Audio levels when recording

Hardware: Focusrite 2i2 (2nd generation) + Behringer C-1 (XLR 48v phantom power) + pop Filter
Software: Audacity 2.10 + Windows 10

How can I record loud and clear audio ?

I have the gain knob on the 2i2 set to about 70% and the monitoring level in Audacity dances around the 54 mark without me talking. When I record with the mic about 6-8 inches away the levels go up to about 30-27. The waveform is almost flat with the peaks struggle to reach 0.1 if I talk loudly
Sure, if I turn up the gain on the 2i2 to 100% it might go up to 0.2 or 0.3 if I talk loudly, but this introduces loads of background noise and the microphone picks up everything.

So, if I amplify in audacity , the loudness goes up, then I have to do noise reduction to loose the background “hum” but now my audio sounds artificial .

I have tried a Rode videomic plugged into the on-board soundcard and have exactly the same problem. I have tried 2 PC (both Windows 10) same issue.

How can I get a nice “Fat” waveform without the problems with background noise which I then have to clean up ? Am I missing something ?

Looking at tutorials on youtube etc everybody seems to get audio that peaks at 0.5 with hardly any background noise whilst speaking normaly . Am I doing something fundamentally wrong ?

First things first. The C1 is a directional, side-fire microphone. You should be speaking into the little red light (which should be on), not the back and not the end.

The C1 is not a fire-breather; it’s not particularly loud. So I’m not surprised you need to present a little aggressively.

It’s the 2i2’s job to boost the microphone volume and convert it to USB for the computer. It’s not that unusual to need to run the volume controls all the way up. If it works there, then that’s where they need to be.

this introduces loads of background noise and the microphone picks up everything.

Nasty surprise: that’s what your “studio” sounds like. The microphone picks up the room. You need to arrange so you’re the loudest thing in it. This is where people start messing with soundproofing—or looking for a quieter room.

I got insanely lucky. The guy before me played drums, so the third bedroom is soundproofed. It has special tile on the walls and carpet on the floor.

You don’t have to treat the whole room. There are ways to create a tiny studio or sound tunnel to announce in. I posted a thing about this for someone else…

Let me go look.

Koz

Found it.

https://forum.audacityteam.org/t/i-passed-acxs-qc-but-the-end-result-sounds-horrible/42749/8

This is part of an on-going conversation, so it goes on before and after this post.

No, this is not unusual. The thing about parking a microphone on your desk and reading audiobooks straight-off is something of a fairy tale.

Koz

Thank you so much ! Just speaking into it the right way helps an enormous amount (surprise surprise !). I had it pointing backwards so it was picking up my computers hum with me shouting in the background . My levels are now hitting with 0.3 level easily and sound so much better.There is so much to learn about audio, going to be fun.

The microphone wasn’t the most expensive and I have a dirt cheap stand for it so I know that the quality wasn’t going to be brilliant, it was the audio levels that threw me. At least I know what bad sounds like, so I can now head towards good.

Again, Thank you so much for that.

What’s the goal? Read audiobooks and be internationally famous? Podcast?

Computer noise can be a serious problem. ACX AudioBook has some serious restrictions for background noise, and it’s checked by a robot, so you may not even get past the bouncer with the bulging muscles.

We have some tools and processes for audiobook readers, if that’s where you’re headed.

Koz

No, nothing that fancy. Toying with corporate videos and instructional videos for work. You just have to start tinkering with Audio and you appreciate how much hard work goes into good sound. I was listening to Stephen Fry doing one Harry potter book that went of for 11 hours !! When you realise how long it takes to do a 5 minute podcast with not umming, you really start to appreciate the skill of the actors and technicians behind audiobooks.

You just have to start tinkering with Audio and you appreciate how much hard work goes into good sound.

I am so going to have to lie down for a minute.

You are miles ahead of all the posters who appear here wanting us to make them “sound more professional.”

There’s two leaps right away. You know about needing to be careful about the original shoot, and you know Editing Is Everything.

One forum note:
This is a volunteer forum, not a help desk. That means our timing can be a little ratty as our availability ripples across nine time zones.

The New York Times has relatively recently graduated to video interviews. Given who they are, they never had trouble attracting guests and hosts, but only in the past bunch of months have they had actual audio with actual microphones and not top-of-the-camcorder, recorded in the bathroom sound. I didn’t pay a lot of attention, but I believe I saw a credit for a teleproduction company. It was a welcome change.

As we go.

Koz

When you realise how long it takes to do a 5 minute podcast with not umming

Most people do Ummm. That’s what gives you the 25 minute editing time for a five minute show, to correct the mistakes.

If you don’t do that, you get an “On The Fly” podcast which is painful to listen to because it’s just uncorrected ramble.

