it is adobe… so yea, most likely you are going to pay per syllable, the enhanced version did sound more robotic to me.. so it would need much more work
One thing I have seen is people saying stuff like "oh RTX Voice already does this in real time". But no, they don't get it. RTX Voice will remove noise yes, but any room reverb will remain. RTX Voice even has a beta reverb removal mode, and it doesn't even work well. This Adobe tool can take a basic webcam room mic and all of it's reverb glory and make it sound like a condenser mic in a sound booth. It's insane. Removing background noise is one thing, but I'm not aware of anything else on the market that improves the voice to this degree. It's not perfect and sometimes if there is an insane amount of noise or reverb, the voice will sound really weird. I think it is literally reconstructing the voice from scratch.
I might not have noticed if I didn't know, honestly, but it does sound a bit flat and processed, yes. I am not a picky listener to podcasts - I focus more on the content - so I would not mind personally if you sold your expensive setup and used this instead. It is not the same though. The high-end mike makes your voice sound richer and more natural. But the Adobe output is a big step up from the tinny, distorted sound, absolutely. Thanks for the heads up!
I'd say it's like 80% the way there. It just needs to fix the high frequencies instead of totally lopping them off. But once that's done, it will obviate the need for expensive mics (unless you are doing live streams).
I used a heavily damaged audio, recorded on an old laptop and with the microphone of that laptop. I was impressed with the result after using Adobe Podcast, the audio quality increased a lot!
Hey Matt, I know this is somewhat unrelated, but I was wondering if you've heard of Eleven Labs. Their beta still isn't open, but on their blogpost from last month I listened to the most insanely emotionally expressive TTS I've ever heard.
Your cheap mic doesn't sound that bad really and it sounds like this service is meant for something like your headphone mic while you're doing dishes or something. The way you used voicemeter blew up the high frequencies a bit too much while adobe voice meter reduced them and sort of overall normalized the audio frequencies. This made it sound much better and closer to something like a modern joe or lex podcast. It sounds a bit like you want to go for a more punchy sound like a radio show from around 2005, but this tool works against that a little bit. It does sound great to get a base sound though so you can add some punch after the fact.
Pretty good considering it will be buyable, but iZotope RX10 advanced is better although it's a $1k software, look it up, and maybe make a video about it (it uses AI in even more stuff!). I did notice that Adobe Podcast changed my client's voice slightly (upper tones as you mentioned). I hope it's not a part of the noise the AI recognizes, and that it's fixable in a later version. I enjoy your videos. Thanks Matt.
Davinci Resolve, including the free version, has an AI dialogue isolator now that works better than RX10 in some situations. Same for Final Cut Pro. It seems like 18ish months ago everybody cracked this problem at the same time.
Hi. I am using RX 10 advanced and I’ve been using RX since forever as a professional. How can you say RX is better than this? Please explain! Of course RX is not only for enhancing speech and make bad speech recordings sound better. It’s a lot more powerful and customizable. But only on voice like this this is miles better, doesn’t matter what RX modules you use and how you set them, the result will not be at this level. Curious to hear why you say that. Thank you!
@@BBDRecordsStudio first, Adobe podcast doesn't do much for reverb compared to "dialoge de-reverb". Second, I think using spectral recovery + dialoge isolate + EQ match might produce similar result. I'm not sure though for this particular audio, I'm not an "audio expert". For my client, I'm sure at the current state of adobe podcast it doesn't sound as natural, and if I were to depend on it my client would surely notice/complain.
@@DaCarnival Didn't know they released this. I tried it and yes, it is indeed comparable if not better. However, I still can't give up RX, has a lot more stuff. I need EQ match and de-reverb mostly.
@@Masterx3 I agree that de-Reverb is better in RX. However it Really depends on how bad your audio is. For example if it’s recorded far away from the mic, spectral recovery won’t make the voice sound close to the mic as this AI can. But yeah, it sounds a bit unnatural to be honest, sometimes it sounds like 80% the real voice and 20% some generated vocal speaking.
