Dude, you have no idea how much pain and suffering you and adobe removed from my life. I live in an empty home with high ceilings and no insulation or any type of sound proofing. I spend hours trying to make my audio sound descent and it still sounds horrible. I literarily just tested this with a piece I'm editing right now and I'm in tears... thanks so much dude!!
I learn so much from this channel! I subscribed for the art/design tips, but I'm loving these videos that branch out into audio, AI and new software too. Keep up the great work!
Glad you like it, Dan 😎 I love these topics, and sometimes, I am concerned whether the photographers or designers who watch the content would like this. You've put my worries away, thank you.
@@PiXimperfect It didn’t remove background noise completely. I also used a hammer on a pot lid while I was speaking. You can still hear a “clanking” sound in the bg, however it wasn’t very loud. And it turns your voice into a robotic voice.
Yes and no, one thing he makes clear is that it's a great tool for, as you say, terrible sound, but it kinda ruins well recorded one, so it would be useful for UA-camrs with no audio equipment and low skills for audio editing mostly
I edit my audio by myself because there are some frequencies that can't be automated, you have to adjust them manually for a crisp voice... And also it doesn't take that much efforts to edit them manually, there may be some difficulties when you're editing your audio for the first time because you have to find best frequencies for your type of voice, and after that just make a preset of it and you're good to go...
I'm a professional Voiceover artist with a great appreciation of getting it right from the get go when recording professionally (Not so much for UA-cam content) What a fabulous looking video 👏🏽👏🏽👏🏽👏🏽. 10 out of 10 for video quality. As for sound quality of the enhancements, I found it all very muffled and lacked the natural sizzle that human voices have. I think there's way too much made of getting rid of sibilance. Sibilants is a natural part of speech clarity. I'm not impressed at all with the tools. As you absolutely rightly stated my friend, you can't beat a treated environment married to a quality microphone used correctly. Very engaging video. Thank you for the effort you made in putting it together 👍🏼. It's a personal preference of course. 🙏🏽
Thank god for this comment! 100% right, sadly it seems like people are going with it not appreciating the beauty of detail and doing things the right way. I can definitely see the use of this in certain situations and for social media but not more than that.
@@99boardwalk Scroll through the comments and you'll see it's the podcasters and UA-camrs who are excited by this AI. A voiceover artist can't use it. Not for commercial, animation or anything that pays over $50
Your channel is always informative and entertaining. You do such a good job keeping the pacing just right, explaining things, testing things, and showing how to use Adobe's tools/products. This blew me away! Thank you for another great video. :)
tip: edit your audio first with a audio editing software and remove some of the noise. If your raw audio is filled with too much noise, Adobe AI will decide your voice as noise and clean that too. Learned this the hard way
It‘s not only sound imperfections or noise removal - it’s speech recognition as well. I tried extremely echoey non-English audio. It removed echo but also tried to recognise speech and in the end produced incomprehensible sort of English-sounding file.
Yes, it's not perfect. But I was impressed with most of the audio files I used with it. Exceptions: When people are talking over each other during a discussion, or when talking while there is applause or laughter. It just goes weird. But all in all, a nice tool to have for the times it might help.
I tried it as well. It only works with english. The fact is the AI tries to gather audios from here and there and tries to merge those sounds together with the original to give a "clean audio", but english recognition. I have no doubt they will improve it, but when the time comes they add other languages, they will put a price on it, sadly.
It’s worth knowing for people recently coming to this that Adobe introduced a limit and now you can only enhance one recording per hour - So if you do a test piece of a few seconds just to check what the output would be, you’re going to have to wait another hour before you can upload your track. It’s not a massive problem but it could impact on your workflow if you are up against a tight deadline
First couple of tests went great. I still here a little handling and tapping but most of it is removed. I still heard some sibilance. Sibilance is less but the adjustment is a bit abrupt, like not moving the sliders in blend-if far enough apart. Noise is gone. I agree with you about the breath sounding artificial. Breath can be removed manually in many sequencers. Adobe Podcast does a great job overall, I would say.
I haven’t checked, but I would imagine you can fine a free “de-esser” online, to control your extra sibilance before you use the AI tool. Logic X comes with a de-esser. Hope this helps 👍
I am absolutely floored. This is incredible! It might not produce top notch audio at the moment but one cannot deny how incredible this is for beginners who can't afford expensive gear! After all, not everyone will be able to immediately buy a decent microphone when they start out so this will help* loads of people!
You'll never get top notch audio from source material that is not top notch quality. With good enough processing you can get something that sounds similar to a quality recording from anything, but in the end it'll be an approximation of the source material no matter how good it sounds.
