Setting Input Level for Better Dialogue Audio

[@DaveDugdaleColorado]

Thanks for sharing. You made me aware of a couple of effects I haven't tried yet.

[@curtisjudd]

Thanks Dave. My tiny way of paying you back for all the help I've received from your videos over the years!

[@ocubex]

Another great tutorial, 3 questions... 1. I was told lowering your gain makes the voice print close to the noise floor meaning when you raise it in post you will also raise the noise level. Is this true or does it not matter at all. 2. Whats the different between adding or setting gain to -3db and normalising to -3db? 3. Do you fix your audio before editing in Premiere or do you send a Premiere timeline to audition?

[@liamhardy2875]

Thank you Curtis, I'm just entering the wonderful world of audio recording and am finding your videos very helpful. Cheers, Liam

[@curtisjudd]

+Liam Hardy thanks Liam!

[@MyJeanf]

Thanks Curtis , I have been researching your video's no that I have finished filming my Documentary. Your video's have been invaluable. Great advice as always.

[@curtisjudd]

Thanks! 🎤

[@Realunitedstates]

Curtis Judd Extremely helpful Curtis! I still have a long way to go, but you are helping us immeasurably. Will be giving you a special mention and thanks (shout-out) in our Thursday episode of Behind The Scenes. And yes, I will be using the new AT899 and Tascam DR100-MKII. :-) Thanks! Paul

[@curtisjudd]

Thanks Paul--can't wait to hear how it turns out!

[@Realunitedstates]

Curtis Judd Well, it isn't perfect yet, but it is markedly better! I can't thank you enough. I am still learning and practicing.

[@kbohlson]

Hi Curtis - great videos. I've had a Tascam DR100Mk2 for a few years. What's been a great help is getting an XLR Y-adapter - single mic into both inputs. I set one channel several dbs's lower than the other - a safety track. Saved me many times. The adapters are inexpensive.

[@curtisjudd]

+Kevin Olson Great idea for a recorder which doesn't have a limiter or safety track feature. Thanks for that!

[@cavejelly]

Your comment regarding gain helped immensely. Thank you.

[@curtisjudd]

Thanks!

[@KTHKUHNKK]

Excellent stuff Curtis especially the part about people with their short attention spans Keith Kuhn

[@curtisjudd]

Thanks Keith

[@KTHKUHNKK]

@@curtisjudd You got it Curtis

[@MrMichael33b]

Nice info, I have the akg c414, I used in my music days, can I implement it into videography outside of studio voiceover work?

[@curtisjudd]

It does have a hyper-cardioid pickup setting which may work well for booming just outside of the top of the frame--would be interesting to see how well that works.

[@richardh568]

Curtis: Thanks for another great video! Question.....In the video, you mention that you used a small hand held recorder for this audio. If you were using your Sound Devices mixer/recorder for this, would it change your input levels any? Since the professional decks have better mic pre's and analog limiting, would you still record at -20 to -18 db? Or would you feel comfortable bumping that up a little to say -18 to -12 db? Thanks again......William

[@curtisjudd]

Hi Richard, That's a really great question, People's voices can be extraordinarily dynamic. In a shoot a couple of weeks back, the talent had one of the most dynamic voices I have recorded and I ended up having to gain down after the first take because he was engaging the limiter quite a lot. I would actually feel more comfortable gaining down with higher end recorders like the 633 because of its cleaner preamps. This gives more room in post for compressing the transient peaks and normalizing without introducing too much noise. It is the less expensive consumer grade recorders where I tend to push the input a little more aggressively with peaks closer to -12 because I know that I won't have as much latitude in post. And if things get too loud during the recording, I have to make the call on whether to do a retake or take my chances at de-clipping the audio in post. I hope that makes sense and thanks for the question!

[@richardh568]

Curtis Judd Thanks. Great info.

