iZotope RX Advanced Training for Dialogue Noise Reduction

[@user-hi5sj1jn6j]

Everything i´ve learned from your Videos is a thousand times better than what i´ve learned in university. Thanks for leading me on my way to become a great dialog editor for the last 3 years. I could do a lot of jobs in the quality i did just because of your videos. Thanks!

[@ThomasBoykin]

Glad to help

[@raistanar]

I've worked in audio post for years and still find these tutorials helpful. Always nice to check how someone else handles these kind of problems. Keep it up, great stuff!

[@ThomasBoykin]

Same, I love seeing new ways to take care of old problems

[@trevorhilton1143]

what perfect timing... starting a new dialogue editing project today. Thanks for all the gold over the years Thomas!

[@baptw]

Awesome vid. Here are my own personal tips in no particular order ! As others mentioned with spectral repair, using the vertical interpolation is awesome to get rid of tonal sounds, super useful for plates/keys ect.. ! Adjusting the bands setting to 4096 is the way to go. I have shortcuts switching between 128/Horizontal and 4096/Vertical I use all the time ! -Using the spectral denoise module to get rid of hums/resonances. Turning down the "noisy" reduction to 0 (just use the "tonal reduction") and then selecting only what you want to learn+process is great. I do need to give another chance to the de-hum module though. It seems to have improved a lot in recent updates. I turned my back on it a couple of versions ago, really didn't like it.. -Using the phase module to help Auto-Align when it struggles.. (by the way guys, render auto-align when you're done denoising, it helps the algorithm "see" better in a way !) -Just turning down selections from a few dB.. -De Rustle in low frequencies (wind, plosives, bumps...) -When applying dialogue isolate, take breaths out of the selection so you don't reduce them. I'm so hopping for a setting to separate breaths on dialogue isolate (and others plugs from acon, supertones and all of those). -On a general note, on a scene, I tend to start with getting rid of any short/punctual noises (like clicks, bumps ect..). Then I get rid of any tonal/longer "unnatural" noises (like hums, set/equipment noises, maybe car passes ect..). This way it should already sound quite natural. Still noisy maybe but in a natural way so when you start getting into reducing broadband noise, you actually don't need to do that much. Thanks again Thomas for all the knowledge you're sharing ! You're the only true professional that I know of who's willing to share skills online to that extent !

[@ahguitar1]

Will have to watch this in batches but thanks for your effort! (I subscribed because of your last RX restoration video). Your last video on this really improved my use of RX. It's great to watch a video that isn't edited down. Watching your work flow, shortcuts, constant freq band selection, undoing, trying again, is a great example and I'm now a lot faster as a result. I've not found many real examples of using RX and I'm self employed so learning it solo is a lonely experience. You're always wondering if there's a better way than what you've discovered. So this kind of video is invaluable.

[@ThomasBoykin]

Thanks, it means a lot to hear that! I'm not interested in being a "thumbnail KZbinr" with colored LED's in my studio and silly faces. It's also important to show that cleanup isn't just plowing through, you have to try and try again sometimes to get it right. Some of my videos are long but I hope it helps people learn. Do you think it's better to cut them into multiple shorter pieces, or leave em hanging at an hour if that's what it takes?

[@ahguitar1]

@@ThomasBoykin I'm not sure. I mean, I had 20-30 minutes spare so I listened to the intro, then jumped in to the examples closest to the things I wanted tips on. From a KZbin algo perspective you'd benefit from splitting it into multiple videos. Tags and titles specific to 'mouth clicks' etc to return better in peoples searches. Perhaps title them to be related "Advanced RX Training Part 1: Mouth Clicks" and put them all in a Playlist. But its extra work that way. You'd have to record a little intro dialogue for each one to explain. New thumbnails etc. The KZbin game. Keep doing the videos in longform and I imagine you'll get a smaller but dedicated audience. I appreciate the work anyway. Being persuaded to get the shortcuts setup was important.

[@sleepmeditationyoutube]

I think its tutorial is great for many reasons but one that stands out which is obvious is the fact this Edit is a extremely exaggerated example to show what can be done in such scenarios but in reality dialogue and VO is way cleaner to even start , so its great to see the useful tools that can be used for everyday dialogue editing which would save a lot of time with such methods.

[@ahguitar1]

Also, I literally didn't know what Deconstruct did and an hour after watching this video I was using it on a clients project 😂

[@ThomasBoykin]

Glad to be of service

[@StuartHetzler]

Deconstruct is an underrated module.

[@deafcatrecordings8053]

the tips with the deconstruct module is gold. thank you!

[@ThomasBoykin]

It's a great secret weapon

[@martynasmiknys7134]

Thanks Tom, your videos singlehandedly improved my sound! Thanks a lot appreciate it a ton!

[@ThomasBoykin]

Glad to help

[@ili6065]

Thanks alot for that so detaled explanation of your workflow!!! It`s undoubtedly very very usefull.