Hi Matt... just came across your channel. Great presentation on the adobe voice tool. Have a rather new Aerial video drone business, which has led to more and more requests to do handheld video, interviews, video podcasts, etc. Have Gh5 and a few decent on camera shotgun mics. Starting to look into tools for better audio and voice for my videos. Question... have you played around with Decript? Appears they have a decent AI voice enhancer as well. I'd love to get your take on that tool. Good luck in 2023!
not bad but also not that impressive. it mainly cuts out the high frequencies and does some equalizing like you already mentioned because those are what seperates a good mic from a bad mic. however a good mic doesnt cut off high frequencies it just records them with a way better resolution and if the ai would want to replicate that it would need to add frequencies that havnt been physically recorded in the first place I cant imagine that this would be possible (yet - we know ai is always up for a surprise)
It's kind of like an (IR) impluse response, like digital guitar rigs use for cabinet sound shaping. Its kinda works, but i think you would get the same results with a decent EQ plugin. It doesn't really change the pickup of the mic as its really the capsule that does that.
Nvidia Broadcast does an incredible job removing background noise live using AI, gets an incredible result without having to upload any audio. Nvidia tends to silently update the tool as well, and I think it performs about the same. The tool has both noise removal and background echo removal.
The tool from Adobe not only removes background noise but also does heavy work with equalization, voice synthesis, etc. It's probably a GAN designed for enhancing voice recordings.
@@---pp7tq only as noise cancellation not as voice enhancement (for noise enhancements it requires a card with the new core type that are introduced with the RTX cards)
having the Ai add some reverb or distortion and even over dubs would be sick, and what's really sick is that we're most likely going to see that in the near future, and that's super exciting!
@@JohannesLabusch yes you mean to tell me you've never heard slight distortion on a chorus for effects or throughout the track? As a performer myself we've used the distortion effects, creatively, to get a specific effect. I'm not saying you just add random distortion just to do it. I'm all about taking these new tools and exploring creative uses for them pushing their abilities to create a new style of artistry, feels like so many are pushing back but in the same breath limiting themselves, I'm more confused by that.
Chances are you've already heard songs with distortion and just don't realize it but pretty much you're adding anything to The voice like Auto-Tune or a fax you're distorting the voice so yes distortion ua-cam.com/video/3hg4plSXhBQ/v-deo.html
Problem with this program is that breaks if you try to sing or get louder all of a sudden (you know how youtubers can be dramatic for the screen). I use it as an alternative for Descript for the Voice Ai thing. Tho i enjoy the garbled burp noises that come out sometimes.
Thanks for the video, but are you aware that your high frequency audio is through the roof? Every S you say, even with the SM7B, is tearing my head off. If you are using those Arctic headset things as quality control, then i can understand why.
It is but this seems to build in frequencies that Davinci misses. I ran a comparison today on a soundtrack I'm working with (admittedly well recorded audio) and the combo of running Davinci NoiseR/ Leveler into Adobe. Just. Wow.
It doesn't sound like it's making your CAD mic sound better, just removing noise. I'm not saying there is no use for it. More useful, IMO, to help a recording that has background noise, like a fan, or diminish the recording of sound reflections. But, losing all the high end in your recording isn't going to make your mic sound more expensive. It does make it sound like you recorded with better technique, or in a quiet room. SM7-B's are nice, but you are also not ever using it w/o a pre-amp and your 'lifter'. If you were to take that same USB mic audio recording and run it through your pre-amp, I bet you would have heard as much (maybe more) improvement than the Adobe app. For $30 you could also get a Waves noise reduction plug-in, that also can be used live. I still feel like it cuts some of your highs, but you can adjust the levels to your taste. There are a bunch of 'in the box' pre-amps that you could use, some of which you can find online for free. Several companies have mic emulation plug-ins that can do a better job than this of making your cheap mic sound more like an expensive mic. I would recommend a company name Audio Acustica's LAVA. If you go to their site, you can try it for free for at least 1 month. But I guess that doesn't fit your motif of AI related topics! I think CAD generally makes pretty good quality mics for the price, BTW.