[@davidstrathmore117]

I just discovered your channel and it's a gem! Quick question: I just purchased Sound Devices Mixpre10t and Audix SCX1HC, I love the sound but I'm still confused about gain vs fader volume. Does it make sense to keep the gain imput as low as possible and just raise the fader volume so that I don't raise the noise floor? I did a bunch of tests and I was getting clean sound at imput gain 50db and fader turned up to +10db. My peaks were at -12dB. Any thoughts?

[@curtisjudd]

Hi David, thanks! Normally you’ll want to do the gain staging with the gain setting before you start the recording and then use the fader to fine-tune how much of the audio signal is sent to the mix tracks during the recording. In that light, I usually prefer to start the recording with the fader at 0 so that you have room to move up or down during the recording. Good luck!

[@davidstrathmore117]

Curtis Judd Thank you for answering my question sir!

[@ChanonWangtrirat]

Thank you for your tricks. I usually peak about -3dB normal sound will at -6dB. Because if I crank up to nearly 0dB, Mostly in Cinema Theatre sound system won't adjusting volume to suit the video or film which playing and loud or clipping because they have calibrated all system to optimum. So normal sound at -6dB should be safe for all equipment. Something weird when mix the sound. Some speaker tend to peak on mid-range especially small or cheap one. I normally reduced mid little bit before bounced them out.

[@curtisjudd]

Good ideas--thanks Chanon!

[@123andjump]

thank you so much, the most helpful video I have heard and seen so far. Greetings from Denmark

[@curtisjudd]

Thanks Michael!

[@TheHotshizzle25]

So isn't it better to immediately record a louder signal with the limiter on in the recorder and then there is less editing? Better to record a quieter signal and then amplify it in post-production? I do not know if the noise signal is not greater in the second case after all? I am also interested in how to set it when there is a large amplitude of levels - that is, some sounds are much louder, and others are quiet - on the same recording. it is not always possible to record each instrument separately, sometimes there is no time for it or the recording is very dynamic ... so what then? In this case, I set the gain around 6.5-7 / 10 on the Zoom H6 and uses a limiter there, which prevents clipping. Perhaps I am doing it wrong? After your movie, I started to wonder ...

[@curtisjudd]

Yes, if you have a good limiter and the signal is just hitting the limiter periodically and not constantly. The H6 limiter is pure digital and doesn't prevent distortion so I'd be hesitant to rely too much on it if you want a clean, distortion-free recording.

[@cristianroman4128]

Hi Curtis. Nice tutorials, I like it when you deliver the information without spending too much time talking. So, I have a question here: if you aim for -12 to -16db when recording and then you normalize it to a higher gain. Won't that increase the noise in your recording? I'm learning audio recording for filmmaking and I thought it would be hard to cancel noise in an outdoor shooting situation, where your mic will also record ambiental sound like cars, foot steps, etc. Thank you, Cristian

[@curtisjudd]

+Cristian Roman hi Cristian, yes, you are right. The trick is that on the flip side, you'll get clipping distortion if you run the gain too high and the talent suddenly laughs or speaks louder. So it is a compromise. Fortunately the newer recorders are getting cleaner in terms of self noise so it is not nearly as much of an issue as it was in the past.

[@cristianroman4128]

Curtis Judd thank you for your reply. I bought myself 2 days ago the Zoom H1 recorder and it seems it has a lot of noise, especially when I'm using an input of 70-80. Maybe is something wrong with it?

[@curtisjudd]

That depends. Is it self noise from the recorder or ambient noise from the space in which you're recording? It should not produce a lot of self-noise which often sounds like hissing.

[@cristianroman4128]

It sounds like hissing to me, but I'm not sure of it. Is there a way to test if it's ambient space or self noise?

[@curtisjudd]

Start recording with the H1 and put it on a bed with about 20cm of blankets on top of it in a quiet room. Let it run for 30 seconds. If that recording still gets lots of hiss, then there's probably something wrong with the recorder.

[@DCifuentesJ]

Would you recommend removing hiss before or after normalizing? Would love to see a video about that

[@curtisjudd]

Either is fine as normalization is just a mathematical addition of amplitude.

