Dialogue Audio Compression
Вставка
- Опубліковано 11 бер 2015
- How can you get clear, clean, present dialogue audio for your film or video projects? There are several things you can do but one of the fundamental things is to use a compressor. An audio compressor levels out your audio so that the loudest parts are not too much louder than the softer parts. Once you have compressed your dialogue audio, you can then apply a variety of other effects and at the end, loudness normalize it so that it is loud enough for playback for your audience.
Visit us at learnlightandsound.com for more updates on how to improve your lighting and sound for video.
Also be sure to subscribe to get weekly episodes!
Gear in this episode:
-----------------------------
Shure SM-58 Cardioid Dynamic Microphone (used for voice over)
Amazon: geni.us/WS1cBF
B&H Photo: www.bhphotovideo.com/c/product...
-----------------------------
Focusrite Saffire Pro 24 DSP Audio Interface
Amazon: geni.us/PFVZph
B&H Photo: www.bhphotovideo.com/c/product...
-----------------------------
Music copyright Curtis Judd 2012
Ethics statement: Some of the links above are Amazon.com, B&H Photo, or other affiliate links.
This is a great exemple that demonstrates how several instances of light compression (done serially) yields a far more transparent result than a single compressor doing all the heavy lifting in one shot.
Thanks Daniel.
Thanks Curtis! Really appreciate you taking the time to make this. Will make a big improvement on my videos :)
Thanks for the feedback Taylor!
Compression very nicely detailed and made understandable. Thank you Curtis.
***** Thanks BF! (that's not the same as BFF, but almost, except one F)
Finally able to grasp the concept of amplitude and frequency. Thanks for this. Continue the good work.
👍 glad it helped!
Took me a couple of tries before I actually finally understood it all, but I do now! This is soooo helpful. Thanks, Curtis!
phatcorns Glad it helped!
Very insightful as always. Thank you, Curtis!
VictorAlarconFilm Thanks Victor.
EXCELLENT tutorial! Even I can understand compression now! Thank you!
Dan D. Dirges Thanks!
great video Curtis!! We started using the NT1A to record the audio of our classes (no camera mic anymore!!! :D) and I'll try to use this tips on our next music class.
***** Thanks and best of luck
Just love your content mate, it's really helpful for a beginning vBlogger who had no idea where to start with what he had in mind :D
Also as a music producer, I loved how you explained compression here ;)
Kurt's Journey Thanks Kurt!
Great tips Curtis. Thank you!
Joe Baker Thanks.
Great tutorial. Thanks for sharing! However, I had your video full screen and I freaked out when I looked at the clock haha.
@CurtisJudd my podcasting audio sounds much better because of this tip! Thank you!!
👍
Great info! I was compressing after normalization (my video editor has an option to normalize audio after import), but this workflow makes way more sense.
Vegan Ostomy Glad it helped and good to hear from you again!
Curtis Judd You're my "go to guy" for audio/video! I just wish I had a more controlled environment to work in when filming so I could get more consistent (and predictable) results - ran into an unexpected challenge where I could hear the birds in my daughters room coming through VERY LOUDLY on the audio track of a video I had finished recording, and it was so bad that that I almost had to scrap the video all together. Fortunately, I found out about "spectral audio editing", which resulted in nearly 8 hours of fiddling with the trial version of Adobe Audition (which I've never used) to remove those chirps from the audio track! The results were far less chirping, but slightly lower (but acceptable) audio quality - I must say that it was quite the adventure! Always learning new things! haha It also convinced me to save up for a future subscription to get Adobe Audition, which is way more powerful than I expected.
Vegan Ostomy Good for you, venturing into new territory with spectral audio editing!
Awesome
👍
BEST TUTORIAL EVA
Thx
Ahhhh. The dreaded LOUD and *soft* talker. This is just my preference, but if I'm going to compress a track, I do the whole track or segment of dialog. In a pinch you do just part, and can throw in a cross-fade to help smooth it out. I never liked loudness filters - too crude of a tool at times. IMHO a graphic compressor really gives you a lot of flexibility with controlling the curves. Love the videos - keep it up!
GeorgeGraves Great thoughts, thanks George.
Love your videos! I'm considering buying a mixer like the Yamaha MG12 to be able to hook up two or more mics (fx Sennheiser EW-100 and Rode NTG3) and do levels and compression on that. What are your thoughts on that?