@@gutterg0d the only thing is this is A.I. so it has the ability to fill in gaps and create whats not there. I agree with you as of now. But A.I. can potentially do some crazy cool shit. This is probably free for now so their machine learning can get better and better. Plus this shit can learn so much it might create vocals from nothing
While I'm using RX by iZotope on a daily basis (which is a totally advanced and professional software), I must admit that this is a really great tool for everyone who needs some help e.g. with their YT videos, but have no knowledge about the whole audio processing part.
@@TimberWulfIsHere Whats the more advanced version of RX? Also Ive been using RX for years and have never gotten such clean low & mids like the AI does
Love the curves and appreciate the attention to detail on the tests. Always a joy to watch and learn. Love the growth of the background and lighting compared to the early days. Nice yluse of color and symmetry.
Thanks for another great video 🙂 I think this is usable and can save some recordings - but it does not replace a good microphone and studio. Just like with computational photography it cant yet replace large sensor, good lenses etc
Thank you so much for this! I started doing vlog's in my car on the way to work and needed something to clean up the road noise. This seems perfect for my use.
The only shortcoming is that, it doesn't work well on non English audios. I've tried with some indian languages and Bengali, most of the time, noise, echo are fixed, but main audio is a gibberish and it isn't understandable anymore. I guess this software uses english speech to text at some point of editing, and maybe the voice models AI uses is in English.
I was wondering how it would handle accents/foreign languages myself. Not entirely surprised it has issues, but that may be why it’s a bit of a demo at this point.
08:33 the original is better also because it compressed the tonality in your voice, removing its texture and personality. However, I have often (even recently) had really bad audio that this may have helped a lot to get something usable.
oh man, this is a next-freaking-level tool. Audio went bad for some of my shorts and I just found this tool through your video right in time. Can't be more thankful today!
tip: edit your audio first with a audio editing software and remove some of the noise. If your raw audio is filled with too much noise, Adobe AI will decide your voice as noise and clean that too. Learned this the hard way
This tool from Adobe is insane, I tried with two audio files from different microphones and compared to the same audio file with treatment on Premiere, and the results are awesome, one of my lav mics was echoing a lot and the processed file removed all echoing ambient sounds which became way better than the other one. Also you can still apply audio filters on top of the processed file from Adobe, I used the same preset filters I use on my microphones and the audio became even better. Now I'm really thinking of using this tool to my other videos too.
Somebody needs to run all the voice files for older games with garbage compressed audio through this and then put them back in the game. (Assassin’s Creed 1, Far Cry 2, CoD World at War, etc.)
Actually I prefer the processed audio from adobe even with the nice mics in studio mainly because it was softer and more pleasant less harsh. This might be simply because I am listening on iPad speakers but I think with any hi fi headphones I would actually feel stronger about because it would likely sound more intense. I notice that your mixing preferences seem to be a bit sharp for the my tastes so using only your own audio samples might be making it sound better than it actually does because when I strain I can detect a sort of text to speech artificial sound to the processed file. It just sounds better because it’s easier on my ears, if your mixing was more bass forward perhaps I would be more annoyed than relieved.
I hear you, but you can use other methods to take the edge off the sound if that's what you want; EQ, Soothe, McDsp SA-2 etc. The extra degradation to the audio isn't worth it if that's all you're trying to achieve.
For AI to do this without subband modulation or spectral smearing is pretty impressive. I'd imagine there's a good chance those listening on smart device speakers will even prefer the processed audio to the original.
Thank you! I saw Adobe AI on another channel, but they failed to say what the Web address was and, since they also failed to give the proper name of the Website (and Adobe won't let you search for it on their site) I couldn't find the proper page. This is a great help as I have just begun doing audio books and a neighbor with dogs has just moved in. Sigh.... Keep up the great work!
Definitely not a replacement for capturing audio with high quality mics at the correct distances, but if something tragic happens and your audio is ruined for some weird reason, this could be really useful to at least get usable audio. Still I would do a little EQ and mixing to make it sound good but this could be great for removing handling noise and plosives, which are a nightmare to remove.
I think it makes a massive difference in all areas EXCEPT the sibilance. I think in every case of sibilance, it makes it worse as it sounds like its boosted mid which gives it a weird chamber sound. Just as trends happen in graphic design, I feel like the trend for Podcasts at the moment is to have a warmer sound overall, which probably makes it accurate for Adobe AI to be Adobe Podcasts. It's not my personal preference, as you say sometimes it's a bit abrupt and unnatural, taking breaths away etc, but it might suit a podcast sound.
OMG, what is happening to this world? We`re witnessing so many mind blowing discoveries that are constantly appearing soooo damn rapidly. AI is developing too fast, but that`s cool anyway 🤙 What else can it be applied to? ;)
HOLY CRAP! This is a savior for me!! I just recently started voice acting, and my mic stands out like a sore thumb in comparison to my fellow VAs mics!