[@tanzexun]

Helpful tips, Curtis. Thanks!

[@smalltalk.productions9977]

still relevant! thanks for the effort and the sharing. thumbs up.

[@curtisjudd]

Thanks!

[@BobPenzien]

Hi Curtis, I plan to shoot Full HD video for KZbin using a Panasonic GH3 camera. Also, I plan to record audio separately using the Zoom H1 and then sync in post using Adobe Premiere Pro CC. I would like to produce the highest quality video and audio that KZbin can handle. Therefore, should I use MOV 1920x1080, 30p with 72Mbps All-INTRA for my video format or something less? The camera also can shoot in other MOV formats, MP4 and AVCHD. I am concerned that the video file size might be too big file for KZbin to handle. For best sound from the Zoom H1, should I record in WAV at 48 KHZ 24 bps? I look forward to your recommendation. I am a real estate broker that you would like put out better than average KZbin videos for marketing my business? Thanks in advance! Bob

[@curtisjudd]

Hi Robert, 1080p30 at 72 Mbps should be great--that's a very good, high quality capture format. After you are done editing and doing any color correction or grading in Premiere, you will export using the KZbin 1080p30 preset which will optimize your final video for KZbin. The idea is to capture your footage using the highest quality you can and then after editing and grading, export to the lower bit rate format that is optimized for KZbin. You've got the right idea! Same for sound--yes, record in 48KHz 24 bit, sync that to your footage in Premiere, and then on when you export using that KZbin preset, you'll be all set. I've got an episode that covers the details of exporting from Premiere over at kzbin.info/www/bejne/oGPNZpekoamgpdU Good luck and let me know if you run into any issues!

[@BobPenzien]

Thank you very much Curtis for your advice!

[@GeorgeGraves]

The *worst* is soft/loud talkers. "oh hi, I'm here to SAY SOMETHING AMAZING!!!!: Those guys drive me nuts. But if you learn you equipment, it's not a big deal.

[@johnmellor932]

Thats why 32 bit float is so great. You can't even ride the gain in advanced mode on Sound Devices mox pre recorders.

[@joshua3113]

Do you have any videos on bag mixing and what type of mixes editors or producers would want to see? Trying to find info on the subject but it's proving elusive.

[@curtisjudd]

Nothing specific to bag mixing, but we will cover things soon over on my other channel: kzbin.info/door/A5_6sZEVufCP2AlIzuYCHQ I will also have an episode coming up on mixing in general with a review of the Zoom F-control mix surface.

[@joshua3113]

Cool, look forward to it.

[@curtisjudd]

Here's the piece on the Zoom F-Control mixing surface for Zoom F4 and F8.

[@rufinorosado]

Awesome video Cutis. Thanks a lot. This was very helpfull.

[@RKRK-ci2vv]

Hai Curtis ... Now I am going to use Gaint Squid Audio Lab lav ( New version ) and Zoom H 1 . Which of the fallowing setting is the best for a better audio ? * 44/16 - Input level (peak ) 37 * 48/24 - Input level ( peak ) 65 Thank you

[@curtisjudd]

+rk theinspirer for video online? I usually do 48/24 and set the input level between 37 and 90 depending on how loud the persons voice is...

[@RKRK-ci2vv]

Thanks a lot Curtis Namaskar

[@AWPPrivate]

I know this is an old video, but If your main speakers were in front of all the microphones then you will be able to minimize feedback in a live recording situation. Also the amps for the base and guitars should be more in the middle behind all the mic's. and behind the performers. Also if you set your recorder so you just barely see the noise floor on your levels, then you will kill a ton of noise. You should also be able to record a solid voice even if it doesn't reach 12db, because you didn't record as much noise. So when you volume up on Audition there is less noise to start with.

[@curtisjudd]

Thanks for the tips!

[@Abbey9000]

FYI Folks: Snagit, is better than Smart Screen Capture, by far. But I am still uncertain what software you are using for the Audio adjustments. I am having difficulty with lavalier microphones using the Tascam DR60-D......