Your cheap mic sounded quite bad to me (tinny sound), the enhancement was much better but kind of flat. I guess that the professional sound is geared towards the producer's preference since I watch a channel produced professionally (redacted) where I think the bass is way too emphasized. You said that you like to keep the high-end sound for your recordings but I get sharp peaks which I would trim down a bit. Anyway, you sound better than me so it is ok.
If you have a RTX Nvidia GPU you can check out Nvidia Broadcast. It Is AI powered and removes all unwanted background noise. I have been using it since it was called RTX voice. Its great.
It’s nice to give people some ideas on how to improve their audio. That said, it’s important to explain the physical limits here. Anything added on electronically is fundamentally different from getting it right at the source. You can’t coax something out of a signal that’s not in there in the first place. If a limited microphone picks up a source, those limits are baked into the result. Now, I have no doubt that technology keeps improving, but I hope people don’t think you can polish a turd until it becomes a nugget. I understand how you'd decide on a nice click bait title, but yes, you DO need a good mic. Nowadays, that dream can become reality for around a hundred bucks. You can come A LOT closer to a Shure SM7B with a 100$ Shure SM57 and some EQ skills. Why fiddle with synthetic enhancements that steal your time and never give you the same result?
Well ai can polish a turd until it becomes a nugget (have you seen img2img AI's turning ugly drawings into photorealistic images!?). It just won't be very accurate or true to life. It will fill in details based on statistical probability. So it will 'make up' the details that aren't there at source. I have a feeling it can become fairly good at this when it comes to vocals. It doesn't seem the hardest job to predict harmonics etc. out of range of the microphone based on what the mic CAN pick up. It will probably have trouble with more unique noises and sounds.
@@nathanpersonal You clearly know more about this stuff than I do. I just think it's a misguided road to go down. Everything is getting more synthetic and artificial. Many people seem to have lost the ability to tell a natural acoustic guitar sound from the plastic-y honk of piezo pick ups, for example. Why is that? Well, they've listened to major stars with shit acoustic sound on stage, for decades. So now that has become, in the majority's mind, what an acoustic guitar sounds like. I wouldn't be surprised if something similar happens to what we feel is an accurate recording of the human voice.
@@JohannesLabusch Given that we’re all watching UA-cam videos through the “lens” of many different types of microphones, I don’t think that this voice enhancement really changes how we perceive voice to any real degree. Even in video itself, many techniques are used by cameras and compression algorithms to get the end result of what we watch. Everything digital has been changed to better make do with the limited hardware we have. Drawing an arbitrary line to separate “real” media from “synthetic” media when it’s all synthetic anyways is just plain weird.
@@diophantine1598 Interesting take. My beef is not with digitally altering or enhancing audio. My beef is with claims like "You Don't Need a Good Mic". I don't think this is a serious claim.
@@JohannesLabusch Though you kind of don’t? I agree it would always be preferable to have a better mic, but in the same way we don’t need as good of a GPU for games because of FSR and DLSS, we are enabled to not have nearly as great a need of a good mic through AI tech. You can’t make something out of nothing, but if most of all the data needed is there, AI can just fill in the gaps the mic can’t.
Why would you present anything, VoiceMeter here, with so much dead space around it? You should look at your analytics and see how many of us are using smartphones. Now I understand the new pinch to zoom UA-cam added...
Once you can use AI to clone your voice they can either just use that from zero to recreate the recording, or use it to help remake the "professional sounding recording"
yeah seems impressive but its not, thats old tech. Adobe has been behind in AI as if they are trying to catch up instead of competing and innovating. Take a look at photoshop and their auto upscaling: not nearly as good as some websites and github repos ive used. Adobe only offers uploading audio instead of live transcription? Please tell me im wrong, because that is sad. AI is advancing in every aspect/use case RAPIDLY everywhere else so whats going on?
All the software in the world won't be able to replace a great preamp with a great compressor and great equalizer. But then again, we'll all be listening in on our Bluetooth ear buds. Software can't emulate tube warmth.
Garbage in garbage out (fancy garbage out in this case). If you can than get a good mic: You really wanna record more details as possible of what makes you unique and not level things out by using AI imho.
