This is one of those Ai, when you say Expectation and Reality. It seems like it just scramble the voice with revurbs and other filters. Video 5+, Ai -1
No, this is basically the same as a AI voice changer. They just trained it with instruments instead of other peoples voices. But in both cases you need to have some skills to make it work.
@@missoats8731Your absolutely right. This is definitely NOT procedural manipulation. He should have tried a recorder or keyboard into it. I already subscribed to this and got mind-blowing results Anyone who can actually hit notes WILL get fantastic results. I'm not bashing MattVid!! You are awesome. But any minor hobbyist in music production can see the problem (and potential!) here.
@@ShitMental Another aspect to consider is that you have to make sounds that are at least somewhat possible to recreate with the desired instrument. For example, Matt was way too low with his voice for some of the instruments. If you do it like that, the AI will not know what to do, since it simply doesn't know such a sound. It's the same problem with human voices. That's why men should artificially pitch their voices up when they use a female singer model and vice versa.
@@missoats8731 This is my experience with these kinds of tools also, especially the voice part of this site. I can't just swap out a male for female vocal without transposing up 5 semitones. Also, I don't think true realtime coversion is entirely possible, as the ai probably needs a bunch of samples to preempt the note properly. Anyone who's used things like harmoniser peddles, which are far simpler and procedural knows it's never 100% tight real-time. (who knows what's coming though!). All this stuff is so fascinating, I could talk peoples heads off with the potential and just guessing what's coming next, or even inevitable makes my head spin!
Now we just need a second ai to "tune" your spoken or sung notes before converting to instrument so that you don't get an untuned instrument track back as a result! lol
Adobe has a pretty decent AI audio Enhancer I usually run my AI voices and text to audio stuff through that then I use the fair light tab in davinci, it's not perfect per se but you can def clean it up a lot. Between denoising and dehissing with a little compression then finally a tiny bit of reverb and I can get sometime palatable 😊
@@AvizStudio I have used it for regular audio before, like like they did switch it up now it just says voice but originally you could use it for any audio. I'm sure there are plenty of audio enhancers out there that can help clean it up a little and if you're familiar with DaVinci Resolve you could always use the fairlight table to tweak it even more
Just a tip…it’s going to copy your pitches so if you’re not in a tune you will get a not so desirable result. Try using audio files of some acapellas in key and then you’d get a better gauge of this tool in my opinion. Love this channel and always learning a lot. Peace and love guys 🙏
Yep - I saw the original video and I gotto say, it's a scam, because it makes all of the instruments sound perfect (I suspect they used real instruments for that 'demo'). This video shows what it really sounds like and I gotto say - it's not great, at all. I do think, however, that it should be possible to use voice to dictate what an instrument sounds like - just obviously not this app - perhaps another developer will design something along those lines in the future.
@@realitycheck1086 No. The original guy did have an actual singing technic or at least knew how to sing in pitch and a key unlike this guy. No hate on this guy.
I saw that video on YT long time ago (in AI world long time ago). The original video is from Emmet Ozcan, he is an incredible Artist, and he can do some crazy beatbox/singing. That is why that demo worked so good!
Yeah, I got the feeling that the original vocals/instrument has to sound close to the instrument you are converting it into in order for it to work properly.
It sounds like it is putting filter effects on the voice, rather than replacing the voice with instruments, because I believe I can hear your voice coming through. The "master" seems to just apply a stronger filter.
Clearly that’s what it sounds like it’s happening when Matt does it but it’s more than that obviously: Unless the original demo faked it it’s not effects
He sang in pitch that the instrument he was trying to mimic could not play in. Also, he needs to learn what the instrument he's trying to get atta this AI actually sounds like. He said an acoustic guitar yet proceeded to sing sustained notes with bends, which just is not what acoustic guitars do. If you can get that right, this actually does create convincing instrument sounds, just like in the original demo.
You need to take 2 points into account while you compare yourself to the original demo: 1. In the demo there are some post effects like reverb. It written there. 2. You need to have good rhythm and some musical talent to make the sound close to what you want like any voice conversion process
Yeah seriously. the original demo guy did make sounds that mimic the characteristics of the instrument he tries to get with his voice, where as this guy doesn't know how to sing, nor does he follow the characteristics of the sound of the instrument he tries to get. You have to have that down. Then it creates pretty convincing instrument sounds just like in the original demo.
I've seen this guy's full video, and it's actually 10-15 minutes. He explains how he got the sound he did. Part of the reason you aren't getting the same quality as the guy in the demo video is that he was running the audio into a DAW to process the raw track (adding minor pitch correction), then adding effects in the DAW to the audio this tool produced. It sounded that good mainly because he's an actual audio engineer. This tool is more meant for people who already know what they're doing in a studio environment and want to work out ideas they hear in their heads without having to hire session musicians. :) Love your content man, and really appreciate the deep dives. Keep doing what you're doing!
