As a bit of friendly feedback you have processed the audio way too much on this - doesn't sound natural at all - way too dead on the gating, sibilance and artefacts present.
Thanks for the feedback. Can you tell me more about the artifacts you are talking about? I actually do very little in terms of processing, so I'm concerned that this is what you're hearing. From the perspective of artifacts. the only thing I'm hearing is the vocal tics that I have. I am aware of the resonance that I get from the big monitor in front of me. I am still looking for a solution for that. I do apply some de-essing because I have very strong sibilance. The problem with strong sibilance is you can only take the edge off. It can't be eliminated, but I'm all ears if you have some solutions or techniques I can try. In terms of gating, the only time I use a gate is on tutorial videos to eliminate mouse clicking. I spent a good amount of time fine tuning my room and set up to eliminate all noise from my system specifically to clean up my noise floor and to reduce the amount of processing needed.
@@jesse.mccune It's possible that you have deadened your room too much, but to me it sounds like you might have run a voice isolator or similar and set it too harshly. RUclips does compress the audio but not enough to be noticeable usually. I loaded your audio into RX10 and the noise floor is zero which just sounds weird and unnatural, and I can hear the noise bleeding through when you talk which makes me think you are using an online voice isolator such as the one built into Davinci Resolve or some other adaptive noise reduction (would recommend batch processing using an empty space as noise sample so you don't hear the transients and hear it through speech).
I went back and checked my session. I forgot to turn off Sonible smart:Gate because it was on. That would explain the lack of noise floor. My chain for this video was: • Sonible smart:Gate • Mouth De-click • Voice De-noise • Softube Lil' Freq for EQ and De-ess
I’m in for a quick comparison video from you as well as something more detailed. I stumbled across Mouth De-click in RX6 and it positively changed my life for VO work I was doing at the time. While I’m always open to adding a new tool (or upgrading in this case), VEA soured my excitement on new Izotope announcements. As I’m getting into some podcast work and trying different tools, Izotope will have to really impress me with RX11 to get access to my credit card.
Thanks for the feedback, Mike. Mouth De-click is a necessity for me. Not just for my own voice, but for working on client work too. I haven’t been overly impressed with RX since they’ve moved to annual updates. I know this it’s been 16-18 months since RX10, but the updates have been lackluster. I truly feel this is a make or break release for Izotope from the perspective of podcast editors. If their new Voice Isolation doesn’t best what’s already out there, what reason is there for us to upgrade? I’ll keep using RX10 until it’s no longer functional on my computer or someone comes out with better options for cleaning up mouth noise and plosives.
@@jesse.mccune The best Mouth Declicking I've heard for broadly applying across the entire audio is Auphonic's Denosier. They've recently added Mouth noises to the Neural Network training of their AI denoiser. The downside is it's tied to the Denoiser's DeReverb and general Denoising.. And it's an upload-to-a-cloud-to-be-rendered type service. So not a lot of flexability. I was really hoping RX would add AI training to their individual modules. People praise the RX Mouth Declick, but it definitely can be improved and I really don't like the sound of it when applied broadly to audio, such as podcasts. I would pay good money for a Neural Network Mouth Declicker plugin. Also, I don't know if you've tried it in this way, but dxRevive is pretty effective at cleaning up plosives. I know you've cut out Spectral Editing from your workflow to save on time, but dxRevive has been sort of my go to for quickly making a selection around the plosive in RX and processing it with dxRevive. It's really saved me time and is infinitely better than RX's Deplosive. You could probably do something similar in the editor if you have dxRevive Pro (make a selection, processing with dxRevive using only the low band.) In terms of RX 11, I have confinidence that the quality of their new Denoisers will at least be competitive. I'm hoping it has some sort of unique functionality; i.e, Dialogue Isolate has Ambient Preservation which allows you to denoise the audio but leave the broadband noise. I am also hoping the 'High Quality' processing mode for 'Less than Ideal' recordings means it's doing some sort of synthesis similar to Adobe Enhance or dxRevive, although it could just mean an updated offline algorithm. But both versions of their Spectral Recovery has been poor, so who knows. It's interesting, because so many people doing this work have already bought the alternatives. I really expected Izotope to come out with a Realtime AI Denoiser for RX 10 and was really surprised when they didn't. With RX 11, I'd of at least expect them to come out with something unique to seperate themselves from the competition, but instead they seem to be playing catchup and going head to head with their competition this iteration instead. Hopefully there's more they announce with it in the next few weeks! As for I, I've been getting by with RX, Auphonic, dxRevive, and Cedar DNS (that VoiceX looks pretty interesting but damn is it expensive in the age of rapidly improving AI), but have been holding out on Supertone Clear and Accentize Deroom for RX 11. So hoping it competes.