Great demo , a lot of work there. I don’t like the sound of the untreated cheap microphone but I like the sound of it less , actually after assessing the Ai Adobe treatment. I prefer the treatment achieved by the first app you demoed and currently use. The Ai conversion sounds dull,over flat and more distant rather like the effect I often achieve when I get my EQing wrong. It’s a bit like the effect of the old Dolby technology . Factually any basic EQ is going to improve the sound quality for any Podcaster even with budget mics. More up market solutions such as Fab. Filter’s Pro Q or the Gullfoss VST at this point in time dramatically out perform the Beta Ai in my opinion and ears. 🙏🙏🏿🎩
Close your eyes and imagine if you will ... Davinci Resolve 2028? All of this AI stuff will be built right in. If you think it feels like magic now ...
It’s how the headset works, you can’t see it but there is an elastic cloth that stretches across the top of my head and thats what my head touches. It’s a steel series arctis pro
Yeah. And that's also why everyone buys a photoshop license instead of building muscle on the gym, or buy a autotune plugin instead of going to singing classes, right? You can't make it better if you don't already start with something decent, you re just putting lipstick on a pig.
Try it for yourself! Let me know how it goes! I hope there is a FREE option once this is released
it is adobe… so yea, most likely you are going to pay per syllable, the enhanced version did sound more robotic to me.. so it would need much more work
One thing I have seen is people saying stuff like "oh RTX Voice already does this in real time". But no, they don't get it. RTX Voice will remove noise yes, but any room reverb will remain. RTX Voice even has a beta reverb removal mode, and it doesn't even work well. This Adobe tool can take a basic webcam room mic and all of it's reverb glory and make it sound like a condenser mic in a sound booth. It's insane. Removing background noise is one thing, but I'm not aware of anything else on the market that improves the voice to this degree. It's not perfect and sometimes if there is an insane amount of noise or reverb, the voice will sound really weird. I think it is literally reconstructing the voice from scratch.
yo thiojoe is here?
does this do it live?
@@DOTFilthy no
Matt always out here with the best AI news! TY
Thank You!
I might not have noticed if I didn't know, honestly, but it does sound a bit flat and processed, yes. I am not a picky listener to podcasts - I focus more on the content - so I would not mind personally if you sold your expensive setup and used this instead. It is not the same though. The high-end mike makes your voice sound richer and more natural. But the Adobe output is a big step up from the tinny, distorted sound, absolutely. Thanks for the heads up!
I'd say it's like 80% the way there. It just needs to fix the high frequencies instead of totally lopping them off. But once that's done, it will obviate the need for expensive mics (unless you are doing live streams).
Exactly what I needed! Thanks for sharing man :)
I used a heavily damaged audio, recorded on an old laptop and with the microphone of that laptop. I was impressed with the result after using Adobe Podcast, the audio quality increased a lot!
Hey Matt, I know this is somewhat unrelated, but I was wondering if you've heard of Eleven Labs. Their beta still isn't open, but on their blogpost from last month I listened to the most insanely emotionally expressive TTS I've ever heard.
Thanks so much mate! I've been waiting desperately for software with this type of audio enhancement.
Same here tbh! Happy to show it to you!
@@MattVidPro 🍻
Your cheap mic doesn't sound that bad really and it sounds like this service is meant for something like your headphone mic while you're doing dishes or something. The way you used voicemeter blew up the high frequencies a bit too much while adobe voice meter reduced them and sort of overall normalized the audio frequencies. This made it sound much better and closer to something like a modern joe or lex podcast.
It sounds a bit like you want to go for a more punchy sound like a radio show from around 2005, but this tool works against that a little bit. It does sound great to get a base sound though so you can add some punch after the fact.
Thanks Matt. i needed this now.
i just tested it out and, it made my 20 dollar headphone mic audio actually sound pretty good, great vid
Pretty good considering it will be buyable, but iZotope RX10 advanced is better although it's a $1k software, look it up, and maybe make a video about it (it uses AI in even more stuff!). I did notice that Adobe Podcast changed my client's voice slightly (upper tones as you mentioned). I hope it's not a part of the noise the AI recognizes, and that it's fixable in a later version. I enjoy your videos. Thanks Matt.