That’s the Five to One editing rule. It takes, on average, five times the length of the show to cut it. That’s a nasty surprise for people cranking out two hour shows. That’s over ten hours of editing, if you bother.

There is one tool that can help. Audacity has labels and you can apply them at the goofs on the fly during recording. When you get done presenting, go back and look for all the labels.

http://manual.audacityteam.org/man/label_tracks.html

If you need to stop and let the dog out, you can Pause the recording with the P key, or you can Stop dead and pick it up again at that exact point with Append-Record (Shift-R).

Audacity will not Crash Record. You can’t go back, start recording at an arbitrary time in your track and wipe out all the trash you produced. That’s a popular Feature Request. There is a sorta crash-ish kinda-record. You can go back in your track and start recording and Audacity will start a fresh, new track at that point. Then it’s up to you to delete the old trash manually.

Audacity will smash it all together into one file when you Export.

Koz

thank you very much for those very useful tips. The labels one is very interesting.

Can you point me in the direction to a newbies tutorial on setting up a workflow when using Audacity. At the moment I seem to be stumbling around. I don’t know whether you should apply noise reduction before or after amplifying or even if using noise reduction and amplify are the right tools to use. I am trying to not have to fix low or noisy audio in “Post” but have yet to find the sweet spot. I suspect I will need to pay more than £40 for a microphone to realise that dream.

I will need to pay more than £40 for a microphone

That’s a popular New User Lament: “If only I could get a much better microphone, all my problems would be over.”

Probably not. Unless you have a Specialty Recording Problem such as Drums, Landing Aircraft, Windy Beach, or a Multi-Party Roundtable, what you have, unless it’s broken, is probably serviceable. I generally use this illustration of what you can do for soundproofing using furniture moving blankets, but in this case, regard the microphone.


That’s an SM58 rock band microphone playing into a simple sound mixer and into a Mac.

That’s it. I shot multiple voice tracks for an animated production with that.


We have generic tutorials.

http://manual.audacityteam.org/

That’s the on-line documentation. Scroll down until you get to Tutorials.

You really want the semester on entertainment recording. What exactly are you doing in that 25 minutes required to get your 5 minute show ready for publication?

I don’t have a good answer for you. People wonder why we don’t have YouTube video instructions on how to do all this stuff. Some people have actually produced documentary videos on some of the Audacity tools and how to use them. Nobody wants to hear about the five to one editing rule and nobody with a pulse wants to hear about it when we find mistakes and want them to do it over.

Terrific video. Really. And it will be perfectly valid until the new software version comes out next week.


Your question rolls back to why this process usually involves you screwing up and then us fixing it. It’s usually the most efficient that way. There is a rainbow of different things you can do wrong and it’s just not practical to preempt all of them.

ACX does have a couple of videos. They start with how to record right after you install your sound proof booth. That’s where this is taken from.

They know bare floors just kills most people, so they start just after that. That’s how I do it. Soundproof third bedroom. No bare floors.

If you record just right, you can present a chapter, cut off the ends, make fluff corrections, reset the volume (capture volume and ACX volume are different), Export a WAV backup and the MP3 ACX submission and go make tea. There is no list of corrections, filters, effects and other production tricks.

I shudder when people describe doing tonal corrections to each word in their production. Their third audiobook is going to be a lovely thing, but take them right up to retirement.


This American Life has an associated group called Transom. They have some very good works on “how to make radio.” They take it right down to equipment part numbers and how to shoot in various conditions. I think that’s where I got this on-the-run interview picture from.

http://transom.org/

And you can always submit a raw clip and we’ll tell you an opinion of what to do with it. It’s rarely a good idea to submit processed work. We can’t take processing out and we may need to.

Did I post this?

http://www.kozco.com/tech/audacity/TestClip/Record_A_Clip.html

Koz

Thankyou for the advice. I have attached the a 20 second clip (with 2 sec quiet at the beginning) we me talking normally about 6 inches from the mic and a pop filter. I have not done any processing to it.

I am not sure if it is loud enough or will it need enhancing ? I am going by the height of the waveform being shorter than what I am seeing in tutorial videos. If I drop the gain in order to make the background hum disappear then the waveform gets even shorter.

I am not sure if it is loud enough or will it need enhancing ?

No and yes. It sounds like you’re presenting in a room with a hundred small window fans. Reducing the gain on the Focusrite only appears to hide the noise. The first step in correction is to boost the volume and the noise just comes right back again.

I got the clip to pass ACX AudioBook compliance, but I had to use very aggressive Noise Reduction to do it. It will never pass Human Quality Control because of the noise that’s left. If I bump up to higher reduction, it will start to damage the voice. There are limits to what can be done.

ACX-Check is a tool you can add to Audacity. It tests for the three sound qualities I wrote about here.