Lol, Matt, this vid was hilarious! I was having dinner while watching and it almost made me spit my food out. As a musician, I’ve got to point out that one of the main reasons why it didn’t convert your voice as great as you’d like it to was because most of the time, you were singing in the same pitch or within the same few pitches. Kind of like monotone. But it’d work great if you made a big difference between the notes/pitches you were trying to sing. Maybe give this tool a try again in the future and try your best to sing an actual existing song, maybe you will notice the huge difference. 😊 cheers!
This is such a good idea! With some refinement, I could really see this taking off! I'll have to check it out when it's a little more refined. Great video!
Definitely seems like either the demo they used were heavily cherry picked, or there is actually some fine-tuning going on with the voices of the people in the demos. What was shown here was substantially worse than what was floating around Twitter for sure.
Bro, remember that this thing transforms your voice into instruments, Matt put zero effort into making a proper melody, that's why the result is shit. This could be the best AI, but if you don't put any effort into making a sound that's coherent with the chosen instrument, it will sound much better. I'm impressed
No. the original demo guy did make sounds that mimic the characteristics of the instrument he tries to get with his voice, where as this guy doesn't know how to sing, nor does he follow the characteristics of the sound of the instrument he tries to get. You have to have that down. Then it creates pretty convincing instrument sounds just like in the original demo.
This sounds absolutely nothing like the ad they showed. Actually, this sounds exactly like RVC but trained on dry indtruments. The voice always comes through. They charge you for this cr@p?
At the moment it's pretty much oriented to people with good capabilities to emulate the components of an instrument, not for the average joe until it's able to just translate the intention of the rough notes and smartly recalculate the whole thing to the actual instrument, like using midi sampled libraries.
This might be very heplful for me in my case. There's a song that I've had stuck in my head since childhood. I only remember the melody. With this tool, I can actually create the melody and hopefully use that as a way to find that song.
I tried with actual sung vocals and it couldn’t handle the volume. Came out weak and clipping. Would have been cool to have an instrument accompaniment to my vocals.
As a music producer for over 29 years and musician for over 40 years...this will be a game changer in the way i can express my artform. The guitar AI demo at rhe beginning was able to interpret intonation and expression and that makes for some next level music form. I'm blowned away about the time in human history that I'm living in. 🤕
The AI demo you saw in the beginning is a scam - they used a real instrument to play that bit to sell an app, pretending that it was your voice converted to that instrument. I do think it's possible to create an app which will be able to use voice to play any instrument, as is the case in the demo. The demo gave a good idea for developers - but the demo was also a scam, since we can see that the real end result is nothing like what they were stating it would sound in that demo.
@@Eric-dd8bk th-cam.com/video/lI1LCfTx2lI/w-d-xo.html So, according to the demo dude, he trained his own instruments - he did not use the original app sample - AND he used a crapload of additional effects to make it sound passable. Curiously enough, even that explanation video could be BS, though I'd love to see someone try to copy his steps to see if it's legit.
Dude, I have 40 years in music - You're singing out of pitch and changing pitch (drifting) all the time, thats whay it can't track you! But thats good as it shows both it's microtonal capbilities and that the Charles Babbage rule of GIGO is still true! Nothing personal ;o)
Oh lol that's my guitar model from like june. I've noticed that human voices work best with the electric guitar models. Otherwise, I think they just put all the instrumental models in one place, it's just plain ol RVC :)
I think this app seems like it's for musicians or attuned users to get the most out of. The original demo guy was a professional composer. While I love your videos Matt and will continue to enjoy them, maybe the quality of the output reflects the input? It gave me a chuckle anyway! 😂
You mean that I can, for example, train an rvc model with samples of trumpet guitar solos or something else, and use it as a model in the Real Time Voice Changer software? 😲
Your efforts are amazing, but the result is the same mean than your singing. But I think that if someone is good in matching the notes, it will work. Love it.
Good video. Thanks for trying out the premium for us... so we don't have to. HAHA. I think their examples in the demo might be tough to replicate actually using the software, especially as an untrained musician. Still early days with this technology, though... interesting to see how this particular type of software will advance over the following years. I imagine by leaps and bounds.
Thanks for covering this. Honestly though, this is... not great. Way too much of the voice comes through and just a hint of the instrument. Like someone else said, sounds like a audio filter. Not at all what the promo vid showed.
Music producer here. Omg!!! 🤣🤣🤣😭😭😭 I laughed so hard on the violin part. Let’s have a whole live stream doing this. So much fun!! But I would not pay for it quite yet.