Davinci Resolve, including the free version, has an AI dialogue isolator now that works better than RX10 in some situations. Same for Final Cut Pro. It seems like 18ish months ago everybody cracked this problem at the same time.
Hi. I am using RX 10 advanced and I’ve been using RX since forever as a professional. How can you say RX is better than this? Please explain! Of course RX is not only for enhancing speech and make bad speech recordings sound better. It’s a lot more powerful and customizable. But only on voice like this this is miles better, doesn’t matter what RX modules you use and how you set them, the result will not be at this level. Curious to hear why you say that. Thank you!
@@BBDRecordsStudio first, Adobe podcast doesn't do much for reverb compared to "dialoge de-reverb". Second, I think using spectral recovery + dialoge isolate + EQ match might produce similar result. I'm not sure though for this particular audio, I'm not an "audio expert". For my client, I'm sure at the current state of adobe podcast it doesn't sound as natural, and if I were to depend on it my client would surely notice/complain.
@@DaCarnival Didn't know they released this. I tried it and yes, it is indeed comparable if not better. However, I still can't give up RX, has a lot more stuff. I need EQ match and de-reverb mostly.
@@Masterx3 I agree that de-Reverb is better in RX. However it Really depends on how bad your audio is. For example if it’s recorded far away from the mic, spectral recovery won’t make the voice sound close to the mic as this AI can. But yeah, it sounds a bit unnatural to be honest, sometimes it sounds like 80% the real voice and 20% some generated vocal speaking.
Would be nice to see what it does for your nice mic
Hi Matt... just came across your channel. Great presentation on the adobe voice tool. Have a rather new Aerial video drone business, which has led to more and more requests to do handheld video, interviews, video podcasts, etc. Have Gh5 and a few decent on camera shotgun mics. Starting to look into tools for better audio and voice for my videos. Question... have you played around with Decript? Appears they have a decent AI voice enhancer as well. I'd love to get your take on that tool. Good luck in 2023!
not bad but also not that impressive. it mainly cuts out the high frequencies and does some equalizing like you already mentioned because those are what seperates a good mic from a bad mic. however a good mic doesnt cut off high frequencies it just records them with a way better resolution and if the ai would want to replicate that it would need to add frequencies that havnt been physically recorded in the first place I cant imagine that this would be possible (yet - we know ai is always up for a surprise)
i think you should have done a mobile record voice. that woulod show the contrast well for this video and clarity
It's kind of like an (IR) impluse response, like digital guitar rigs use for cabinet sound shaping. Its kinda works, but i think you would get the same results with a decent EQ plugin. It doesn't really change the pickup of the mic as its really the capsule that does that.
It denoises, removes reverb, sibilance, and levels the audio as well as EQ.
It cuts high frequencies because noise removal works like this. The most hiss is in the high frequencies so it cut it but other sounds too a little.
Do these programs need a computer? Is it web/cloud based? I am trying to use this on mobile..I use a IPad Pro..I got sick of viruses on computers..
ahhhh the REAL test was to check how much better your best setup improved.... why did you not?!
Nvidia Broadcast does an incredible job removing background noise live using AI, gets an incredible result without having to upload any audio. Nvidia tends to silently update the tool as well, and I think it performs about the same. The tool has both noise removal and background echo removal.
only on rtx cards though … on anything lower it is basically the same as discord’s krisp noise canceling
The tool from Adobe not only removes background noise but also does heavy work with equalization, voice synthesis, etc. It's probably a GAN designed for enhancing voice recordings.
@@kuromiLayfe RTX Voice works also for GTX 1XXX as well.
@@---pp7tq only as noise cancellation not as voice enhancement (for noise enhancements it requires a card with the new core type that are introduced with the RTX cards)
having the Ai add some reverb or distortion and even over dubs would be sick, and what's really sick is that we're most likely going to see that in the near future, and that's super exciting!
Huh? Why on Earth would you want to add distortion???
