I’ve attached a short sound clip of my issue below so people can get a feel for what I’m about to describe.
I’ve got an extremely faint clicking sound that’s also extremely consistent. It’s very, very quiet. The recording is pure, so you will have to probably blast the volume to notice it. But it’s there. And I can’t figure out what it is. It is something to do with my computer without any doubt of my mind. It’s not just effecting recording. When I have my headphones on, I can hear it in them when I’m not recording and when the mic is disconnected.
Any ideas? I’m trying to give recording an audio book a try and that’s when I noticed it, once I got everything as quiet as I can manage.
The clicks are almost exactly one second apart. Perhaps software conflict ?:
Audacity momentary interrupted as other software which tries (and fails) to take control of the audio ?.
Giving Audacity “above normal” CPU priority is worth a try, so it is less likely to be interrupted by other software.
I gave giving audacity more priority, but it didn’t have an effect on the issue. However, at the very least, it should prevent issues later down the road if some weird processing thing happens. But it did make me look through my task manager. I closed a couple of errant apps, and also looked in my sound settings on the Windows Sound Settings → App Volume and Device Preferences and noticed my input was, for some reason, my Line (Elgato Sound Capture). I changed it to my mic. Afterwords, I made a new sample recording and get this new lovely sound. No clicking. Of course, my girlfriend is awake now so the audio isn’t as clean as it was when I recorded at night, but that’s to be expected.
I have a feeling it was the input setting capturing weird noise from my capture card, because I can still hear the click in my headphones when the volume is maxed. However, it is no longer being captured by Audacity, which is the tiniest little annoyance off my shoulders.
However, one little sample I recorded now had me hearing what sounded like a cricket outside the window. Weird, since it’s the middle of the day now. My recording space is far from perfect. Would hanging a blanket from the window help with the bug noise? I feel like it would. Any other neat trick I could do to lower the room noise just a little more too?
I am in actually a rather large home space. It’s a room about 10’ by 12’. It does have some empty walls, but there are a lot of objects in the room to mess with echos currently. Televisions on the walls and things. Unfortunately, my desktop tower is in here too and that noise is picked up, I’m sure. Although as far as ACX requirements go, when I record my voice and some room noise, add in EQ, RMS Normalization, and a Limiter in that order per a guide I found on the wiki (https://wiki.audacityteam.org/wiki/Audiobook_Mastering) everything seems to pass the robot check. I don’t know if it would pass a human check per se, but I think at this point my biggest enemy may actually just me working with my own voice.
desktop tower is in here too and that noise is picked up
That’s what I think I hear in NonClicking.wav. It’s a combination of motorway trucks going by in the distance and gentle rain in the trees ffffffffff. The hiss (fffff) is almost certainly coming from your microphone system and is normal. The 18 wheelers (heavy lorries) in the distance is the fans from your computer. It would be good to get rid of that because it puts your live speech performance right on the edge of not passing technical compliance.
If you can tell your computer is on just by listening…
ACX noise specification means, in English, your noise level needs to be 1000 times quieter than your voice. A lot of home readers fail noise.
There are mechanical ways to help. I finished a Kitchen Table Sound Studio design using hardware store parts.
If you’re recording in a large bare room, you’re going to have echo problems and this will help with that, too. There is no software fix for echoes and reverb.
You can build a pillow or blanket fort around your computer as long as you don’t restrict air flow.
I think at this point my biggest enemy may actually just me working with my own voice.
It’s a little discouraging when after weeks of struggling with room noises and computer distortion we tell performers they just passed the first of two sound checks. Next comes Human Quality Control where you go to die if you’re a clinical asthmatic or stammer.
Everybody gets the first step wrong. Do Not move or breathe during that first two seconds.
I’d think people would miss that splitting to mono step.
I went through and put a little bit of a barrier between the PC and myself. It audibly made a difference to my ears, but makes less of a difference to my mic. I might go out now and find those supplies and whip it up. The biggest issue is that I don’t have a flat surface to put the mic down on, but rather an extendable arm holding things in place. However, I think I can design something that can work with that based off your design anyway.
Next comes Human Quality Control where you go to die if you’re a clinical asthmatic or stammer.
More of what I assume is a lisp. A half lisp. It’s more if I get ahead of myself speaking that it comes through on the audio. You can hear a bit of it in the sample I put in one line, and less in the other. I know this isn’t done after my modifications either, but then I’ll be able to compare and hear the difference too. I’m not just trying to succeed, but learn how to do it better and it’s hard to improve without record of failure.