[@ThomasBoykin]

You're very welcome!

[@ili6065]

Great!! So many tips and tricks. Thanks allot for so detailed explanation. Despite the video is long it's very second is precious.

[@ThomasBoykin]

Glad it was helpful!

[@wojciechstecyszyn6040]

Thanks for the video, but ooooh, it was pain to discover at the very end of this looong video that I could do it along with you right from the beginning . Please next time give us this information about files to download at the very start of the video ;) I will go from the beginning anyway cause its best way to learn but it hurts .... I wanted to go for a walk, but I will sit here for another hour ;) Maybe already you could edit this video so next people watching will know right away. But anyway, thank you so much for sharing it with us. Cheers :)

[@RDAudioReviews]

Really enjoyed this. Looking forward to RX 11. I hope the tools are on par with other offerings, since I like working in RX Adv round tripping from Davinci Resolve.

[@alistairfarrant9155]

So helpful, especially your problem-solving strategies. thank you!

[@ThomasBoykin]

Glad to hear it!

[@bertomil]

Great tutorial. Never used the deconstruct modue. Will try it. One advice for you. I often use an ambience match, instead of copy-paste, to remove things that other modules ignore.

[@ThomasBoykin]

Good tip!

[@christopherbrock7934]

Wow dude. Thank you. Best Christmas gift ever.

[@Yadadamizzle]

I’m v grateful for your channel

[@ThomasBoykin]

My pleasure!

[@tiagolorena]

When cleaning up plates (keys, earrings, etc), you should start by removing the resonant ringing frequencies using vertical interpolation in the Spectral Repair, and then clean up using horizontal interpolation, you'll end up with fewer artifacts and it's easier for the algorithm to filter out stuff. I have never used Deconstruct that way tough and it makes total sense, I think it'll work better than Spectral repair in some cases. That DeBleed tip is also great, BUT you have to be careful with weird artifacts creeping in when mixing both signals. I have in the past kept the clean dialogue from a lav and then mixed in just the props handling noises from the boom without any weird artifacts, so that's an alternative.

[@ThomasBoykin]

Awesome tip, I'll try this out

[@makisoundeditor]

Yippee!!! Thank you so much for doing this!

[@ThomasBoykin]

No problemo

[@ChrisPFuchs]

Thanks for these videos as always. It's always cool to see what others are doing and learn a few things. I especially like the way you mention blending the raw with the processed audio. I've done this with Dialogue Isolate, but haven't really done it like you did in the plate example. Also, using the deBleed and setting it up that was was new to me. I might try that actually with SpectraLayer's Imprint function as well and see how it works compared to using DeBleed for that. A few of my own thoughts: You should try giving Dynamic DeHum a shot with the Learn function, as oppose to using it in Adaptive mode. In adaptive mode, it can incorrectly identify what is dialogue and what is hum. Especially the Truck Beep example it would have made a difference as a starting point. (I pulled the audio into RX to confirm). Sometimes it can be tricky getting it to 'Learn' what you want it to, but you can usually play around with it a bit until it does. Adaptive Mode is super nice when the hum is changing in pitch over time. Learn is fantastic when the Pitch is consistent. Both are Dynamic, as oppose to the old Static Mode. Also if you haven't, you should check out using dxRevive when Spectral Editing inside of RX. Not as like a broad Denoiser, but on fine selections where you might need resynthesis along with your other attenutation/NR methods. Kind of hard to explain over text but it's been a really powerful tool when used in this way. Also, I have a keybind that literally just does -3dB or reduction haha. I find I use it all the time on small selections that I just want to push down a few dB. That and copy pasting selections of roomtone.

[@ThomasBoykin]

Great tips, and yes I use plugins inside of RX, really great to hone in one specific issues that way

[@MarcelloOG]

Fantastic video Thomas! Thanks a lot!

[@jobinjayan2153]

thank you so much Thomas

[@inforobob]

Thanks much, this was very informative.

[@ThomasBoykin]

Glad it was helpful!

[@soundragon_]

Good day, thanks for your videos! With help of your videos I edited and mixed my first small movie.

[@ThomasBoykin]

Keep it up

[@Salantsoundstudios]

Great video thanks thomas!

[@ThomasBoykin]

If you watched the whole thing I commend your patience

[@Producersfader]

@37:21 soundflow will remember these settings if you have a macro for it :) Thanks for the great video

[@ThomasBoykin]

Great tip, but iZotope should fix it instead of me having to setup a macro in another program to fix their bug.

[@Producersfader]

@@ThomasBoykin Sure buying another software is not ideal.But for the moment it does remedy the issue.Souldflow comes with a bunch of other nice feature.Its worth a try. Any chance you can share this example session? Thanks again.

[@adelaudio]

@@ThomasBoykin plugin developers dont have influence on that. it's kind of random before pt 2023.9. i talked to accentize developers about that.