@@JohannesLabusch artist use distortion in creative ways, not saying just to add random distortion. It's been used creatively pretty extensively
@@stylezmorales On voice? 🤔 In the rare cases where that is done, I don't see the need to have it done by AI.
@@JohannesLabusch yes you mean to tell me you've never heard slight distortion on a chorus for effects or throughout the track? As a performer myself we've used the distortion effects, creatively, to get a specific effect. I'm not saying you just add random distortion just to do it. I'm all about taking these new tools and exploring creative uses for them pushing their abilities to create a new style of artistry, feels like so many are pushing back but in the same breath limiting themselves, I'm more confused by that.
Chances are you've already heard songs with distortion and just don't realize it but pretty much you're adding anything to The voice like Auto-Tune or a fax you're distorting the voice so yes distortion ua-cam.com/video/3hg4plSXhBQ/v-deo.html
How can I use enhancer but for Phone Microphone???
the CAD M37 sounds pretty decent which is the XLR version of the U37.
anyway to use this with obs?
Check out the Audio Repair > Studio Sound in Descript. Adobe seems be aggressively going after Descript.
Jackson was recorded in SM7. Not SM7B. FYI.
Been waiting to try it out, would you say that it's better than Studio Sounds from Descript?
I would say it sounds better
Thanks for sharing this, Matt! I didn't even know Adobe had a tool like this. I'll definitely have to check it out.
Can it remove the mouse and keyboard sounds that my omnidirectional lav mic picks up or we're not there yet?
Haven't tested it.. but my intuition says yes
The Nvidia & AMD voice filtering both do a great job of excluding keyboard noise
MJ’s “Bad” was actually recorded with an original SM7 !
The SM7B is a remodel. The original SM7’s are vintage and cost much more.
Apparently davinci resolve now has a free ai voice cleanup tool
I wonder if you tried out Steelseries GG's Sonar. Pretty good competition as for Voicemeter for realtime microphone sound enhancer.
sounded kinda muffled to me, removed the crispiness. Im sure it will soon get as good as pro tho
Problem with this program is that breaks if you try to sing or get louder all of a sudden (you know how youtubers can be dramatic for the screen). I use it as an alternative for Descript for the Voice Ai thing. Tho i enjoy the garbled burp noises that come out sometimes.
Thankyou for bringing this to our attention. I have an Adobe subscription, and this would be excellent for clearing up all manner of recorded audio.
Really awesome video! Thanks!
Your audio is great! thanks so much for sharing!
Thanks for the video, but are you aware that your high frequency audio is through the roof? Every S you say, even with the SM7B, is tearing my head off. If you are using those Arctic headset things as quality control, then i can understand why.
Can i do it realtime? On twitch stream, discord etc? if not can you guys pls help me ?
Davinci AI noise removal is really good too.
It is but this seems to build in frequencies that Davinci misses. I ran a comparison today on a soundtrack I'm working with (admittedly well recorded audio) and the combo of running Davinci NoiseR/ Leveler into Adobe. Just. Wow.
@@andymcquade Yeah. Adobe make an extra enhancement that Davinci didn't add yet.
@@cesar4729 Yet...(hopefully)
try the new Voice Isolation & Dialogue Leveler in Resolve Davinci studio 18.1 and compare to this
Stable diffusion v2?
pBrilliant pInsight! 👍 Thanks for the peek inside. 🤝 Super informative and useful. 🙌
Mike Russell has great audio and voice tips!
I don't remember subscribing to you but I must of done at some point. I enjoyed the video.
Thank you for watching! I used to make minecraft videos so.. it was probably from that
@@MattVidPro probably.
You really should do a segment in a turtle neck and sued jacket because you are the spitting image of young Carl Sagan!