The only part I can’t force to work is the big room echo.
At 9 seconds there is a serious echo.
“Where you redeem this card-ard-ard-ard.”
Turn the volume up a bit when you listen.
You got lucky. Most of the room noise is lower pitch than 100Hz which is the cutoff for the “Low Rolloff For Speech” filter, first step in mastering. That will get you to ACX passing, but by the thinness of a sheet of paper. I applied Noise Reduction of the Beast (6, 6, 6) and we’re done…except for the echo-o-o-o.
Are you going to build the kitchen table studio? That’s pretty good at killing room echo and reverb. If it also kills some of the room noise, so much the better. ACX doesn’t like Noise Reduction.
So this is just simple construction and arrangement. If you’re close to a Home Depot, not even that much construction. Tip. If you ask for furniture moving pads, they will send you to those tiny sticky round things you put on furniture legs to keep them from scratching up the floor. Ask for blankets.
I predict [holding fingers to forehead] you are going to get to the end of your first book or presentation, listen to the first few chapters, make a sour face and read most of it again. You start the read a Class A Newbie and finish weeks later a seasoned professional.
The long story is that I have to figure out an efficient way of using this thing. My current room layout doesn’t really jive with it. I had to move a bunch of things to stupid places just to get it to fit somewhere. That means I basically shoved everything on my desk onto the floor. Now, I did another sample recording which I’ll include. My ears might not be as trained, but I don’t hear any echo now. That said, it didn’t really reduce my noise floor as far as I could tell. I believe, with things setup like this, that I will still need to reduce the noise with the number of the beast.
Now, this might still be improvable. There’s a massive CRT in here that, if I can move it, I can free up space for this frame and things will be great. But it’s too large and heavy to move on my own so I’m pretty stuck with things as it is, at least until tomorrow when I can wrangle up some help maybe.
Current problems include not being able to see my monitor, and having to have my mouse, keyboard, and monitor actually inside the box with me. This seems a bit counter productive, but I’m not sure how much noise that’s actually producing. I’ll go ahead and submit a file of this thing too.
What you can’t see is that I am using 3 blankets, but they aren’t as heavy duty as yours. I weighed mine, they are 5.2 lbs a piece. There is one across the bottom, and two sharing the space on top. Additionally, because I had some spare blankets, I tossed some more fabric up there. As per specifications, all sides are covered from top to bottom. No external light is bleeding through the sides or the back, and it seems pretty sealed up.
Also, typing on the keyboard in here sounds extremely pleasant.
The design is different than the one you posted too because I don’t have a large enough, flat surface to set the design up in. Instead of four legs, it actually has 4 more 3 way elbow joint connectors and a box frame beneath it. Additionally, there is square of 3/4" plywood fitted into the bottom of the frame so there isn’t empty space where my desk can’t fill it up.
I had to move a bunch of things to stupid places just to get it to fit somewhere
In other words, you’re making a sound studio.
There’s a massive CRT in here
Really, glass monitor? Cathode Ray Tube? Not a flat LCD monitor? Takes two people to lift it? If it is a glass monitor, those things can radiate powerful signals from all the internal magnets and scanning electronics.
That appears to be the end of the room echoes, or they’re far enough down not to be noticeable.
There’s a noise in the first two seconds I can’t identify. Starting just after the half-second mark, there is a rumbly noise that comes and goes. The attached clip has the two second segment intentionally boosted. It doesn’t really sound like that in real life. What in your studio could sound like that?
I will still need to reduce the noise with the number of the beast.
I agree, but that’s for the microphone hiss (fffff), not the rumble. Noise Reduction doesn’t work on sounds that change.
having to have my mouse, keyboard, and monitor actually inside the box with me
There’s another rule here. If you struggle for longer than about two or three weeks with the computer, stop using the computer.
This is a stand-alone sound recorder, the Zoom H1n. It’s the lowest end of the series. One poster is well on her way to setting up a Zoom H5.
That’s my photo setup. Really, you would be moving that whole thing into your cotton sound cave. In my case, I replaced all that wood with my messy, but silent third bedroom. What you see sitting there will produce an ACX compliant sound clip after simple mastering and no noise reduction.
There is good news. As far as I can tell, you don’t have The Yeti Curse, also called Frying Mosquitoes. That’s a bad interaction between the microphone and a ratty USB connection.
This is a Zoom H4 (no “n”) inside a cotton sound cave.
