Getting basic mic recording setup correct, can someone advise?

[RH22_user]

Hi,
I am actively recording audio for videos as part of my current job and although I have used Audition for quite a few years now I am still barely above novice level in terms of audio production. 

I use a RØDE NT-USB, it's what my employer provided. 
I should note that I am connecting it to a Windows PC via a USB port in my external monitor since the native USB ports on the PC are all USB 3 which seems to cause interference when using this mic. I have no idea if this affects how Windows/Audition sees the mic, I assume not. 

So using the RØDE NT-USB and Audition on Windows I find myself struggling to get a basic setup working for recording narration for videos that is clean and normalised for e.g. YouTube - which I learned only recently should be -14 to -16 LUFs. 

I am attaching an image which illustrates how the mic is setup in windows (I have my headphones connected through the 3.5mm jack on the RØDE NT-USB) and a recording comparison I did of my setup vs what the great Mike Russell showed in the video below
Mike Russell - YouTube - How to Make Your Voice Sound Better in Multitrack (Adobe Audition Tutorial) 

1. I have my mic and speaker settings in Windows at 75 - this sounds the best to me and I have read many posts online that also suggest this. And in the screenshot below I also include a my Audition Audio Hardware settings.
   
   
2. I setup the Multitrack Session as Stereo and record in Mono - this I read seems like a normal thing to do for voice over recordings. (I record my narration audio separately and then add to video in Premiere)
3. I also assume the Multitrack Session > Sample Rate should be 48000 Hz which seems to be the default and is I understand what the RØDE NT-USB supports.
   
   
4. In Mike's video in e.g. Track 1 the audio output is Master where for me it is Mix. I don't know what Master refers to nor where to find it. I am guessing this relates to Mike using an audio amplifier/mixing desk?
   
   
5. I did a test recording using the first Mono selection similar to how Mike did it. 
   
   
6.  When Mike records the wavefrom is tiny when compared to mine - does this indicate that something is wrong? I am including the Amplitude Statistics for that recording in the screenshot below.
   Note: Mike appears to be using a "Signature 12 MTK" in the video which I guess is somehow amplifying the audio signal into Audition or something like that - again total novice here. 
   
   
7.  I used the control in the lower left corner of the track to slide the audio level to make it louder, and when it's at ~21.5 dB my waveforms are huge and the audio is very distorted. For Mike this is not the case. 
   
   
8.  I then also did undo on the previous action and used the Match Loudness as Mike did for -16 LUFS and where again my results are distorted. 

Is there something obvious I am doing wrong here or is this just how it is when one only has access to a USB microphone?

[RH22_user]

As an edit to the orginal post above, I should also add that when recording I normally get clicks, buzzes and distortions included in my recordings which I need to then go through and clean up by silencing, auto healing or completely rerecording. Some of the sounds are the usual mouth noises that come with recording, the distortions I don't understand. I have always assumed it was an issue with audio processing from the microphone, through the computer hardware/OS and into Audition. Needless to say it makes creating a clean recording a lot longer than I would like. 

[SteveG_AudioMasters_]

It's not worth messing about with the size of the waveform display - that will just mislead you. It's normal when making a recording to set the mic gain so that there's enough 'headroom' to take account of anything that gets slightly louder than you intended - these days between 12-15dB lower than the 0dB peak. To set the record gain with this mic involves using the Windows Sound Control panel. Under the Recording tab you select the mic, which should be listed there, then Properties and then Levels, where there is a slider to adjust. Yes that's a right pain to get at, but you shouldn't need to do it very often, as it should remember the settings you finally make. The thing to do is to adjust the levels using the 'monitor input levels' facility in Audition (right-click with the mouse held over the meters and you'll get that) and set it so that your recorded peaks are around -12 to -15dB, as noted above.

Now your recorded level should appear to be more healthy. As for all the rest, we'd need to hear a sample recording to see what you are up against, but you have to get the levels right first, and ideally sort out what the interference, etc is. All this business about LUFS and everything else comes after that.

[RH22_user]

Hi Steve and thanks for the reply!
I have a shortcut for the Windows Sound Control panel on the desktop as I need to open it regularly. I find that Windows will sometimes adjust the levels, I'm not sure why, so I always need to check it before using a microphone. 