@ChrisPFuchs I haven't tried Auphonic's de-noiser because I'm always looking for efficiency and avoiding cloud-based and subscription-based services. Being tied into their dereverb and denoising functions is another no-go for me. It's the reason I wasn't a fan of dxRevive when it was released. At least they fixed that with a recent update so we can turn off the EQ it was applying. That's an interesting approach for plosives. What issues do you hear with RXs De-plosive that makes this approach better? None of my clients are willing to pay extra for that level of service where I'm manually fixing each plosive. I will keep this in mind if I have a big budget or high profile use case where this level of work is warranted. With RX 11, I'm not expecting anything from it, so that I may find myself surprised if they deliver something special, but won't be let down if the only draw is the new real-time voice isolation plugin and it turns out to be a middle-of-the-pack tool. Since they're calling it a voice isolation tool, I don't think we'll see an ambient preservation option since the way these tools are trained would tend to make that a challenge unless they take the approach Supertone did with Clear. When I read about "High quality processing mode" I had the same thought that it's going to be something to dxRevive and Adobe's voice destroyer. It would need to be an improvement over those two monstrosities to be a selling point for me. The biggest issue with those types of technologies is that it changes the voice. Once they can figure out how to implement it without changing the inherent qualities of the voice, it will be a useful tool, but that is not the case today. They are better than Spectral Recovery, though. Side note, Adobe swears they aren't using synthesis in their voice destroyer algorithm, but I don't buy it because I've heard instances where words are created that weren't there in the raw audio. At this point, my expectation is that I'll upgrade RX when the plugins no longer run on my OS or they have introduced enough new features and improvements to warrant the upgrade price. It will be a tough sell to upgrade RX Advanced simply for an average quality voice isolation tool. As you said, most pros have already bought alternatives, so that anchor is set. RX Advanced is now going to come down to how well the voice isolation plugin compares to the others already on the market. At $399 list for an RX Advanced upgrade, someone could buy Clear and DxRevive Pro and have most of their bases covered for noise and reverb reduction. We all have our preferred tool kits and there's still a place for RX in mine as long as they're plugins. VoicEx definitely isn't for people who aren't working with clients. It's not an everyday plugin for me, that role is filled with Supertone Clear. I look at Clear as a screwdriver and VoicEx as a drill. A screwdriver is much more handy around the house and gets a lot more use, but sometimes, a project comes up where a drill will make things easier and save time. If RX's new plugin does something that my other tools don't, it'll get added to the toolbox. Noise and reverb reduction tools are a money maker tool and I have no problems having multiple options if they all fill a specific need.
@@jesse.mccune Yep, I agree Auphonic's limitations is definitely less than ideal. I'll use the Denoiser for long format podcats that have a ton of Mouth Clicks to save time. I point it out though because it really is significantly better at avoiding parts of the speech traditonal Mouth Declicker's dig into- Transients of P's and B's and sometimes even upper frequencies on vocal frys. And it attenuates (based on the amount of reduction) instead of removes, which is quite natural sounding. As for dxRevive, I agree although in dxRevive Pro, Retain avoids reverb at least. RX Deplosive just isn't that great. You can set it gentle enough that it's grabbing some of the low frequency bumps but as soom as you try to increase the strength and up the frequency range, it'll start digging into parts of the voice fundamental. Also, some plosives persist for a bit longer than the short bump that Deplosive looks for. On really bad plosives that completely disrupts or masks the harmonics of the voice, dxRevive will synthesis that missing content back. Definitely time consuming/surgical work. But this is why I mentioned I was really hoping for RX to update their individual modules with Neural Networks; that way it's a little bit cleaner when broadly applied to the dialogue and the time/quality compromise isn't as bad. *Because my comments are getting deleted for whatever just wanted to elaborate here; Deplosives works alright on broad, subtle settings- I'm speaking more about the plosives where you'd normally have to go into the Spectral Editor to clean out that dxRevive can be effective and save some time. I'm not sure what Adobe means by not doing Synthesis, but I believe you can literally upload audio to it with missing high or low content and it'll fill out the missing spectrum? I've seen it add high some high frequency content. I'll have to test that again out of curiosity. I suppose it probably doesn't fill in gaps of missing dialogue harmonics like dxRevive would. I'm actually a little surprised you view Clear and VoiceX like that. From what I've heard, VoiceX really manages to retain the audio quality of the voice, but sometimes isn't always better at handling the noise. So I'd think the go-to would be VoiceX as the general Denoiser and Clear for the Reverb Reduction or when the Neural Network training of VoiceX doesn't agree with a certain noise. Neither of them really give a ton of finesse. But I don't own VoiceX so I could be wrong. I know a lot of people don't have a lot of confidence in RX right now; the best features of the last two upgrades for me is something nobody talks about- feather mode and Return Selection to Previous State haha. But AI is just progressing so fast right now and Izotope has had 2.5 years (let's be honest RX 10 doesn't count), to create a competive modern AI dialogue Denoiser that works in realtime. It has to at least be competitive, for their sake. Personally I value improvements to the RX editor as well as I use it everyday for work.