I've been watching you for years but hadn't seen a video of yours in a while. I've always bookmarked old vids for reference. So I'm pleased to see how far you've come in video quality and lighting setup. Your information and the way you teach hasn't changed and has always been king. But the look and flow has shown a very pleasing and study growth matching the Information you share. Just wanted to say congratulations 🎊 and I hope your success continues. You and your team have earned it. Cheers from Sacramento California
sounds amazing! I've been searching for years for something to remove the echo and fix the proximity problem. No need to process high quality audio through the AI. The AI made it sound "over-processed". It still sounds great, but sounds "uncanny valley" in some way.
To me it sounds like it would be a useful tool if you are really struggling with background noise or using an absolutely terrible/broken microphone. But if used on anything even barely passible, I feel like a lot of personality is taken out of the voice. From my own testing, this seems to be especially true for people with higher-pitched or breathy voices. This video was done excellently by the way. Very comprehensive, well presented, and thorough!
Yeah, I felt the same when trying some of my own audio. The AI made a terrific job, but sometimes it's a bit too agressive and cut out a few sounds that were not noise, but were under a certain thresold level, like a harsh noise gate.
Unmesh, it's wonderful that you have a U87 for your audio but I would recommend getting a pop filter or something similar in front of it. It's very sensitive to moisture and eventually it will start sounding dull. It's a great mic & investment, protect it. ;-)
Thank you Unmesh, it really made a difference. I tried it will you were explaining all about it. And I loved it. It will help me alot!🙏🏻 1 thing though it make somewhat like Artificial like a robotic at the end result but that's its awesome, thank you again.
Pretty cool. Davinci Resolve also has a new speech processing feature that works pretty well for free. I recently botched my lav mic recording and clipped the recording into the camera, and it cleaned it up pretty well. I wouldn't count on it all the time but it can be a lifesaver.
I've tried this ai. I think if the audio is in English this will give good results. The ai was trying to figure out what my subject was saying and converting it. My subject is a Filipina now she sounds Chinese hehe!
As a sound engineer I'm always confronted with the statement "We'll fix it in the mix.". Although we can perform little big miracles with the technology available today, best would be to do it properly at the time of recording. That of course includes the fact that for instance in music the arrangement (comparable to composition in photography) and the performance need to be of the highest quality possible. Otherwise we're just polishing a t*rd 🙊 Great content as per usual and Adobe's AI audio tools could be handy in some extreme cases here! Keep it up! ✌🏻
wow, audio gate, gain, roomtone removers, reverb reduction and more with a click. I love it. This will save me a lot of time of mastering audio for my voice acting
You have created an incredibly useful instructive piece! For example, every UA-cam user should see an echo test so that they can record professional soundtracks without paying anything extra. The room echo prevents listening to audio tracks louder, for example, while doing dishes in another room. You bring out all those important points like sibilance. Absolutely amazing work, not to mention the great tool you mentioned, which can do wonders in the right context. This video will immediately enter my hifi test folder for speech department. Sounds like the AI enchancer has been taught according to SM 7B. In future we can all sing like Elvis. But they will first cash the AI in a form or another.
I like the Adobe processed sound, but in every case, it sounds like the person speaking is situated inside my ear. You CAN produce sound that bears or reflects “distance” from the listener without echo or hollow room sound. However, when your eye sees the speaker sitting at a desk 9 feet away from the camera, but the voice sounds like the speaker’s mouth is directly beside your ear, there is a mismatch that makes it look like the sound was dubbed.
Thank you very much for all your efforts and super useful information you provide us in Photoshop section. Your job is amazing!! On the other hand, i can't be saying the same for Adobe's ''A.I.'' new feature. There are dozens and dozens of free softwares out there which can do exactly the same. The result is equally bad. There are certain things that can be done and others that simply can't!!! For some folks, that boomy result can be impressive, for others not. Have spent so many hours setting up my mic trying countless combinations and different configurations. It simply can't be done!! If someone believes that using some compressors, noise suppression filters, expanders and noise gate filters can achieve studio recording ''miracles'', i would say...well, keep dreaming. If on the other hand he's happy with the result, i am happy for him!!
was looking for a recommendation for noise cancelling and better mic. Right after hearing this review, kinda go for Adobe Speech Enhancement. Thanks! ❤
If someone were to sing and it was echoing and you would edit the audio recordings with Dolby Atmos it would actually sound really good but that’s only depending on what you’re making your style song but if it’s just someone talking, you don’t really want to do that so it’s gonna sound bad
Definitely cleans it up pretty good, but makes it sound a bit compressed and in some situations it seems to boost the 100-1500 hz band a bit, making things sound a bit nasal. Great tool for cleaning up poor quality video, but not yet a substitute for recording under better conditions if you are searching for commercial quality sound.
I tried it out earlier today before watching this video and the result I got was terrible! There was weird mid-range EQ and compression which lowered the audio quality and there were weird artifacts if you laugh or chuckle. It didn’t know what to do with that. I think it has a loooooong way to go yet.