Sorry I don't fully understand the concept of setting the gain headroom to allow louder audio between 12-15dB lower than the 0dB peak. I usually set the gain so that the audio sounds loud enough and does not have so much gain that the audio sounds distorted and that is pretty much all I know to be honest. And in terms of setting the levels in Windows sould I be using percentage (e.g. 75) or decibels? See screenshot below.

And also I don't understand where in the Audition UX the monitor input levels facility you refer to is located.
The thing to do is to adjust the levels using the 'monitor input levels' facility in Audition (right-click with the mouse held over the meters and you'll get that) and set it so that your recorded peaks are around -12 to -15dB, as noted above.

If I enable recording (R) and input monitoring (I) then the Levels panel below becomes active but I don't like using the input monitoring as there is always a delay between the input and output. Is this what you are referring to, as shown in the screenshot below? And how do I set it so that the recorded peaks are around -12 to -15dB?

[SteveG_AudioMasters_]

Yes, use the I (Input) button - they disabled the right-click option, and I forgot... It's no good setting levels so that it 'sounds loud enough' - that means nothing. It's the level that is fed to Audition from the mic that is important, and you adjust the gain so that this level peaks at somewhere between -12 and -15dB on the meter. You don't have to keep monitoring it, you just set it for your level at the time, and Audition will then be recording a signal that isn't disappearing into system noise and interference, and with sufficient level to show in multitrack. Setting that input level is important. You've got to get this right first before you can sort out everything else. The record input level is the only one you care about - all the others are just listening levels, and don't relate to your recording at all.

[RH22_user]

Hi, okay so I should enable the I (Input) button and then use the Levels panel adjusting the mic gain in Windows to check that the mic input is between -12 and -15dB correct? And once the gain and levels are correct I can disable the I (Input) button. I would keep it active but I don't like the input/output delay. 

And basic questions, should the green or orange bar be reaching between -12 and -15dB in the Levels panel and what is the yellow line that appears in the Levels panel indicating? Actually that line appears in all the level meters in the UX it seems. 

[SteveG_AudioMasters_]

By default, the transition from green to orange happens as -18.1dB. That's the default, but you can alter this in Preferences>Playback and Recording. There's something to be said for setting that transition to -15dB and then, when you are speaking normally, you should only get the occasional flash of orange. Once you've set the level so that this is happening, you're good to go, at least with the basic recording.

Incidentally, the right-click monitoring option still works in Waveform view, but don't record in waveform view unless you are desperate; there are other things that can go wrong if you do, because it's not direct to disk, like Multitrack recording.

[RH22_user]

Thanks Steve!
Ok, I have set the Preferences > Playback and Recording > Meter Color Crossover Levels > Yellow caution level to -15dB. I was not aware of that option. I will create some test recordings using that. 

Sorry I checked right-clicking in Waveform view but I don't seem to find this 'monitor input levels' facility you refer to. Does recording (R) and input minotoring (I) have to be enabled in Mulitrack view for it to be active or visible? I'm guessing I'm just totally missing something 🙂

[SteveG_AudioMasters_]

The mouse pointer has to be on the meters before that rightclick option appears. Right-clicking is actually a way of doing a lot of things faster in Audition - the options offered vary according to where you do it.

[RH22_user]

Hi Steve, sorry the moust pointer has to be on which meters exactly and you mean in the Waveform view correct? Could you possibly illustrate what you mean with a screenshot, in particular for this 'monitor input levels' facility. 

[SteveG_AudioMasters_]

[RH22_user]

Hi Steve,
Thanks for that, I see it now. 
In Multitrack mode Meter Input Signal is greyed out and to use Levels panel one must enable input monitoring (I) which is not preferable due to the delay effect. 
So Meter Input Signal is only available in Waveform mode however you mentioned that one should not record there as things can go wrong.
So is the idea that it's available there so one can monitor the levels while checking the waveforms of a recording and not for when actually recording?

[SteveG_AudioMasters_]

As far as I'm aware there's no difference. Some latency is inevitable, simply because of the way the processing has to work. As far as why recording in Waveform view is there - that's a whole different story.  At one stage early on in Audition's history (when it was Cool Edit), recording in Waveform view was the only option - multitrack direct to disk recording simply wasn't there at all. So it's there for historical reasons, mainly - and for very quick fixes to existing files.

If you've set the level correctly for a recording, then there's absolutely no reason to need to keep monitoring it whilst making the recording itself. If you've left sufficient headroom then you'll get a clean recording, using all the dynamic range available.