Any alternative recommendations?
Flemming Eiberg Thanks! I haven't done much recording on mixers that are designed for live sound reinforcement but of course it will work. If you've already got it, go for it! If not, recording with a field recorder or audio interface and then applying the processing in post is how I usually work. When shooting live action, I record with the Tascam DR-60DmkII and when doing screencasts like this, I use my Focusrite Saffire Pro 24DSP audio interface. On the Focusrite, I do sometimes use the DSP based compressor while recording which is very much like what you are suggesting here. In the pre-digital days, people recorded with hardware compressors all the time so it is legitimate. But you're sort of committed because the processing is applied in the recording and you can't back it out. So if you get the technique down, you can save yourself some time in post.
Great video Curtis! Do you have any tips on how to bring the lows up? Or is it better to just bring the gain up and then compress the highs? Thanks!
***** Hi Sam, just to make sure I understand when you say lows and highs, do you mean quieter and louder or do you mean lower frequency (bass) and higher frequency (treble) sound?
Sorry quieter and louder, I knew some this was off with that question!
***** No worries, just wanted to make sure I answer the right question! Yes, the typical way is to compress the louder peaks and increase the overall gain. Of course that can introduce additional issues (when you increase the overall gain, the noise gets louder as well), but we'll address ways to manage that in future episodes. Thanks for the question Sam!
Thanks for the info, looking forward to it!
This further helped me understand what you were doing in a previous video but I'm curious. 2 Questions from this: Is applying one global effect essentially the same as highlighting the highest peaks and lowering each one's amplitude manually?
Don't you run the risk of raising the amplitude of unwanted low frequency background sounds when you do this, especially when you initially raise the entire amplitude level? Is it better to record dialogue at a baseline lower amplitude, then bring it up in post even if that brings in unwanted noise, vs. recording at a higher frequency and using noise reduction in post? Thanks for responding to everyone's comments.
Zach Boyce Good questions again! Highlighting a section of audio and manually fading or attenuating it affects the entire audio waveform whereas using a compressor only affects those parts of the waveform that exceed the threshold. And that can sound a little different. They're both legitimate and useful tools. Compression can manage the peaks without affecting the overall loudness or body of the audio. But manually fading is useful if you want to lower the amplitude of everything.
Compression followed by normalization will definitely raise the noise floor. That's a calculated risk. If you use decent quality gear (doesn't have to be through-the-roof top tier), including a decent mic and pre-amp and analog to digital converters and record in a decent acoustical space, I find that I typically don't have to use any noise reduction filters. And while this is up to endless debates in audio forums, 24 bit encoding may also help here. But if anything in the chain isn't working well, noise reduction may become a necessity after compression.
Thanks for the questions, they're insightful!
Curtis Judd Ideally what decibel level should I be peaking at when recording dialogue, both in a controlled space, or in the field? I always thought -12 dB, but maybe I am wrong.
Zach Boyce I aim for -12 dB.
Thanks for the good instructional video. I'm curious, I've watched several of your videos on this topic and you say that you don't like to compress too much, usually between 2:1 and 2.5:1 and yet you compressed it twice at 2:1. Doesn't that fall into the aggressive range you say you like to avoid? Or is it that too much compression at once causes artifacts?
Sorry just read through another comment where you address this. Your response to that leads me to a further question, though. If 2:1 doesn't get it down enough, why not undo it and redo it at a higher ratio, etc, until you reach the levels desired?
Bob Adams You certainly could do that, but if you're working with a 24 bit audio recording, it doesn't seem to make a practical difference in my experience. In essence, I'm sort of finessing the peaks to where I want them a little at a time so that I don't go too far. Also, what I didn't show in this case, sometimes I use a different threshold on the second pass which gives a different result than a single, more aggressive pass because in the second pass you're compressing a different portion of the peaks, if that makes sense. Thanks for the question!
Curtis, great video. Is there a difference between running the compressor twice at 2:1 versus running it once at 3:1?
Ah, I was going to ask exactly the same question. Like any type of audio processing, I would imagine each pass introduces artifacts, so cumulative applications of the compressor may theoretically result in lesser quality than a single pass. Theoretically, at least, if not in practice. On the other hand, that may be complete rubbish.