[@curtisjudd]

I use Adobe Audition but any digital audio workstation app should work.

[@curtisjudd]

+JennyFTOL I use Adobe Audition for editing audio. On the open source side, Audacity. It is not as easy to work with, to be honest, but it is free. Just recently, Avid has also made a free version of ProTools available called ProTools First. The learning curve is a little steeper on that. On the less expensive commercial side, there is Reaper which I have used quite a lot in the past.

[@snouuman]

Funny story. My boss went over my last dialogue recording and said I'm doing it wrong and that -18dB is way way way too low (This is how I've always recorded in order to avoid clipping). So he suggested I come to this channel to see how it's done. He basically just proved himself wrong lol. I'm going to send him this video and hopefully he doesn't see my comment.

[@curtisjudd]

It depends on a few factors. If you're working with a pretty weak preamp and with just one input, -18 dBFS is probably not ideal. In that case you'll probably want to push closer to -12dBFS. But if you've got multiple inputs (e.g., interviews), that 18dB of headroom is often pretty important, especially when people get excited and talk over each other. So there isn't a hard and fast rule. You and your boss are both right. 😉

[@ryanw7044]

I watched your video on best input setting for the Zoom H1. I tried 37 and my talking voice is at -24db or less on the zoom. If I crank it up to input 80 on the zoom I can get -12db if I talk a bit louder. It seems the lower my original recording is, the worse it sounds when I normalize it to -6db like in your other video. In this one you normalize to -3db. So when I set the zoom h1 to 20 input and my mic to +20db talking from the same distance I get between -12 to 15db but it sounds louder and fuller when normalized. I'm just trying to use this for audio when vlogging so that I sound good but also anything I record sounds good as well when the camera is not facing towards me without having to constantly adjust the input level. I guess what I'm trying to say out of all of that is that the lower the recorded level I get, like -24db showing on the soom h1 screen, the crappier it sounds when I normalize it afterward. Maybe I should just stick to using auto levels! Cheers.

[@curtisjudd]

Hey Ryan, thanks for the info. This was a bit more of an academic test. If your settings sound good, then they are good. Best wishes!

[@ryanw7044]

My settings never sound good. I can either set them really high on the mic and low on the zoom (or high on the zoom and lower on the mic) and risk everything that isn't my voice being too loud, or set everything really low and normalize it later and it sounds bad. Maybe I should have gotten a lavalier mic instead. But at least either way sounds better than when it is plugged directly into the camera. All you get then is HISSSSS! Cheers.

[@curtisjudd]

Are you using a shotgun mic like the RODE Video Mic Pro? If so, do you have it on top of your camera or boomed above your head? You can boom it and dramatically improve the sound like this: kzbin.info/www/bejne/ioaonoR4j7CieNU

[@ryanw7044]

Yeah, it is a video mic pro. It's on top of the camera, but it's on an action camera grip so the mic is 6-8 inches above the camera. It's for vlogging and traveling so there is no permanent option for boom (though I do have it for at home, and use it!) Both point down towards me slightly. Only about 1 foot from my face. In order to get register -12db or so I almost have to stick the mic into my mouth unless I put it on +20 db. But I also think I just have to watch more videos and test different things out. I am getting better results now with a really low recorded level, even if it is -24 and normalizing it to -6db. Cheers.

[@DerekRoff]

Thanks for a useful and interesting video.

[@sosexv1]

Another helpful video. thank you

[@klajdi9999]

I think that with our cheep and noisy recorder's.....-12 to - 18 is not a good idea. If we record near to 0 db the difference to dialog and floor noise will be better. For this reason I have bought a Sound devices mix pre-D. Thank you very much for you youtube channel and your contribution! Greeting to you!

[@BruceAvilla]

That Sound Devices Mix Pre-D is $899. A bit steep, eh??