@ChrisPFuchs The main reason I view Clear and VoicEx the way I do comes down to efficiency. Clear doesn't have much impact on render times while VoicEx does, so if I can get acceptable results with Clear, that's what I use. I'm simply looking to get the job done quickly while sounding good enough that no one will stop listening due to the audio quality. I could A/B test things and I'd probably find similar things to what you've described above, but for podcast editing work, there's no incentive for me to dive that deep into the weeds. Clients just don't pay enough for it. I'll set up my mix and start editing. If I hear something during the edit that isn't working well, I'll adjust or replace the plugin. Are there times where de-plosive might rob the fundamental of some low end? Probably, but if I'm not listening for it, I wouldn't be able to say. This is one of the other reasons I wanted to get out of the RX editor. Working with plugins allows me to make judgments from the listener's perspective rather than an engineer's perspective. It forces me to focus more on the things that matter to the listener and pulls me out of the headspace where I would otherwise obsess, A/B test, and overanalyze everything. There is a time and place for the RX editor because it offers a lot of power that can't be replaced by plugins, but for me, podcast editing isn't one of those scenarios.
It’ll be interesting to see what it is. My best guess is some sort of metering or loudness presets, but that will apply more to music production than podcasts.
As a bit of friendly feedback you have processed the audio way too much on this - doesn't sound natural at all - way too dead on the gating, sibilance and artefacts present.
Thanks for the feedback. Can you tell me more about the artifacts you are talking about? I actually do very little in terms of processing, so I'm concerned that this is what you're hearing.
From the perspective of artifacts. the only thing I'm hearing is the vocal tics that I have. I am aware of the resonance that I get from the big monitor in front of me. I am still looking for a solution for that. I do apply some de-essing because I have very strong sibilance. The problem with strong sibilance is you can only take the edge off. It can't be eliminated, but I'm all ears if you have some solutions or techniques I can try. In terms of gating, the only time I use a gate is on tutorial videos to eliminate mouse clicking. I spent a good amount of time fine tuning my room and set up to eliminate all noise from my system specifically to clean up my noise floor and to reduce the amount of processing needed.
@@jesse.mccune It's possible that you have deadened your room too much, but to me it sounds like you might have run a voice isolator or similar and set it too harshly. RUclips does compress the audio but not enough to be noticeable usually. I loaded your audio into RX10 and the noise floor is zero which just sounds weird and unnatural, and I can hear the noise bleeding through when you talk which makes me think you are using an online voice isolator such as the one built into Davinci Resolve or some other adaptive noise reduction (would recommend batch processing using an empty space as noise sample so you don't hear the transients and hear it through speech).
I went back and checked my session. I forgot to turn off Sonible smart:Gate because it was on. That would explain the lack of noise floor. My chain for this video was:
• Sonible smart:Gate
• Mouth De-click
• Voice De-noise
• Softube Lil' Freq for EQ and De-ess
I’m in for a quick comparison video from you as well as something more detailed. I stumbled across Mouth De-click in RX6 and it positively changed my life for VO work I was doing at the time. While I’m always open to adding a new tool (or upgrading in this case), VEA soured my excitement on new Izotope announcements. As I’m getting into some podcast work and trying different tools, Izotope will have to really impress me with RX11 to get access to my credit card.
Thanks for the feedback, Mike. Mouth De-click is a necessity for me. Not just for my own voice, but for working on client work too.