You don’t have enough to worry about, so how would you fix the reading error at about 14-1/2 seconds? One way is to make a long pause (curse if you wish) and read the whole sentence again. Edit later. Some readers clap or snap their fingers to make the error easier to find later.
“Which contains more than twenty times…” Do Not try to fix one single word.
The newest Audacity version has Punch and Roll, where you can stop, back up the cursor and fix the whole sentence in real time.
Really, glass monitor? Cathode Ray Tube? Not a flat LCD monitor? Takes two people to lift it? If it is a glass monitor, those things can radiate powerful signals from all the internal magnets and scanning electronics.
Really. I used two foot pipe sections to make a giant cube cave. The television would not fit inside. Not that I can lift it inside anyways! I cast retro gaming on Twitch and I use the thing to play NES/SNES games on.
That appears to be the end of the room echoes, or they’re far enough down not to be noticeable.
Yeah. When I compared to my original clip to this, I never heard anything quite as dramatic as card-ard-ard-ard in the first place, but I could definitely hear what I would call the hum at the end of the word. The cave does a good job of getting rid of that. It sounds cozy in there. It’s too bad the hardware store that sold me the blankets also has the blankets smell a little like wood varnish right now. I’m sure it’ll go away shortly. I hope.
You don’t have enough to worry about, so how would you fix the reading error at about 14-1/2 seconds?
I haven’t gone out and done any extremely long versions of audio yet. I read out an audition for fun (not submitting that) to get a feel for the software a little more. How I did it then was that I’d select some natural points where I could rerecord and split the track at those points with CTRL+I. I’d re-record the lines, snip snip, and put the newly recorded line in place. That’s if I didn’t notice while I was recording in the first place. I’ll have to give the punch and roll thing a try on one of these lines in the future. Maybe intentionally make one up soon just so I can try to edit it out.
If I knew I read the line wrong, I’d stop for a second, then just repeat the line. Going back over the recording, I’d just delete the incomplete and misspoken sections. I’m quite obviously extremely new to the process and haven’t come up with some kind of thing to introduce to the track to make my mistakes more obvious to find. Although, I’m also new enough it’s more of a challenge to find the parts that aren’t filled with mistakes!
Honestly, that H1 Zoom looks right up my alley. It’d open up a few venues in the house where I could set this cave and work. It’d make it just so I could shut off the computer in this room and probably be more than fine too. Most of the floor is definitely the tower doing its processes. I can audibly hear them at any time of day it’s noisy enough. The problem is coming up with a quick hundred for the improvement. I might see if I can scrounge it up though. It’d really just be easier to get rid of the PC in the process and only need to go back to it for editing.
As for that rumble noise though, I believe I have recaptured it’s wonderful essence right here for you.
That noise appears to be the noise of me giving my desk a bit of a shake. It’s more dramatic in this clip because I put my hands down on it and gave it some more oomph. I didn’t know what it was when I made the clip, but being made to think about it made me think of all the processes I had to do to make it. I had to reach into the box. I had to extend my arms pretty far to hit the R key. I had to withdraw my arm from the inside. Then I probably shifted my body to get closer to the mic in the box. Any one of those things could have resulted in a gentle shake of the desk. All the processes should hopefully vanish by replacing the CRT with the box since I won’t have everything stuffed in here then.
I decided at this point in the message I’d try out punch and roll too since I’m already set from shaking the desk.
I feel like I did this right. When I reached the little pause in the sentence, I flubbed up and said two dozen. Found the area, did the punch and roll, heard my audio going and was barely prepared for when I needed to speak. So there was also an awkwardly long pause in the middle. I trimmed that out and that’s what you’ve got there.
Personally, I listen, and I can’t tell if I should trim out a couple more milliseconds of dead noise or not, but I don’t think my untrained ears could pull out that it wasn’t done in one take either.
Now, very interesting to me, is that why my RMS is a bit lower than it has been because I didn’t lean up as close to the mic while doing this, it appears my noise floor is at about -70.9db in raw. That’s quite a bit quieter than it was last night with no changes.
After applying audiobook mastery to that retake (with RMS normalize at -19db) it still at -60.7db, within range. It makes me wonder why it’s suddenly quieter.
Now, I didn’t make any further edits to that second take. If my newbie analysis is correct, the sharp foil noises at the beginning are me shuffling the magazine I was reading (I think I realized this happened as the work happened because I left myself an extra bit of time after to start speaking. Good job, me. I’d get rid of that by either clipping it or just replacing it with some otherwise silent room noise. At the end of the clip there’s another noise, which I think is me reaching in and touching the keyboard without properly bopping the spacebar right away to stop the recording. That one I wouldn’t replace with anything since it’s at the end of the audio and I’d just clip it out altogether.
