a fuzzy rule, we don’t recommend specific microphones. It’s good to have actual experiences with the stuff recommended and there just aren’t enough of us to do a good job. The H1n was the sweet spot for me. Good recordings if I couldn’t drag out the big guns.

That was a radio broadcast sound shoot. It’s taken in a fully soundproofed conference room with sound mixer, boom, blast filter and backup recorder.

We didn’t need the backup. The station accepted the track with no corrections or comments.

Another reason for the H1n is to have a recommendation if someone just couldn’t get computer recording to work right. How many dollars are required to do a reasonable job.

You can get sticker shock with some sound jobs.

He’s holding about a grand USD.


Thanks for providing so much helpful insight and devoting all this duty-free time to my silly projects. I’m certainly learning a lot here.

As a passing question, is there a chance that a De-esser could very, very subtly mess with the background “hiss” in a recording? I was listening back to the first and only recording I’ve De-essed from the Zoom, and the rainforesty hiss became more of a gentle … almost faint static hiss? You need to have the volume cranked and earbuds to even halfway notice it, but it’s there, and I’m not sure if this is something ACX will flag.

I assume it’s the De-esser anyway. I’ve done several recordings after this one that turned out just fine - gentle rainforest hiss, no static at all. The settings I used on the Spitfish De-esser were just lifted from a YouTube video, so maybe that’s where I went wrong?

And the RH just approved said recording. So perhaps I’m overthinking things, but it’s pretty jarring to me.

You need to have the volume cranked and earbuds to even halfway notice it,

Diving for Noise doesn’t count. Set the playback volume for comfortable voice listening, roll the performance back to the noise and don’t touch anything. Still hear it?

Does the work pass ACX Check with a noise -65dB or quieter? That’s all you need to pass technical conformance That is assuming the noise is natural rain-in-the-trees shshshsh sound. There are other noises which stand out, are very annoying and take extra attention, but microphone/electronic noise isn’t one of them.

The recommended DeEsser has no significant affect in volumes quieter than the Threshold setting. That’s why you have to Master first to put the performance volume in the right place and then apply DeEsser.

I’ve never used Spitfish. You are now the expert.

To catch up: Noise, Volume (RMS) and Peak are the basics for acceptance. A robot checks for those and if you don’t make it past the robot, then somebody will post a nice “try again” message and reject you. If you pass the robot, a human reviews your work and that’s where the theater is important. They hate distractions, such as SS sounds that will cut glass, stuttering, gasping, wheezing, lips smacks, etc. As a fuzzy rule, you can’t get a machine to judge that.

So that’s where you are now. Avoid obsessive nit-picking. Nobody is going to turn the volume all the way up and inspect atomic-level random molecule noise. That said, they might be concerned if what noise you do have changes as you talk. That’s can be distraction and they just hate that.

Did you like your work with Mastering and DeEsser applied as I posted? If you started with loud enough performance, that should slide right through.


Well, the noise passes as our fabled rainforest hiss until about 90% earbud volume on sound-proof headphones. Then, while still not much of an interference, it is noticeably unlike the room noise I’ve had in any of my recordings prior to now - just a little rough, almost crackly. I certainly don’t notice it on car or computer speakers though. No changes in the quality of the sound throughout and it doesn’t fluctuate as I speak, no. It’s consistent with the rainforest hiss, but is just crackly for some reason.

Nitpicking is extremely in my nature but I realize I’m not doing my mental health any favors by sitting here and obsessing over this. More than likely. Probably. Whoa, my vision’s blurry…

That said it passed ACX check with flying colors. I did as you said, mastered and then de-essed. Overall it’s an improvement, but for my hopefully inconsequential nitpick.

You can attenuate the rainforest (when you are not speaking) with an expander plug-in e.g. the free version of couture.

If you’re not done obsessing yet, I found a way to use Noise Gate without doing too much damage.

You have to download and install it.

Noise Gate suppresses sound lower volume than a setting.

This message thread describes how I did it.

I used Mastering to get the work in range of Noise Reduction, and then gentle Noise Reduction to get the work within range of Noise Gate. Both Noise Reduction and Noise Gate are set for very modest affect, but still, between them you can get about 18dB of noise reduction, serious noise suppression and nobody can hear what you did.

Have we filled up your dance card yet?

Let’s see. Announce the work > export a safety WAV > cut or edit and correct it > Master it > DeEss it > Noise Reduce it > and Gate it. These steps should not affect ACX Check, or if they do, it should leave the work still within range. DeEssing in particular might reduce the peaks slightly. Peaks are a limit, not a range. As long as they don’t go louder than -3dB, you’re good to go. Essing should not affect RMS (loudness) at all.

If you miss any of the steps, you have to start over because some of the corrections depend on the order of tools.

It’s strongly recommended you export each raw performance as WAV and move it to a safe place as backup.

If you do read a book, the chapters must match, so you have to do all that to each chapter exactly the same way.


just a little rough, almost crackly.

Post some of it.

Did you speak too loud? The blue waves are never supposed to go all the way up or down. View > Show Clipping. That will produce red marks at all the places there is overload sound damage.

How’s the battery life holding up? Did you get the wall power supply? I don’t know what it looks like when the batteries start to go flat or what the warning is. I didn’t use it that much yet.