Nick Laws Good catch and yes, they are the same. Once I get better at this, I hope to be able to get there more quickly in one pass. For now I like to apply it in layers to make sure I finesse it because it can be over done. Also, because I recorded at 24 bit, I'm not worried about adding artifacts like we might normally be concerned when working with 8-bit footage or 16-bit audio. One thing I sometimes do, but didn't get to here, is use a different, less aggressive threshold on the second pass. In that case, the effect can be a little less extreme because then the compressor affects less of the waveform. Still experimenting on that, but that's the reasoning.
Scott Vanderbilt That's a good point. I think with lower bit depth audio that can be a concern. With 24 bit recorded material, probably not a lot of artifacting introduced.
Curtis Judd
A helpful follow up. Thanks. :)
***** Glad it helped!
Hey Curtis, I've been using the Giant Squid lava mic to record in front of the camera and open space, but now I need to create tutorials in front of the computer and this mic is getting a lot of keyboard and mouse noise, like keystrokes and clicks. Do you have any recommendation on how to tackle that, either before or after the recording? Hope you can help :)
+Leandro Martinez Hi Leandro, unfortunately, lavaliers, most of which have omni-directional pickup patterns, are not great for rejecting sound near the mic. Ideally, one would use a dynamic microphone with a cardioid pickup pattern like this: ua-cam.com/video/EHgSC3C0pTw/v-deo.html
Mouse clicks and keyboard noise can be manually removed using spectral noise reduction but that's a rather tedious process and you have to go find and remove each one individually, and often you won't get it all. You'll also need an app with this capability (e.g., Adobe Audition, Izotope RX). I'll look to add a tutorial on how this works.
Sorry I didn't have better news that didn't involve buying a new mic or doing a lot of tedious work. :(
+Curtis Judd Actually you helped a lot :) You helped answering (rare thing on a lot of youtube channels) and helped with all of your videos. I thought on giving a try and ask you about the lava just to cover all possibilities here, because these kind of equipment here in Brazil are so expensive. I was able to buy the Giant Squid looking at your reviews and just as a friend was in the states, so thank you much for that :D
I imagined that I would need some omni-directional mic and lots of pages on the internet have recommend the Blue Yeti. Just for you to have some comparison, the Audio Technica AT2005USB that is about $65-$99 on Amazon, would cost me $200 dollars if I buy here (converting from my currency) :/
I'll wait some friend to travel again and grab me one of the Audio Technica's model. Thank you so much again :D
+Leandro Martinez Thanks Leandro and best wishes!
Just passing to let you know I've got a AT2005USB and it's awesome! Thank you so much for all your time & videos. It really helped me a LOT. Cheers :D
Hi Leandro, lavalier microphones typically have an omni-directional pickup pattern so they will simply be more likely to pick up keyboard and mouse noise than other more directional mics. The most effective options would be to go to a more directional microphone (e.g., a dynamic microphone) or use a quieter keyboard and mouse. I use a dynamic microphone with a cardioid pickup pattern to record my voice-overs like in this tutorial. This one was recorded with a Shure SM-58. I've also had good luck with the affordable Audio Technica AT2005 which can be used as a USB mic so you don't even need an audio interface if you do not already have one: ua-cam.com/video/EHgSC3C0pTw/v-deo.html Best wishes!
Why you apply the compression twice with x2 ratio instead of do once with x4 ratio?
Finessing it. Technically you could. But I'm trying to do as little compression as possible, just the necessary amount so that I can loudness normalize without clipping.
Why do you increase the overall gain? Does that option not affect the final outcome?
If not then why not us the scale on the right that's adjustable with the scroll wheel? Or is that a new feature?
+HollowProductions Increasing the gain makes the audio louder and more "present". It definitely does affect the final outcome.
The scale on the right can be zoomed with the scroll wheel but that just changes the view, not the gain of the waveform. That does not affect the audio, just lets you see it up close.
would you still raise the gain if you were to do a loudness maximization?
+Curtis Judd And how do you make sure that you haven't raised the gain too much? (besides it audibly being too loud or distorted)
+HollowProductions No, loudness normalization is raising the gain. That is all you need. :) You're on the right track, keep it up!
+HollowProductions Loudness normalization. :)
Link not working. Great video. Thx
Thanks. Sorry, which link isn't working? Just clicked those in the description and all seem to be working from here.
At 10.05 Loudness and Normalization. Thx
Thx..