Some time this afternoon I’ll get my gear in the right place.
blankets smell a little like wood varnish right now.
Leave them spread out in the sunshine for a while.
I used two foot pipe sections to make a giant cube cave.
Design is a juggling act. The original 20" pipes made it so you didn’t have to pay attention which direction you laid the blankets. Either direction would work. With the pre-cut two foot pipes, you have to put the blankets on so sideways was the 70" dimension, or the blanket wouldn’t reach the table. Making all the pipes the same was intentional. Can you imagine the craziness if certain pipes always had to be in certain places on the cube?
You can totally make the thing much larger, but you would need more blankets and make sure they overlapped in use. The original design didn’t have a blanket on the table and it didn’t work nearly as well as two blankets, one top and one bottom.
People have used bedding and packing material for soundproofing. In both cases there is a premium on not weighing anything and that doesn’t work for sound. Heavy is good. Sound has to move the blankets to get in, or in the case of echoes, get out and then get back in again. That’s why your sound “tightened up” and got clearer when you started using it.
That noise appears to be the noise of me giving my desk a bit of a shake
Home performers all fail noise. It’s hard to think about background sound being 1000 times quieter than your voice. Sitting here right this second I can hear traffic outside, the fridge running and a jet going over—and this is a quiet neighborhood.
The Zoom series of recorders need a memory chip to work. It’s not optional and it doesn’t say that anywhere in the advertisement or promotions. Also note this is a battery operated recorder—it’s isolated from anything that can cause problems. They will sell you a power supply to plug into the wall. I didn’t get one so I have no idea if it affects the sound. You can be the test case.
Don’t fall in love with that number. You should hit at least -65dB to be unconditionally safe. One poster had a comment from ACX which I thought was cute.
“Yes, your submission passes noise, next time see if you can make it better.”
They have been known to reject noise not because the submission actually failed noise, but for the software patching the reader had to use to get it there. That’s why you shouldn’t fall in love with Noise Reduction, either.
Analyzing a post-production sound clip isn’t all that useful because we’re never quite sure what you did to it and if there’s a mistake anywhere, we could be chasing each other for days. Clean, raw samples only.
Punch and Pepper.wav volume is too low. If you watch the bouncing sound meter, it has sound peaks in the -18 range. It should be in the -10 range. Sound goes double and half every 6. so you’re worse than half the volume you should be.
There’s another trick. Can you overload Audacity? NEVER blow into a microphone, but can you speak or yell loud enough to make the Audacity sound meter go all the way up and turn red? You won’t damage the Yeti. It has a maximum sound volume rating just short of a jet engine. If you have to scream to do that, it means the volume control on the back is too low. I would not be shocked if you settled on max or close to it.
There are microphone controls on the back of the Yeti. Make sure your pattern is #3 (heart or kidney shaped). Just to cover it, you should be speaking into the grill just above the company name. It’s a side-fire microphone, not end-fire. Don’t talk into the top.
Home sound systems tend to “gentle” volume because New Users can get into a lot more trouble from overload than quiet sound.
There’s another trick. If you place the microphone off your left or right cheek instead of straight in front, you can get a lot closer and louder. Spacing could be as close as one power fist.
Also, it sounds like you’re starting to nod off at the end, there…zzzzzzz
I’m actually a very loud person. All of these recordings have been done from what I’ve gathered is a shaka, or greater, away, with my mic gain set to minimum.
When I use my mic, and I put myself at one shaka away, my voice peaks at around what I assume is that range when I say things like card. We had a funny thing there before. I have the gain knob on the back turned all the way down. If I do a powerfist on minimum, I peak saying similar things at about -5db.
Here are samples, because I may be reading into things wrong. The only thing I’m judging is an approximation of where the little blue line ends up at the end. Both at minimum gain, but both at the filenamed distances instead of further away.
Also, it sounds like you’re starting to nod off at the end, there…zzzzzzz
You have a good ear. I woke up after like two hours of sleep, saw the reply, and decided to get some work done. After, I went back to bed. So you are probably entirely correct.