ACX Sound Specifications

ACX-Check

Download and Drag Acx-check.ny to the Audacity plugins folder. Restart Audacity. ACX Check appears under Analyze and will tell you if your sample is going to make it past the automated testing robot at ACX AudioBook. It requires 3/4 second or more of quiet to tell you the background noise value. That’s what the two seconds of silence in the forum format does.

ACX AudioBook minimum requirements are also Broadcast values and not a bad place to start if you plan on hiring out. You need to figure out where those noises are coming from. At first pass, I would say your computer fans.

Koz

Unless you plan on investing in expensive, exotic microphones…

…changing the microphone isn’t going to help. You have enough equipment to go with and just need to remove the noise.

Koz

good call Koz. I am so used to using my PC I had forgotten how much noise it makes.

I have done another test using my surface pro 3 which is a lot quieter.

Many thanks for your invaluable help and insight into this.

That’s more like it. We’re left with the gentle fffffffffff rain-in-the-trees hiss in the background and no fan noises.

I Normalized to -3.2, Equalized with LF Rolloff and noise reduction at 9, 6, 6, more gentle correction than earlier, and LF Rolloff isn’t doing as much work.

I had to be careful to avoid that clicking noise you made at about the 1.5 second mark for the Noise Reduction Profile. You really have to freeze and not do stuff during that time.

Just for giggles, run the Focusrite all the way up and do a similar test. Forget the “Focusrite is at 80%” thing. The Focusrite’s job is to boost your voice, not to have a pretty knob. I have several Microphone Preamplifiers that require running all the way up to do their job.

Koz
Screen Shot 2016-07-18 at 14.25.07.png

Koz, I have attached a recording at 100% .

I have also played around with the ACX plugin and managed to get my audio to pass whilst playing with the noise reduction and compressor. I am slowly understanding what each effect effects the waveform. having fun and feeling a bit more in control.
Many many thanks for you guidance in this.

Noise Reduction lives or dies on the profile step. That’s where you let the tool sniff pure noise so it has a good idea what to do. If you allow the slightest bit of voice in the profile, Noise Reduction will try to delete the voice, too. That’s why I made a fuss about making noises during the clip two seconds.

“Normal” correction is the Reduction of the Beast: 6, 6, 6. As the noise gets worse and worse, the first number, reduction, needs to go up. Note your last submission was 9, 6, 6. If you get much over 12, 6, 6, your recording or microphone might not be up to the job.

Note these are not the Audacity default settings and will not work well on music or other production.

This is the generic setup I use.

Noise Reduction
– Drag-select Room Tone, silence or the flat area between spoken phrases.
– Effect > Noise Reduction: Profile
– Select the whole clip or show by clicking just above MUTE.
– Effect > Noise Reduction: Settings 6, 6, 6 > OK



I have a trick with Effect > Compressor. Normalize first so the applied sound is at standard level and then normalize again afterward, so the accidental volume change is invisible. If you do that, then the pure compression effect is the only thing that changes and all you have to do is run the ratio up and down to get what you want. 2:1 is good in most conditions. I had to go o 2.5:1 on one of the messier submissions. By the time you get to 3:1, you should start to inspect your recording conditions. That’s correcting for a seriously sloppy recording.

Audio Compressor
– Select the whole clip or show by clicking just above MUTE.
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2 > OK
– Effect > Compressor: Thresh -20, Floor -40, Ratio 2:1, Attack 0.2, Release 1.0, > OK
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2 > OK

managed to get my audio to pass whilst playing with the noise reduction and compressor.

You can always force a submission to pass. The next thing ACX does is let Human Quality Control listen to it and if it doesn’t sound human, you get “excessive processing” rejection. So no, you can’t submit cellphone voices, talking into a wine glass or recording in a barn.

Koz

When you play that raw clip, please note the bouncing light sound meter tips between -6 and -12 which is the ideal recording volume.

ACX Submission peaks are not to exceed -3 and the Audacity recording peaks should be just about where you have them, so chances are there will always be some correction. Even in my tour de force recordings, I always had to change volume.

You have 50Hz wall power hum leaking into your recording, so we’re stuck with Equalize > LF-Rolloff rumble filter.

Normalize
Equalize LF-Rolloff
Noise Reduction 6, 6, 6
No compressor

I did something a little different. Effect > Amplify: Amplification [1] > OK.

Attach.

You may need to change those settings to be a bit stiffer when you get into this, but I wanted to see what I could get away with. The amplify step is silly. Your RMS (loudness) is just about to fall off the bottom and the other readings had a little elbow room left, so I boosted everything very slightly. This is completely compulsive. You would never do that in production.

Now we’re fooling with the limits of an inexpensive microphone. I think that’s where much of the ffffffffff is coming from.

I don’t think we can do any better.

Koz
Screen Shot 2016-07-18 at 18.08.48.png