I haven’t been overly impressed with RX since they’ve moved to annual updates. I know this it’s been 16-18 months since RX10, but the updates have been lackluster. I truly feel this is a make or break release for Izotope from the perspective of podcast editors. If their new Voice Isolation doesn’t best what’s already out there, what reason is there for us to upgrade? I’ll keep using RX10 until it’s no longer functional on my computer or someone comes out with better options for cleaning up mouth noise and plosives.
@@jesse.mccune The best Mouth Declicking I've heard for broadly applying across the entire audio is Auphonic's Denosier. They've recently added Mouth noises to the Neural Network training of their AI denoiser. The downside is it's tied to the Denoiser's DeReverb and general Denoising.. And it's an upload-to-a-cloud-to-be-rendered type service. So not a lot of flexability. I was really hoping RX would add AI training to their individual modules. People praise the RX Mouth Declick, but it definitely can be improved and I really don't like the sound of it when applied broadly to audio, such as podcasts. I would pay good money for a Neural Network Mouth Declicker plugin.
Also, I don't know if you've tried it in this way, but dxRevive is pretty effective at cleaning up plosives. I know you've cut out Spectral Editing from your workflow to save on time, but dxRevive has been sort of my go to for quickly making a selection around the plosive in RX and processing it with dxRevive. It's really saved me time and is infinitely better than RX's Deplosive. You could probably do something similar in the editor if you have dxRevive Pro (make a selection, processing with dxRevive using only the low band.)
In terms of RX 11, I have confinidence that the quality of their new Denoisers will at least be competitive. I'm hoping it has some sort of unique functionality; i.e, Dialogue Isolate has Ambient Preservation which allows you to denoise the audio but leave the broadband noise. I am also hoping the 'High Quality' processing mode for 'Less than Ideal' recordings means it's doing some sort of synthesis similar to Adobe Enhance or dxRevive, although it could just mean an updated offline algorithm. But both versions of their Spectral Recovery has been poor, so who knows.
It's interesting, because so many people doing this work have already bought the alternatives. I really expected Izotope to come out with a Realtime AI Denoiser for RX 10 and was really surprised when they didn't. With RX 11, I'd of at least expect them to come out with something unique to seperate themselves from the competition, but instead they seem to be playing catchup and going head to head with their competition this iteration instead. Hopefully there's more they announce with it in the next few weeks!
As for I, I've been getting by with RX, Auphonic, dxRevive, and Cedar DNS (that VoiceX looks pretty interesting but damn is it expensive in the age of rapidly improving AI), but have been holding out on Supertone Clear and Accentize Deroom for RX 11. So hoping it competes.
@ChrisPFuchs I haven't tried Auphonic's de-noiser because I'm always looking for efficiency and avoiding cloud-based and subscription-based services. Being tied into their dereverb and denoising functions is another no-go for me. It's the reason I wasn't a fan of dxRevive when it was released. At least they fixed that with a recent update so we can turn off the EQ it was applying.
That's an interesting approach for plosives. What issues do you hear with RXs De-plosive that makes this approach better? None of my clients are willing to pay extra for that level of service where I'm manually fixing each plosive. I will keep this in mind if I have a big budget or high profile use case where this level of work is warranted.
With RX 11, I'm not expecting anything from it, so that I may find myself surprised if they deliver something special, but won't be let down if the only draw is the new real-time voice isolation plugin and it turns out to be a middle-of-the-pack tool. Since they're calling it a voice isolation tool, I don't think we'll see an ambient preservation option since the way these tools are trained would tend to make that a challenge unless they take the approach Supertone did with Clear. When I read about "High quality processing mode" I had the same thought that it's going to be something to dxRevive and Adobe's voice destroyer. It would need to be an improvement over those two monstrosities to be a selling point for me. The biggest issue with those types of technologies is that it changes the voice. Once they can figure out how to implement it without changing the inherent qualities of the voice, it will be a useful tool, but that is not the case today. They are better than Spectral Recovery, though. Side note, Adobe swears they aren't using synthesis in their voice destroyer algorithm, but I don't buy it because I've heard instances where words are created that weren't there in the raw audio.
At this point, my expectation is that I'll upgrade RX when the plugins no longer run on my OS or they have introduced enough new features and improvements to warrant the upgrade price. It will be a tough sell to upgrade RX Advanced simply for an average quality voice isolation tool. As you said, most pros have already bought alternatives, so that anchor is set. RX Advanced is now going to come down to how well the voice isolation plugin compares to the others already on the market. At $399 list for an RX Advanced upgrade, someone could buy Clear and DxRevive Pro and have most of their bases covered for noise and reverb reduction.