[@klajdi9999]

Bruce Avilla Yes! Very hi price for a 2 channel mixer, but I work for indy film. What can I do without Sound devices?! Greetings

[@curtisjudd]

You make a good point. There are two assumptions I didn't mention which are pretty critical: 1) that you are using a clean preamp (which many of us are not--not sure whether a Zoom H1 qualifies and most DSLRs probably do not quality) and 2) you are recording in 24 bit (which most cameras are not) so that you have enough "resolution" to process it in post. It may be that my issue wasn't that I was recording too hot but that I was just mastering too hot. It seems like in my most recent experiments that mastering so that the final peaks are at around -3db works best, at least when it comes to most online platforms. What is your preference on that?

[@klajdi9999]

Curtis Judd I use a zoom H6 and if I record at input level 7 is very noisy. Zoom is noisy recorder compared to professional recorders. Anyway I have find the solution! I go out from sound devices mixer with XLR and input to the zoom H6 with mono jack 1/4 (unbalanced) This way can help me to record at max input level 3 and about near of zero. Work perfect! I work in post production too but if the audio file is noisy, when I compress and normalize the level of noise floor grow up. Excuse me for my english!

[@BrianMarcWhittaker]

What about sample rate and bit depth? When exporting from Avid Media Composer or Premiere I've read to send 48 KHz with a bit depth of 16. Would I record at 96 KHz & bit depth of 24?

[@curtisjudd]

Yes, much like on the video side where we typically try to capture higher quality. However, for dialogue recordings, I usually record at 48kHz, 24 bit unless I'm recording sound effects that I will be processing heavily (or slowing down and resampling) in which case I will record at 96 or 192kHz (always at 24 bit).

[@ocubex]

-18 to -15, isnt that too low? I thought -6 was the target for peaking?

[@123andjump]

I have heard -18 to -12 is good

[@protoman247]

boost in post. setting gain to peak at -18 to -15dbfs gives you a good signal to noise ratio and allows plenty of headroom during the recording process.

[@crawlinginfilm9683]

I've been stressing over such matters, even before coming across Curtis's great videos (including this one), despite having worked for a couple of decades by 6-12 dB headroom as a rule of thumb, occasionally 18 dB (only in uncertain or nerve-rackingly important situations). But it's always worried me. Either I set 6dB and sweated over possible clipping (e.g. repeatedly revisiting the meter while operating a video camera), or I set bigger headrooms and still sweated, this time over possible quality loss. So, having some free (ish) time, I recently did some subject research (googled a lot and kicked a few calculable tyres). For a linear audio recording (e.g. as in a 16-bit WAV-PCM file), each successively greater bit represents a correspondingly successive 6dB increase (in electrical hence audio power) over the previous (lower) bit. Optional Explanation of the above: Each successive bit corresponds to an equally successive doubling of (analog) voltage (from an analog-digital convertor), i.e. a 3dB change in that voltage. However it is the dB change in -power- that we're interested in, and that is proportional to the square of voltage. Given that dB are logarithmic, squaring becomes multiplying by 2. So each successive bit corresponds to a 2*3 = 6dB increase in the amount of power it represents. Therefore with 18dB headroom we lose the use of 18/6 = 3 bits. Consequently for a 16-bit recording only 16-3 = 13 bits are left to represent the expected biggest waveform (the one you're setting up against during sound-check etc.). Of course that could change if (feared/anticipated) unexpectedly loud stuff actually happened, or if there were unexpectedly persistent low-level periods. Regardless (e.g. assuming the sound-check levels turned out to be valid) a 13-bit representation implies 2^13 = 8192 quantization levels hence a (power) signal/noise ratio (SNR) of 20*log10(8192) = 78 dB. I understand that figure is in the general range of the best tape recorders of analogue era. Not as bad as I had imagined/projected then... 24-bit offers the same 6dB/bit but there are more bits ! So no headroom risk/cost to fear - bits to burn !. Maybe I should be doing that from now on. My Zoom H6 can record that format and file sizes are no longer the issue they once were. Wouldn't expect any noticeable improvement in quality, as compared to those 13 bits, but "the more insurance the better" i.e. less sweating. I'd be really interested in what Curtis or others know or have experienced in this area. Or indeed any mistakes in the above reasoning, theoretical or practical.