[@bartusenko]

Priceless tutorial! Thank you! I was wondering, maybe you could do a tutorial or maybe some tips here on how to clean that noise that sometimes lavs get when transmitter gets to far or just signal is not good? Kinda like white noise (broadband) that gets loud everytime talent speaks get this kind of issue almost all the time from production sound but can't find no info anywhere Would really appreciate your tips!

[@ThomasBoykin]

Oh yeah, Sennheisers are really bad for that but sometimes Lectros do the same thing. Good idea.

[@gastibarroule]

I’ve had good results with spectral repair selecting only the hiss snowy band, with vertical surrounding area towards the highest frequencies, but just a little of the lower frequencies of your selection. Usually with a high band count. It shapes the hiss into the higher frequency content “silence” without destroying sibilants.

[@bartusenko]

@@gastibarroule thanks, i'll try it out!

[@arifchagla8752]

You can save your rx connect settings as a windows configuration @37:19

[@ThomasBoykin]

Tried, doesn't work for me but maybe does for others

[@19Stride]

Great value!!! Thanks🎉 for

[@ThomasBoykin]

You bet!

[@gastibarroule]

For the difficult plate clip, using the bands and direction of spectral repair can also achieve great results. For higher pitch transient sounds, selecting horizontally with surrounding area going into the desired sound (the words) or silence, a smaller band choice will work better. A lower pitched hum or tonal noises will we better treated with higher bands. I only wish a could see a diagram of the bands used by spectral repair to understand why it works like this. Please Izotope!

[@girlyvoice3210]

With the plate example, when you imported them "clip by clip", shouldn't it then be possible to now subtract them, either spectrally or with phase inversion?

[@ThomasBoykin]

Theoretically possible but rarely works in practice

[@allvalley]

Hey buddy, i love this, have you done a video on how you have it mapped out? i feel like i could shave a massive chunk out of my editing if i got my head around this

[@ThomasBoykin]

Yes, I have a video on module chains coming up. Should help clear that aspect up.

[@The58845]

Hello. Could you explain the "2 seconds handle" at 5:01 please ? Why 2 seconds and not none ? Thank you in advance and thank you for this video.

[@ThomasBoykin]

I always have handles so I can tweak the edit afterwards

[@adelaudio]

how many days do you get on a show like the last of us for dx editing? i assume you only do the technical adr spotting, and the director's/production adr spotting and edicue lists is done by the adr editor?

[@ThomasBoykin]

45 minute show is usually dialogue edited in 5-7 days, just production sound

[@adelaudio]

@@ThomasBoykin yeah thats what i get on a standard budget tv show or movie. 45min/5 days, 90min/10 days. but i honestly dont have too much time on ALTs finding and digging too deep into every single click manual with this 10min per day shows. i was honestly shocked when you said 1 week per episode in the video. i thought, ok you guys will get 10 days or so on a netflix show. you are fast then. but i do adr spotting, edicue + adr editing on my 10days/90min shows. anyways, refreshing to see you guys dont take shortcuts by default like some obviously do.

[@bertomil]

Hello Tom. One general question. Why are you always ignoring those frequencies (mostly from handling noise) and not taking them off? Is it on purpose? The mixer should cut them with EQ, but it leaves some marks anyway.

[@ThomasBoykin]

It's easy to take off too much, especially in the dialogue edit. I would always rather leave some texture in, I rarely get asked by a mixer to do more processing!

[@adelaudio]

pt does remeber clip by clip and individual (aka last used setting) in pt 23.9

[@chinpou]

Looks great, thanks for making this. Is anyone else having trouble placing the order for the noisy wavs? The "Place Order" button isn't clickable for me.

[@ThomasBoykin]

Hey I just tested on Safari (iOS) and it works fine for me.

[@chinpou]

I put in more time on it but eventually downloaded by checking out as a guest. Excited to get learned!@@ThomasBoykin

[@santhoshraina7360]

Hii tom,My RX 9 version takes very long time to compute and display the spectogram after any passes i do even small piece of spectral repair, any solution to this?

[@ThomasBoykin]

What kind of computer are you using? Where is the audio file located (external or internal HDD)?

[@santhoshraina7360]

@@ThomasBoykin Im using HP omen laptop with audio files located internally

[@lordsnot2268]

"some of you will be on (version) 11 or 12" Well, maybe in the future. RX 10 is the latest version to be released. 😄

[@ThomasBoykin]

I leave my videos up for a long time!

[@lordsnot2268]

@@ThomasBoykin LOL. Classic. Well played, sir! 😎

[@Paulkruz1]

So we giving away game for free now?

[@ThomasBoykin]

I can show you how to kick the ball but it’s up to you to score the goal

[@Paulkruz1]

Its not soccer by the way, it's FOOTBALL @@ThomasBoykin

[@Blueblackngold]

This is stressful