I've gotten this before! but I personally don't see it.. LOL
@@MattVidPro You're not supposed to. It's like hearing your own voice for the first time. Trust me - I know what I'm projecting on you. lol
try NVIDIA broadcast together with voice meter
It doesn't sound like it's making your CAD mic sound better, just removing noise. I'm not saying there is no use for it. More useful, IMO, to help a recording that has background noise, like a fan, or diminish the recording of sound reflections. But, losing all the high end in your recording isn't going to make your mic sound more expensive. It does make it sound like you recorded with better technique, or in a quiet room. SM7-B's are nice, but you are also not ever using it w/o a pre-amp and your 'lifter'. If you were to take that same USB mic audio recording and run it through your pre-amp, I bet you would have heard as much (maybe more) improvement than the Adobe app. For $30 you could also get a Waves noise reduction plug-in, that also can be used live. I still feel like it cuts some of your highs, but you can adjust the levels to your taste. There are a bunch of 'in the box' pre-amps that you could use, some of which you can find online for free. Several companies have mic emulation plug-ins that can do a better job than this of making your cheap mic sound more like an expensive mic. I would recommend a company name Audio Acustica's LAVA. If you go to their site, you can try it for free for at least 1 month. But I guess that doesn't fit your motif of AI related topics!
I think CAD generally makes pretty good quality mics for the price, BTW.
Your cheap mic sounded quite bad to me (tinny sound), the enhancement was much better but kind of flat. I guess that the professional sound is geared towards the producer's preference since I watch a channel produced professionally (redacted) where I think the bass is way too emphasized. You said that you like to keep the high-end sound for your recordings but I get sharp peaks which I would trim down a bit. Anyway, you sound better than me so it is ok.
I have no idea how you produce so much quality content while being a college student.
If you have a RTX Nvidia GPU you can check out Nvidia Broadcast. It Is AI powered and removes all unwanted background noise. I have been using it since it was called RTX voice. Its great.
It’s nice to give people some ideas on how to improve their audio. That said, it’s important to explain the physical limits here. Anything added on electronically is fundamentally different from getting it right at the source. You can’t coax something out of a signal that’s not in there in the first place. If a limited microphone picks up a source, those limits are baked into the result. Now, I have no doubt that technology keeps improving, but I hope people don’t think you can polish a turd until it becomes a nugget. I understand how you'd decide on a nice click bait title, but yes, you DO need a good mic. Nowadays, that dream can become reality for around a hundred bucks. You can come A LOT closer to a Shure SM7B with a 100$ Shure SM57 and some EQ skills. Why fiddle with synthetic enhancements that steal your time and never give you the same result?
Well ai can polish a turd until it becomes a nugget (have you seen img2img AI's turning ugly drawings into photorealistic images!?). It just won't be very accurate or true to life. It will fill in details based on statistical probability. So it will 'make up' the details that aren't there at source. I have a feeling it can become fairly good at this when it comes to vocals. It doesn't seem the hardest job to predict harmonics etc. out of range of the microphone based on what the mic CAN pick up. It will probably have trouble with more unique noises and sounds.
@@nathanpersonal You clearly know more about this stuff than I do. I just think it's a misguided road to go down. Everything is getting more synthetic and artificial. Many people seem to have lost the ability to tell a natural acoustic guitar sound from the plastic-y honk of piezo pick ups, for example. Why is that? Well, they've listened to major stars with shit acoustic sound on stage, for decades. So now that has become, in the majority's mind, what an acoustic guitar sounds like. I wouldn't be surprised if something similar happens to what we feel is an accurate recording of the human voice.
@@JohannesLabusch Given that we’re all watching UA-cam videos through the “lens” of many different types of microphones, I don’t think that this voice enhancement really changes how we perceive voice to any real degree. Even in video itself, many techniques are used by cameras and compression algorithms to get the end result of what we watch. Everything digital has been changed to better make do with the limited hardware we have. Drawing an arbitrary line to separate “real” media from “synthetic” media when it’s all synthetic anyways is just plain weird.
@@diophantine1598 Interesting take. My beef is not with digitally altering or enhancing audio. My beef is with claims like "You Don't Need a Good Mic". I don't think this is a serious claim.
@@JohannesLabusch Though you kind of don’t? I agree it would always be preferable to have a better mic, but in the same way we don’t need as good of a GPU for games because of FSR and DLSS, we are enabled to not have nearly as great a need of a good mic through AI tech. You can’t make something out of nothing, but if most of all the data needed is there, AI can just fill in the gaps the mic can’t.