I just realized as I was recording the files to insert into this comment now that the ACX check plugin tells me what my peak is. Powerfist appears to be -4.7db and the shaka at -7.7db. Obviously in this case the shaka is a better distance, but the peak is still too loud. I can get further away still from the mic, but I can’t lower gain any further. Is it better to be too loud than to be too quiet? Should I meddle with something else to hit a better level? Both files were read speaking toward the blue yeti logo on cardiod mode, the heart, on the mic. I actually just figured out the setting a couple days back while trying to set things up before posting at all.
The entire time I have been streaming on Twitch I had the mic tucked in a corner about three or four feet away from me, with objects in front of it, on bi directional, and the gain up around three quarters, and then the gain dropped in Twitch quite a bit. If nothing else, this whole process has made me sound better in OBS!
First, I moved the giant CRT behind me. It is a temporary home for the old monster, especially since I both like to use it and it makes it so I can pull my chair out very far. However, it free up most of the desk too. I had to move my monitor because the box is so large (I may, in the future, make it smaller so it fits better and also gives me more space for the screen, 2’ bars is very large).
I positioned the yeti in there to make it look ominous as it emerges from its cave. However, I actually position it nearer the front of the box. Fun fact, beneath this box is a signed photo of Terry Crews on a fuzzy rug wearing nothing, but covering up with a basketball. The things you’ll find in people’s offices, eh?
I placed two 5.2lb moving blankets on top, slightly offset to either side from one another, and one on bottom. I gave myself a little wiggle room at the end with the fabric as I tucked in the outer layer under the box. You can see it at the top, there’s actually a third regular blanket, on the heavier side but not as sturdy as the moving blankets, draped over the right side. That’s to help further reduce the fan noise, and it also hangs over the top edge a bit unlike the movers blankets. I thought it created a cool dark space, and it also will probably, very slightly limit my voice leaving the box and bouncing back out on the wall behind me, which is a bit less filled up than the wall that was behind me previously. Additionally, there’s a turtle tank over there and it likes to scratch the walls. I’ll have to consider what to do about him.
The fabric also is draped in the back of the box in a different way. It hangs behind and down below my desk for the outermost blanket, and the inner one is draped along the inside on the bottom board. Again, this is to help limit the computer’s fan noise somewhat, as I figure the small space behind my desk will be filled up a bit more with the blanket, and that vibration will have to travel a long way through the blanket to make it to the box.
Now, with everything out of there, and with things moved and manhandled, I also took another step. A step into the power cord for the PC which turned everything off. Gasp. So I plugged it all back in and noticed my fan was making a different noise. A louder, wobbly noise. I didn’t like it. I opened the case, cleaned the fan off, gave it a little tug, and now it’s smoother sounding than it was the first time and definitely leagues better than the one that drew me to it.
I apologize in advance for this recording, but I’m starved on time to make the post unless I want to wait several more hours for everyone to go to bed. The kids will be home soon. This is the raw audio.
Here I spoke approximately a shaka away. However, my head was tilted down to read the magazine, and I was speaking to the side instead of directly at the mic. I think this made my voice quieter over all, and a bit sloppier. Additionally, I just ate some peanuts so… meh. Sorry. Anyway, the noise floor of this clip appears to be about -71.8db without any adjustments. After putting on audiobook mastery, it’s at 61.7db. These are by far my best numbers yet.
If it’s true and sound goes up double and a half for about every 6 db, I’ve definitely cut my noise floor’s volume in half with these changes. If I understand things right. That’s a big win.
Now, I think, my biggest challenge will be to battle with mic positioning, mouth placement, my own voice, and figuring out how to read something while facing my mic while it is in this giant box. I have no idea what will work yet.
But even if that’s my biggest challenge, I am still 100% open to new ideas and learning more. I’m very patient, and if it comes to the end and I’ve narrated a book and I have the newbie’s curse where the first few chapters are so garbage I want to rerecord them, I’d love it to be just because I’m bad, and not because I couldn’t figure out audio!
If your computer went through Catastrophic Power Failure, you should Shut Down (not restart), wait and bit and then Start. When it settles down, Shift-Shutdown, wait and Start. That should clear out all the debris and orphan junk that collected when the power went away.
Noise is much lower and better behaved. I would apply a touch of Noise Reduction to get it below -65 before submission.
But more much more serious, room echo is sneaking back. Is there more than one microphone in your system? Do you have a headset and do you leave it plugged in while you announce into the Yeti?
What happened when you tried the yelling intentional overload trick? Could you get there? Good engineers know where both extremes of the equipment are, overload and low volume noise.
There is technically a webcam in the mix that can draw do audio. I’ve never used it for audio, but I suppose it could be a culprit of something. Let me try disabling it.