We all have our preferred tool kits and there's still a place for RX in mine as long as they're plugins. VoicEx definitely isn't for people who aren't working with clients. It's not an everyday plugin for me, that role is filled with Supertone Clear. I look at Clear as a screwdriver and VoicEx as a drill. A screwdriver is much more handy around the house and gets a lot more use, but sometimes, a project comes up where a drill will make things easier and save time. If RX's new plugin does something that my other tools don't, it'll get added to the toolbox. Noise and reverb reduction tools are a money maker tool and I have no problems having multiple options if they all fill a specific need.
@@jesse.mccune Yep, I agree Auphonic's limitations is definitely less than ideal. I'll use the Denoiser for long format podcats that have a ton of Mouth Clicks to save time. I point it out though because it really is significantly better at avoiding parts of the speech traditonal Mouth Declicker's dig into- Transients of P's and B's and sometimes even upper frequencies on vocal frys. And it attenuates (based on the amount of reduction) instead of removes, which is quite natural sounding. As for dxRevive, I agree although in dxRevive Pro, Retain avoids reverb at least.
RX Deplosive just isn't that great. You can set it gentle enough that it's grabbing some of the low frequency bumps but as soom as you try to increase the strength and up the frequency range, it'll start digging into parts of the voice fundamental. Also, some plosives persist for a bit longer than the short bump that Deplosive looks for. On really bad plosives that completely disrupts or masks the harmonics of the voice, dxRevive will synthesis that missing content back. Definitely time consuming/surgical work. But this is why I mentioned I was really hoping for RX to update their individual modules with Neural Networks; that way it's a little bit cleaner when broadly applied to the dialogue and the time/quality compromise isn't as bad. *Because my comments are getting deleted for whatever just wanted to elaborate here; Deplosives works alright on broad, subtle settings- I'm speaking more about the plosives where you'd normally have to go into the Spectral Editor to clean out that dxRevive can be effective and save some time.
I'm not sure what Adobe means by not doing Synthesis, but I believe you can literally upload audio to it with missing high or low content and it'll fill out the missing spectrum? I've seen it add high some high frequency content. I'll have to test that again out of curiosity. I suppose it probably doesn't fill in gaps of missing dialogue harmonics like dxRevive would.
I'm actually a little surprised you view Clear and VoiceX like that. From what I've heard, VoiceX really manages to retain the audio quality of the voice, but sometimes isn't always better at handling the noise. So I'd think the go-to would be VoiceX as the general Denoiser and Clear for the Reverb Reduction or when the Neural Network training of VoiceX doesn't agree with a certain noise. Neither of them really give a ton of finesse. But I don't own VoiceX so I could be wrong.
I know a lot of people don't have a lot of confidence in RX right now; the best features of the last two upgrades for me is something nobody talks about- feather mode and Return Selection to Previous State haha. But AI is just progressing so fast right now and Izotope has had 2.5 years (let's be honest RX 10 doesn't count), to create a competive modern AI dialogue Denoiser that works in realtime. It has to at least be competitive, for their sake.
Personally I value improvements to the RX editor as well as I use it everyday for work.
@ChrisPFuchs The main reason I view Clear and VoicEx the way I do comes down to efficiency. Clear doesn't have much impact on render times while VoicEx does, so if I can get acceptable results with Clear, that's what I use. I'm simply looking to get the job done quickly while sounding good enough that no one will stop listening due to the audio quality. I could A/B test things and I'd probably find similar things to what you've described above, but for podcast editing work, there's no incentive for me to dive that deep into the weeds. Clients just don't pay enough for it.
I'll set up my mix and start editing. If I hear something during the edit that isn't working well, I'll adjust or replace the plugin. Are there times where de-plosive might rob the fundamental of some low end? Probably, but if I'm not listening for it, I wouldn't be able to say. This is one of the other reasons I wanted to get out of the RX editor. Working with plugins allows me to make judgments from the listener's perspective rather than an engineer's perspective. It forces me to focus more on the things that matter to the listener and pulls me out of the headspace where I would otherwise obsess, A/B test, and overanalyze everything. There is a time and place for the RX editor because it offers a lot of power that can't be replaced by plugins, but for me, podcast editing isn't one of those scenarios.
My guess is that second thing is just a rebranded version of VEA that is integrated into RX
It’ll be interesting to see what it is. My best guess is some sort of metering or loudness presets, but that will apply more to music production than podcasts.