Just tried this feature. I must admit, the result is amazing. Much better than me applying some EQ, speech enhancer, and de-noise effect on my Davinci Resolve. BUT, it looks like the AI is build on English vocabulary. My audio is in Bahasa Indonesia so, many words are pronounced oddly. Especially those with "R" letter, since Indonesian says R in different way compared to English.
it would be great if you just replayed the part at 0:45 to 1:06 with the speech enhancement. I feel like seem like it would have saved a lot of time. Also, you can compare the exact same audio instead of viewing it as a toggle switch. but overall very informative. thank you!
Seems when the audio has problems it can do something about that. But when it is already quality input it may "overprocess" it so it sounds unatural (in parts). I'd say a valuable tool when it is needed!
Wow! This is like izotope RX for free for people who don’t know how to do it. Really putting me out of a job as someone who helps podcasters clean up recordings.
I tried repairing some bad audio and this tool didn't work well at all. I used Davinci Resolve's audio leveler and repair tools to very high success. It worked better than iZotope RX and faster.
It seems to just lower the highs, and bump the mids and mid-lows. I would have liked to see before and after on a parametric EQ. You accidently didn't create any plosives before testing it, would have been interesting. But what was amazing was taking out the reverberation from the room!
The results I've been getting are uneven at best. It is substituting a quickly constructed AI version of a voice when it has to scrap the actual audio. A lot of times that sounds very unnatural. Also, sometimes the AI adds a voice to certain sounds. Musical instruments are exceptionally bad, as it sometimes actually gives the music a voice. Nonsense words, but still an actual voice. Early days, to be sure. And yes, getting a good recording is always preferred.
The most mindblowing thing in this video is that adobe released something for free
Well nothings for free is it? You upload data. Could be a win-win thou
But adobe have a bunch of free software
@@tommielif exactly win-win lmao
@@alif__ well I was optimistic wasn’t I
It's in beta stage , wait till it's fully released
Dude, you have no idea how much pain and suffering you and adobe removed from my life. I live in an empty home with high ceilings and no insulation or any type of sound proofing. I spend hours trying to make my audio sound descent and it still sounds horrible. I literarily just tested this with a piece I'm editing right now and I'm in tears... thanks so much dude!!
Yay! Happy for you, Brenda 😎
feel free to share some stuff so i can lend an ear
@@PiXimperfect will it work with nvdia Rtx 1650 and nvdia broadcast
You have a closet right?
Still better to make a little vocal booth in a closet or something.
I learn so much from this channel! I subscribed for the art/design tips, but I'm loving these videos that branch out into audio, AI and new software too. Keep up the great work!
Yeah so great, and thoughtful dude for sure must've put link for adobe site.. NOPE. 20 own self marketing sites wtf
Glad you like it, Dan 😎 I love these topics, and sometimes, I am concerned whether the photographers or designers who watch the content would like this. You've put my worries away, thank you.
@@PiXimperfect It didn’t remove background noise completely. I also used a hammer on a pot lid while I was speaking. You can still hear a “clanking” sound in the bg, however it wasn’t very loud. And it turns your voice into a robotic voice.
UA-camrs should use this! There are so many videos out there with terrible sound.
Yes for sure, i use it on my tutorials
youtube has a budget for this
😂really
Yes and no, one thing he makes clear is that it's a great tool for, as you say, terrible sound, but it kinda ruins well recorded one, so it would be useful for UA-camrs with no audio equipment and low skills for audio editing mostly
I edit my audio by myself because there are some frequencies that can't be automated, you have to adjust them manually for a crisp voice... And also it doesn't take that much efforts to edit them manually, there may be some difficulties when you're editing your audio for the first time because you have to find best frequencies for your type of voice, and after that just make a preset of it and you're good to go...
I'm a professional Voiceover artist with a great appreciation of getting it right from the get go when recording professionally (Not so much for UA-cam content)
What a fabulous looking video 👏🏽👏🏽👏🏽👏🏽.
10 out of 10 for video quality.
As for sound quality of the enhancements, I found it all very muffled and lacked the natural sizzle that human voices have.
I think there's way too much made of getting rid of sibilance. Sibilants is a natural part of speech clarity.
I'm not impressed at all with the tools.
As you absolutely rightly stated my friend, you can't beat a treated environment married to a quality microphone used correctly.
Very engaging video.
Thank you for the effort you made in putting it together 👍🏼.
It's a personal preference of course. 🙏🏽
Thank god for this comment! 100% right, sadly it seems like people are going with it not appreciating the beauty of detail and doing things the right way. I can definitely see the use of this in certain situations and for social media but not more than that.