[@TheOnlyExxxception]

@@crawlinginfilm9683 I've found myself stressing incredibly over soft-loud speakers because of this! Not sure about all your calculations but this is interesting to note. I used to record to give myself more headroom in case there was peaking, but then my sound design professor told me that I should be recording two versions if dialogue varies - one at the regular -6 to -12dB and then another take at a lower volume to account for possible peaking. But that just seems like a waste to me and may miss out on great cinematic takes. I still constantly struggle with dialogue that goes from soft to loud, or actors who want to vary widely in their volume. I'm actually looking for tutorials right now for this specific type of varying volume dialogue.

[@martinremixful]

So I will get better and cleaner audio with less hiss if I record at for example -16db and turn it up in post instead of recording at -10 and not be turning it up in post?

[@curtisjudd]

It depends, but generally, it is important to leave enough headroom for when the talent get loud so that you don't clip. Then in post you can compress the loudest peaks and increase the amplitude of the entire audio clip.

[@rebth1st]

Adobe Audition and not Adobe Premiere Pro?

[@curtisjudd]

Same applies.

[@commercialand]

I am ready to export my feature movie of 2gigs, and there are ghost audio sounds, most likely glitches from my system being too slow. Should I export at a higher resolution? I heard something about apple 4444 or something like that for export. Not sure it will matter, have you ever had this issue? I know its not my audio because I can play it over and over in the timeline and somethings its there and sometimes it's not.

[@curtisjudd]

I would export in the format you need to deliver (e.g., if delivering for a film festival, they will specify this in the technical requirements for submission). Then check to see if the glitches still exist in that. If so, you'll want to contact the company which provides the software you're using. Best wishes!

[@commercialand]

@@curtisjudd thank u and I meant to say 2 TB...Lol Should I try removing the Prefs? Not sure if I will lose session data? I recently switched from 2tb to a 4tb but still very slow

[@curtisjudd]

@@commercialand I'm not sure. Are you using Adobe Audition and Premiere? If so, probably best to contact Adobe. Best wishes!

[@commercialand]

@@curtisjudd final cut

[@curtisjudd]

Ok, I’m not sure what deleting preferences keeps. Might be best to contact Apple. Sorry I couldn’t help more.

[@anantsinghal9489]

Hi I recorded some dialogues for a film directly through the h1 at input level 25..I can hear the dialogues faintly in the recorder speaker..its saved in wav format..but when I transfer it to my pc, all I hear is distortion or just wind..please help!

[@curtisjudd]

Hi Anant, how close was the H1 to the person speaking? Also, you'll probably want to set the input level higher than 25 in most cases. If you place the H1 within about 30cm of the person's mount, somewhere in the 40 to 50 range is probably better. Good luck!

[@MEERDERIDEEN]

Nice video, thanks!

[@BobPenzien]

Nice Video Curtis. I was wondering what screen capturing software you are using to show the Audacity footage in this video? Is it Camtasia Studio or something else?

[@curtisjudd]

Thanks Robert. I used Smart Screen Capture, which I believe is free in the Mac App Store. It is pretty basic and I have to add the zooms in Premiere but you can't beat the price!

[@evillepin]

Thanks! Its helpful!

[@curtisjudd]

👍

[@Abbey9000]

Help !

[@mohamadavakian]

your vocal were cut 4 times but thanks for tutorial

[@curtisjudd]

You are absolutely right. I learned with more testing that I have to limit my peaks to -6 db to prevent clipping.

[@curtisjudd]

@Real Fast English Old video, true peak limiters are what is needed now. If you're using Adobe Audition or Premiere, we have some newer videos that are more relevant today: kzbin.info/www/bejne/hXy2iIOoiZqoptE