[RH22_user]

Ok, thanks again Steve!
Actually yes I use the Waveform view for quick edits to recordings when somehthing sounds so bad I need to replace it. And I didn't know Audition came from something called Cool Edit, that's interesting 😉

Sorry for more screenshots but I suspect I am still doing something wrong. I have set things as suggested and when I do a quick test and then do what Mike Russell did in his video, just as an example, apply Match Loudness in Multitrack to -16 LUFS and playback the file the levels are going into the red. I suspect that if the gain and levels are correct I don't need to run Match Loudness at all, I'm not sure at all. 

[SteveG_AudioMasters_]

We haven't finished with the basics yet. The next concept you have to get used to is Normalising. This is something that can only be done in Waveform view, because it affects your file directly. What it does is to place the peak value of your file at a fixed amplitude - everything else will follow automatically. In other words, it just scales the file.

Why is it important? Because every single amplitude-related effect in Audition is designed to work around signals normalized to 0dB. So before you try compressing, or relating files to LUFS settings or anything else like that, you have to normalize your file. The headroom in the file was important whilst recording it, but not any more when you process it - this is when you get it to its optimum amplitude for processing.

As far as setting loudness is concerned, that's a late stage in the process - you need to get the file sounding correct first. With a lot of voiceover files, a first step to doing this would be to limit the peak values - effectively squashing them. You may think that this will make a large difference to the sound, but generally it doesn't. What it does do though is restrict the dynamic range. If you cut off the top 6dB of all of your peaks, you will gain that 6dB of unused space just below 0dB, and if you normalize the file again, you'll find that everything seems louder, and probably more 'even'. Ultimately, once you've got your file sounding the way you want it to, then you can have a look at the LUFS value - doesn't make sense before that.

[SteveG_AudioMasters_]

It's also worth noting that with some processes, you have to renormalize after applying them, simply because they've cut off peaks. With speech you can get around this. If you use an effect like the Hard Limiter, this can sort all the levels out for you. - a look at the sliders will give you a good clue as to what's happening. A good starting point is the 'Light' preset, but you will probably wish to alter this for your specific needs - it's only a starting point, like all effect presets.

[RH22_user]

Hi Steve,
Firstly many thanks for all the help here as I'm a lot further on than I was when I started looking into all this.

My main goal right now is to get the raw voiceover recordings loud enough for YouTube. See below where based on this setup and just taking the raw recording and uploading it to YouTube it is 6dB lower than it should be. 

Last night I was experimenting with making copies of my original cleaned up recordings, then in Waveform view increasing the loudness by +5dB then saving, adding to Premiere, adjusting mix there and then uploading small clips to YouTube to test what the loudness would be.

Right now I have two video projects I need to place online so I need to try and look into this further once those are done. My main problem right now is I find difficulty getting my audio loud enough so that when uploaded to YouTube no audio compression is needed. The "stats for nerds" view from YouTube in the lower right of the screenshot below is what the audio in an upload video should look like. 

[SteveG_AudioMasters_]

You will only achieve that by compressing the audio so that the quieter parts are closer in level to the louder parts - that's what compression and limiting does. Youtube doesn't appear to do too much to the audio you place on it - I do this regularly with music that has a wide dynamic range, and the dynamics pretty much remain - so the loud bits are loud, and the quieter bits remain that way. But for your speech, that's not what you want at all - you want it like it would be on the radio, or a 'normal' program, at any rate.

So for your speech to appear to be louder, you have to compress it and make sure that you renormalize it to 0dB afterwards - you need rather more of your audio to be at a higher level. One other thing you could try is the Speech Volume Leveler - but if you use this, definitely re-normalize the results afterwards. And try different settings, almost certainly with a check in the 'Boost Low Signals' box.

[RH22_user]

Hi Steve,
and thanks again. At this point I could do with some kind of document/video outlining this process so I can try and recreate it myself. As I indicated at the start I'm barely above novice and while I can follow the concepts of what you describe I would need to see the process applied in the software in practice.

(I create software tutorials videos and my sort of golden rule is, and what I always say to people is, never assume the user knows what you do 🙂 )

[RH22_user]

Hi Steve,
I was wondering would it be at all possible for you to compile all your suggestions into list of steps acting as a workflow of sorts that I could try and follow? I have the concepts of recording, checking levels and gain working in practice and it's now the Normalising process for voice recordings that I would need to become more familiar with. 

Or if there is a good video or document/help guide outlining the process that you could recommend that would also help 🙂