@@99boardwalk Scroll through the comments and you'll see it's the podcasters and UA-camrs who are excited by this AI. A voiceover artist can't use it. Not for commercial, animation or anything that pays over $50
Your channel is always informative and entertaining. You do such a good job keeping the pacing just right, explaining things, testing things, and showing how to use Adobe's tools/products. This blew me away! Thank you for another great video. :)
tip: edit your audio first with a audio editing software and remove some of the noise. If your raw audio is filled with too much noise, Adobe AI will decide your voice as noise and clean that too. Learned this the hard way
What audio editing software would you recommend?
@@manassikdar1 yes, audacity is a good option, it's free, open source and really easy to use
@@manassikdar1 Audacity
@@manassikdar1 Audacity, so many professional UA-camrs/Singers use this software for their voice editing and vocal mastering
Yeah this is super interesting - it almost re-synthesizes voices if the audio is too noisy, which results in people sounding like different people
It's brilliant for what it does. You're 100% right about getting it right at source. It sounds slightly clipped - but a great blanket process.
It‘s not only sound imperfections or noise removal - it’s speech recognition as well. I tried extremely echoey non-English audio. It removed echo but also tried to recognise speech and in the end produced incomprehensible sort of English-sounding file.
Perhaps the entire reason it's free is because they're using it to gain AI speech information.
Oh, that's sad to know :/
Yes, it's not perfect. But I was impressed with most of the audio files I used with it. Exceptions: When people are talking over each other during a discussion, or when talking while there is applause or laughter. It just goes weird. But all in all, a nice tool to have for the times it might help.
I tried it as well. It only works with english. The fact is the AI tries to gather audios from here and there and tries to merge those sounds together with the original to give a "clean audio", but english recognition. I have no doubt they will improve it, but when the time comes they add other languages, they will put a price on it, sadly.
I tried this for Spanish and it works just as awesome as English.
It’s worth knowing for people recently coming to this that Adobe introduced a limit and now you can only enhance one recording per hour - So if you do a test piece of a few seconds just to check what the output would be, you’re going to have to wait another hour before you can upload your track. It’s not a massive problem but it could impact on your workflow if you are up against a tight deadline
Let the micro-transactions begin !
Ha ha, you thought it was really going to be free?
i have been trying to upload a 20mins video for hours now
First couple of tests went great. I still here a little handling and tapping but most of it is removed. I still heard some sibilance. Sibilance is less but the adjustment is a bit abrupt, like not moving the sliders in blend-if far enough apart. Noise is gone. I agree with you about the breath sounding artificial. Breath can be removed manually in many sequencers. Adobe Podcast does a great job overall, I would say.
I haven’t checked, but I would imagine you can fine a free “de-esser” online, to control your extra sibilance before you use the AI tool. Logic X comes with a de-esser. Hope this helps 👍
As an audio engineer i Can definitely say that this works wonders and will raise the bar once again for amateur level creatures
👾👾👾
it needsd some work though the pitch is a bit off
I got a email requested maybe I'll be a chosen 1
it does!! you dont know how i struggled to remove noise from my phone recording for a video i made using audacity.
amateur level creatures 👽
Sound is one of the most important details. If the sound is bad, I don't even bother watching the video
fool
Cool story bro
I am absolutely floored.
This is incredible!
It might not produce top notch audio at the moment but one cannot deny how incredible this is for beginners who can't afford expensive gear!
After all, not everyone will be able to immediately buy a decent microphone when they start out so this will help* loads of people!
Where to find this software
You'll never get top notch audio from source material that is not top notch quality. With good enough processing you can get something that sounds similar to a quality recording from anything, but in the end it'll be an approximation of the source material no matter how good it sounds.
Same thing with DxO PureRAW denoising photos so good you might as well have a 1 or 2 stop better camera in low light!
@@gutterg0d the only thing is this is A.I. so it has the ability to fill in gaps and create whats not there. I agree with you as of now. But A.I. can potentially do some crazy cool shit. This is probably free for now so their machine learning can get better and better. Plus this shit can learn so much it might create vocals from nothing
Yes, sir! Absolutely agree.
While I'm using RX by iZotope on a daily basis (which is a totally advanced and professional software), I must admit that this is a really great tool for everyone who needs some help e.g. with their YT videos, but have no knowledge about the whole audio processing part.
Wouldn't say Rx is that advanced but can be useful for sure.
@@TimberWulfIsHere Whats the more advanced version of RX? Also Ive been using RX for years and have never gotten such clean low & mids like the AI does
@@grantsomething don't think there really is until there is a version with complex ai algorithms comes into play
The best Indian UA-cam channel I have ever seen and the best Photoshop channel in the world. Highly recommend 🔥🔥🔥 and love from Sri Lanka ❤️❤️❤️
The processed one sounded great too but the one for the mic sounded *more natural*
Love the curves and appreciate the attention to detail on the tests. Always a joy to watch and learn. Love the growth of the background and lighting compared to the early days. Nice yluse of color and symmetry.