Why would you present anything, VoiceMeter here, with so much dead space around it?
You should look at your analytics and see how many of us are using smartphones.
Now I understand the new pinch to zoom UA-cam added...
Considering it's Adobe it will problably be cheaper to buy a good microphone instead of that software.
The adobe software is free actually
KEEP IT UP MATT!
THANK YOU!
Its called compression & eq and its been a thing for decades
the audio in this video sounds like it could really use a D'esser
Once you can use AI to clone your voice they can either just use that from zero to recreate the recording, or use it to help remake the "professional sounding recording"
I think voice actors are next on the chopping block. imagine just typing and have it read exactly as you want.
@@p5rsona sentiment notes on text are kind of ready to use
Lets drop our audiophile onto the website.
The thing that's missing is low end - sounds less interesting.
i was waiting on you to use the headset mic
yeah seems impressive but its not, thats old tech. Adobe has been behind in AI as if they are trying to catch up instead of competing and innovating. Take a look at photoshop and their auto upscaling: not nearly as good as some websites and github repos ive used. Adobe only offers uploading audio instead of live transcription? Please tell me im wrong, because that is sad. AI is advancing in every aspect/use case RAPIDLY everywhere else so whats going on?
You could have just added warmth to the old one instead of buying all that .
The Ai is helping very much now a days
All the software in the world won't be able to replace a great preamp with a great compressor and great equalizer. But then again, we'll all be listening in on our Bluetooth ear buds. Software can't emulate tube warmth.
Garbage in garbage out (fancy garbage out in this case). If you can than get a good mic: You really wanna record more details as possible of what makes you unique and not level things out by using AI imho.
Hardware market is gonna crash hard. From audio to literally musical instruments. In no time you would be able to Gibson Les Paul a potato.
My idea of shit mic is an old headset, but with 7 post process steps its usable for videos
But thats not new. There are tools like this in Audio for years....
Yes but they don’t work as clean as this. It removed noise even when someone is speaking
“You’re a garbage microphone”
‘You mean a lot to me”
😅
the real question is does this work with a garbage $20 headset?
worth a test! another vid on this is coming soon!
Voice goes from sad to Chad!
Holy shit
Great demo , a lot of work there. I don’t like the sound of the untreated cheap microphone but I like the sound of it less , actually after assessing the Ai Adobe treatment. I prefer the treatment achieved by the first app you demoed and currently use. The Ai conversion sounds dull,over flat and more distant rather like the effect I often achieve when I get my EQing wrong. It’s a bit like the effect of the old Dolby technology . Factually any basic EQ is going to improve the sound quality for any Podcaster even with budget mics. More up market solutions such as Fab. Filter’s Pro Q or the Gullfoss VST at this point in time dramatically out perform the Beta Ai in my opinion and ears. 🙏🙏🏿🎩
hey matt
Heyo!
part of me really thinks this is cool, but my other side is going"fucking hell adoby is garbadge with how they dont let you own shit"
Davinci have a similar AI. Pretty good.
Audio uncanny valley
Close your eyes and imagine if you will ... Davinci Resolve 2028? All of this AI stuff will be built right in. If you think it feels like magic now ...
placebo effect
Totally muffled. High end loss. I would NOT use it until it improves.
Adobe need to keep their grubby paws away from ai
meh,not that good,almost the same results as a normal compressor or noise reduction effect
Audacity might not be the best to your shure mic hahaha
you really need to put some deessing on your audio
adobe competing with izotope lol
Bro why is your headset floating above your head
Push that shit down
It’s how the headset works, you can’t see it but there is an elastic cloth that stretches across the top of my head and thats what my head touches. It’s a steel series arctis pro
@@MattVidPro Weird headset for sure
Yeah. And that's also why everyone buys a photoshop license instead of building muscle on the gym, or buy a autotune plugin instead of going to singing classes, right? You can't make it better if you don't already start with something decent, you re just putting lipstick on a pig.
The ai enhanced sounds worse. This software is hot garbage.