Thanks for another great video 🙂
I think this is usable and can save some recordings - but it does not replace a good microphone and studio. Just like with computational photography it cant yet replace large sensor, good lenses etc
Thank you so much for this! I started doing vlog's in my car on the way to work and needed something to clean up the road noise. This seems perfect for my use.
The only shortcoming is that, it doesn't work well on non English audios. I've tried with some indian languages and Bengali, most of the time, noise, echo are fixed, but main audio is a gibberish and it isn't understandable anymore. I guess this software uses english speech to text at some point of editing, and maybe the voice models AI uses is in English.
I agree, the Dutchlanguage is also not processed well!
But removing echo was great. Is there another tool that removes echos?
Same thing here for Italian, the result was a made up language with English-like style.
I was wondering how it would handle accents/foreign languages myself. Not entirely surprised it has issues, but that may be why it’s a bit of a demo at this point.
Your comment is definitely highlight most useful comment
08:33 the original is better also because it compressed the tonality in your voice, removing its texture and personality.
However, I have often (even recently) had really bad audio that this may have helped a lot to get something usable.
That’s pretty impressive. It sounds a little muffled and crunchy at some parts but still, pretty good
oh man, this is a next-freaking-level tool. Audio went bad for some of my shorts and I just found this tool through your video right in time. Can't be more thankful today!
tip: edit your audio first with a audio editing software and remove some of the noise. If your raw audio is filled with too much noise, Adobe AI will decide your voice as noise and clean that too. Learned this the hard way
This tool from Adobe is insane, I tried with two audio files from different microphones and compared to the same audio file with treatment on Premiere, and the results are awesome, one of my lav mics was echoing a lot and the processed file removed all echoing ambient sounds which became way better than the other one.
Also you can still apply audio filters on top of the processed file from Adobe, I used the same preset filters I use on my microphones and the audio became even better. Now I'm really thinking of using this tool to my other videos too.
I love that you gave a bunch of real world scenarios and audio test types. Thanks!
Another awesome audio tool that's AI powered and similar to this is called "Krisp", it is currently being used in Discord's noise cancelation system!
Love that you got the curves adjustment layer in there! :D
Somebody needs to run all the voice files for older games with garbage compressed audio through this and then put them back in the game. (Assassin’s Creed 1, Far Cry 2, CoD World at War, etc.)
OMG dude!! This changes everything from now on! Thanks for sharing!
Actually I prefer the processed audio from adobe even with the nice mics in studio mainly because it was softer and more pleasant less harsh. This might be simply because I am listening on iPad speakers but I think with any hi fi headphones I would actually feel stronger about because it would likely sound more intense. I notice that your mixing preferences seem to be a bit sharp for the my tastes so using only your own audio samples might be making it sound better than it actually does because when I strain I can detect a sort of text to speech artificial sound to the processed file. It just sounds better because it’s easier on my ears, if your mixing was more bass forward perhaps I would be more annoyed than relieved.
I hear you, but you can use other methods to take the edge off the sound if that's what you want; EQ, Soothe, McDsp SA-2 etc. The extra degradation to the audio isn't worth it if that's all you're trying to achieve.
@@ChrisPFuchs preach
I learned alot about mics and audio recording from this video. Man you gave us good ear test .
For AI to do this without subband modulation or spectral smearing is pretty impressive. I'd imagine there's a good chance those listening on smart device speakers will even prefer the processed audio to the original.
Really...one of the best channels on YT. Thank you!
Why does the cheap 5$ Amazon mic sound so good 💀
Thank you! I saw Adobe AI on another channel, but they failed to say what the Web address was and, since they also failed to give the proper name of the Website (and Adobe won't let you search for it on their site) I couldn't find the proper page. This is a great help as I have just begun doing audio books and a neighbor with dogs has just moved in. Sigh.... Keep up the great work!
Definitely not a replacement for capturing audio with high quality mics at the correct distances, but if something tragic happens and your audio is ruined for some weird reason, this could be really useful to at least get usable audio. Still I would do a little EQ and mixing to make it sound good but this could be great for removing handling noise and plosives, which are a nightmare to remove.
This is great! Saves so much fiddling about. The result is amazing. Let's hope is rolled into the software as a filter!
One of my favourite UA-cam channels, always look forward to your next one. Another great tutorial.
Tried it and it saved one of my audio that i did not recorded right. Many thanks!
I think it makes a massive difference in all areas EXCEPT the sibilance. I think in every case of sibilance, it makes it worse as it sounds like its boosted mid which gives it a weird chamber sound.
Just as trends happen in graphic design, I feel like the trend for Podcasts at the moment is to have a warmer sound overall, which probably makes it accurate for Adobe AI to be Adobe Podcasts. It's not my personal preference, as you say sometimes it's a bit abrupt and unnatural, taking breaths away etc, but it might suit a podcast sound.
I am currently researching the application of AI, thank you blogger for sharing this will be very useful
OMG, what is happening to this world? We`re witnessing so many mind blowing discoveries that are constantly appearing soooo damn rapidly. AI is developing too fast, but that`s cool anyway 🤙 What else can it be applied to? ;)
HOLY CRAP! This is a savior for me!! I just recently started voice acting, and my mic stands out like a sore thumb in comparison to my fellow VAs mics!
I've been watching you for years but hadn't seen a video of yours in a while. I've always bookmarked old vids for reference. So I'm pleased to see how far you've come in video quality and lighting setup. Your information and the way you teach hasn't changed and has always been king. But the look and flow has shown a very pleasing and study growth matching the Information you share. Just wanted to say congratulations 🎊 and I hope your success continues. You and your team have earned it. Cheers from Sacramento California
sounds amazing! I've been searching for years for something to remove the echo and fix the proximity problem. No need to process high quality audio through the AI. The AI made it sound "over-processed". It still sounds great, but sounds "uncanny valley" in some way.
To me it sounds like it would be a useful tool if you are really struggling with background noise or using an absolutely terrible/broken microphone. But if used on anything even barely passible, I feel like a lot of personality is taken out of the voice. From my own testing, this seems to be especially true for people with higher-pitched or breathy voices.
This video was done excellently by the way. Very comprehensive, well presented, and thorough!
Yeah, I felt the same when trying some of my own audio. The AI made a terrific job, but sometimes it's a bit too agressive and cut out a few sounds that were not noise, but were under a certain thresold level, like a harsh noise gate.
You have one of the best info videos. keep going!
Unmesh, it's wonderful that you have a U87 for your audio but I would recommend getting a pop filter or something similar in front of it. It's very sensitive to moisture and eventually it will start sounding dull. It's a great mic & investment, protect it. ;-)
Which microphone is used in studio?
Tell the model, please.
@@seductioncorner Neumann U87
Thank you Unmesh, it really made a difference. I tried it will you were explaining all about it. And I loved it. It will help me alot!🙏🏻 1 thing though it make somewhat like Artificial like a robotic at the end result but that's its awesome, thank you again.
Pretty cool. Davinci Resolve also has a new speech processing feature that works pretty well for free. I recently botched my lav mic recording and clipped the recording into the camera, and it cleaned it up pretty well. I wouldn't count on it all the time but it can be a lifesaver.
In free version?
@@gawr2765 Yep in the free version!
I've tried this ai. I think if the audio is in English this will give good results. The ai was trying to figure out what my subject was saying and converting it. My subject is a Filipina now she sounds Chinese hehe!
As a sound engineer I'm always confronted with the statement "We'll fix it in the mix.". Although we can perform little big miracles with the technology available today, best would be to do it properly at the time of recording. That of course includes the fact that for instance in music the arrangement (comparable to composition in photography) and the performance need to be of the highest quality possible. Otherwise we're just polishing a t*rd 🙊
Great content as per usual and Adobe's AI audio tools could be handy in some extreme cases here!
Keep it up! ✌🏻
wow, audio gate, gain, roomtone removers, reverb reduction and more with a click. I love it. This will save me a lot of time of mastering audio for my voice acting
You have created an incredibly useful instructive piece! For example, every UA-cam user should see an echo test so that they can record professional soundtracks without paying anything extra. The room echo prevents listening to audio tracks louder, for example, while doing dishes in another room. You bring out all those important points like sibilance. Absolutely amazing work, not to mention the great tool you mentioned, which can do wonders in the right context. This video will immediately enter my hifi test folder for speech department. Sounds like the AI enchancer has been taught according to SM 7B. In future we can all sing like Elvis. But they will first cash the AI in a form or another.
I like the Adobe processed sound, but in every case, it sounds like the person speaking is situated inside my ear. You CAN produce sound that bears or reflects “distance” from the listener without echo or hollow room sound. However, when your eye sees the speaker sitting at a desk 9 feet away from the camera, but the voice sounds like the speaker’s mouth is directly beside your ear, there is a mismatch that makes it look like the sound was dubbed.
I appreciate the effort you put in to create this video! creating something like this takes a hell lot of time!
Omg what a great video 😭😭 I needed this so bad!! Will be using for my next video!!
Why wouldn't it be free? They NEED people to upload a crap load of audio to train their AI for free.
I think it's a good addition from Adobe to help beginners who can't afford a professional microphone. I'm eager to try it out for myself
The way you analyzed the audio, shows you are a great mixing engineer too 😃
probably a trap
Oh wow, just tried it, you have saved me between 6 -18 hours of workload a week. Lifesaver, the audio sounds absolutely fantastic. 🤓
Even in a video about audio, you find a way to talk about curves :)
Thank you very much for all your efforts and super useful information you provide us in Photoshop section. Your job is amazing!! On the other hand, i can't be saying the same for Adobe's ''A.I.'' new feature. There are dozens and dozens of free softwares out there which can do exactly the same. The result is equally bad. There are certain things that can be done and others that simply can't!!! For some folks, that boomy result can be impressive, for others not. Have spent so many hours setting up my mic trying countless combinations and different configurations. It simply can't be done!! If someone believes that using some compressors, noise suppression filters, expanders and noise gate filters can achieve studio recording ''miracles'', i would say...well, keep dreaming. If on the other hand he's happy with the result, i am happy for him!!
was looking for a recommendation for noise cancelling and better mic. Right after hearing this review, kinda go for Adobe Speech Enhancement. Thanks! ❤
Hello from Canada. You are very good at what you do!
If someone were to sing and it was echoing and you would edit the audio recordings with Dolby Atmos it would actually sound really good but that’s only depending on what you’re making your style song but if it’s just someone talking, you don’t really want to do that so it’s gonna sound bad
Damn, the difference is ridiculous! Made some tests here and its jaw dropping!
Definitely cleans it up pretty good, but makes it sound a bit compressed and in some situations it seems to boost the 100-1500 hz band a bit, making things sound a bit nasal. Great tool for cleaning up poor quality video, but not yet a substitute for recording under better conditions if you are searching for commercial quality sound.
I tried it out earlier today before watching this video and the result I got was terrible! There was weird mid-range EQ and compression which lowered the audio quality and there were weird artifacts if you laugh or chuckle. It didn’t know what to do with that. I think it has a loooooong way to go yet.
Just tried this feature. I must admit, the result is amazing. Much better than me applying some EQ, speech enhancer, and de-noise effect on my Davinci Resolve. BUT, it looks like the AI is build on English vocabulary. My audio is in Bahasa Indonesia so, many words are pronounced oddly. Especially those with "R" letter, since Indonesian says R in different way compared to English.
it would be great if you just replayed the part at 0:45 to 1:06 with the speech enhancement. I feel like seem like it would have saved a lot of time. Also, you can compare the exact same audio instead of viewing it as a toggle switch. but overall very informative. thank you!
This is such a game changer! I used it for realistic AI voice clip from Syntheses, and the result is it sounded even MORE realistic! Goddamn!
You, extenuating the sibilants in difference, was hilarious 🤣
Loving the long videos!
Can you do a tutorial of your processing your current mic? Your audio sounds great!. What audio processing do you do/use? Thanks!
I just tested it. This is insanely useful!
Seems when the audio has problems it can do something about that. But when it is already quality input it may "overprocess" it so it sounds unatural (in parts). I'd say a valuable tool when it is needed!
Wow! This is like izotope RX for free for people who don’t know how to do it. Really putting me out of a job as someone who helps podcasters clean up recordings.
Sounds amazing! Do you know of something similar for mastering of music?
I had the same question. Trying to refine (or enhance) a crude musical recording. Let me know if you find out or if someone gets back to you!
Good job dear , Please make a video to show how did you setup your lighting studio ?
Thanks Alot Dude Really Helped Alot❤
Thank you so much brother, already worked, it's very helpful me...
Thanks so much for the heads up!!!
I tried repairing some bad audio and this tool didn't work well at all. I used Davinci Resolve's audio leveler and repair tools to very high success. It worked better than iZotope RX and faster.
DaVinci Resolve is remarkably good. I have found it to be the best myself
I tried this but the problem is it makes someone sound as a robot🙁
I'm all for clean audio, but over-processing seems the norm these days. This tool works great for rubbish in, it messed up the pro-mic though.
wow it's making it so much better . dunno how long it'll be free
AirPods Pro audio now sounds like studio 😂😂 I love AI - thanks for sharing
Wow well tutorial step by step ❤❤
This just changed my life!
Wow, this as a vst would be gold
I definitely hear a difference. This tool is amazing.
Amazing sounds ☺♥♥♪
It seems to just lower the highs, and bump the mids and mid-lows. I would have liked to see before and after on a parametric EQ. You accidently didn't create any plosives before testing it, would have been interesting. But what was amazing was taking out the reverberation from the room!
I love your vids so much! They're so helpful
The results I've been getting are uneven at best. It is substituting a quickly constructed AI version of a voice when it has to scrap the actual audio. A lot of times that sounds very unnatural. Also, sometimes the AI adds a voice to certain sounds. Musical instruments are exceptionally bad, as it sometimes actually gives the music a voice. Nonsense words, but still an actual voice. Early days, to be sure. And yes, getting a good recording is always preferred.
I don't think it's meant for musical instruments haha.
Impressive results!
Amazing! Thank you!