Well, to be honest you are not giving the program a good source material to work with :D It is not about you singing in/out of tune or anything like that. But you really have to sing with the intended instrument's capabilities and character in mind. Not really sure what they were trying to achieve with the "cinematic" one, but i.e with your first one - the acoustic guitar, you were humming long sliding notes that are not really what an acoustic guitar is capable of doing ;) It sure is still in the uncanny valley territory, but I was able to squeeze some pretty decent saxophone performances out of kits-ai. Definitely better than what most expensive VST (virtual synths) can achieve with the brass / woodwind instruments! The reason being the opposite - you can not have the same amount of control over the timbre and vibrato's with the keyboard, but with the voice you can, and it translates really well.
A strange mix of your own voice and the intended instrument, it needs work to enable it to be more usable by the masses. Also similar apps have been available on iOS since about 2010, "Voice Band" for instance.
Tested it and i must say, even after serveral tries it just sucked. Either the trailer is highly cherry picked or the audio in there is straight up fake. Wont get my money with that
with Realtime voice changer client, there are some guitar models you can do this with and they sound fine. this version is more complex on how it picks up sounds. (btw the guitar one is a model ive used for it)
Yeah I suspect Musicfy is pretty much just RVC trained on instruments. You can still hear Matt's voice in all the results, so all it's doing is applying a filter, not actually generating instrumental tracks. It's very crude, not like what comes out of most generative audio AI.
Impressive as hell. Not quite the substitute for conventional instruments (yet, perhaps). I imagine it can definitely be improved upon. But, it is cool to make music with these sounds. I'm a millennial , but I imagine electronic keyboards and other electronic instruments got a lot of criticism at first. But musicians made a lot of amazing music with it. Likewise, these ai instruments might get a lot of criticism, but I can definitely see a lot of cool music being made with this.
Impressive? You can still hear Matt's voice in most of them. These results are worse than what can be done manually with some DAW software. There's plenty of TH-cam videos showing how to convert your voice into music, and the results are legitimately good, and sound nothing like a human talking into a mic. It's not accessible to the average person, but neither are the doctored results from Musicfy's demo video.
![[#2024MAMA] G-DRAGON - 무제(Untitled, 2014)+POWER+HOME SWEET HOME+뱅뱅뱅+FANTASTIC BABY | Mnet 241123 방송](http://i.ytimg.com/vi/Ox29z5Nf1Uk/mqdefault.jpg)
This is one of those Ai, when you say Expectation and Reality. It seems like it just scramble the voice with revurbs and other filters. Video 5+, Ai -1
No, this is basically the same as a AI voice changer. They just trained it with instruments instead of other peoples voices. But in both cases you need to have some skills to make it work.
@@missoats8731Your absolutely right. This is definitely NOT procedural manipulation. He should have tried a recorder or keyboard into it. I already subscribed to this and got mind-blowing results Anyone who can actually hit notes WILL get fantastic results. I'm not bashing MattVid!! You are awesome. But any minor hobbyist in music production can see the problem (and potential!) here.
@@ShitMental Another aspect to consider is that you have to make sounds that are at least somewhat possible to recreate with the desired instrument. For example, Matt was way too low with his voice for some of the instruments. If you do it like that, the AI will not know what to do, since it simply doesn't know such a sound. It's the same problem with human voices. That's why men should artificially pitch their voices up when they use a female singer model and vice versa.
@@missoats8731 This is my experience with these kinds of tools also, especially the voice part of this site. I can't just swap out a male for female vocal without transposing up 5 semitones. Also, I don't think true realtime coversion is entirely possible, as the ai probably needs a bunch of samples to preempt the note properly. Anyone who's used things like harmoniser peddles, which are far simpler and procedural knows it's never 100% tight real-time. (who knows what's coming though!). All this stuff is so fascinating, I could talk peoples heads off with the potential and just guessing what's coming next, or even inevitable makes my head spin!
This tool matches your pitch very well. It’s potential with proper direction are actually crazy
Now we just need a second ai to "tune" your spoken or sung notes before converting to instrument so that you don't get an untuned instrument track back as a result! lol
Yeah man, if you can't sing for shit like me you're still in trouble. lol
Lol auto tuning your recording then opning it.
Adobe has a pretty decent AI audio Enhancer I usually run my AI voices and text to audio stuff through that then I use the fair light tab in davinci, it's not perfect per se but you can def clean it up a lot. Between denoising and dehissing with a little compression then finally a tiny bit of reverb and I can get sometime palatable 😊
@legacylee but it optimized for speech, not every sound human can make
@@AvizStudio I have used it for regular audio before, like like they did switch it up now it just says voice but originally you could use it for any audio. I'm sure there are plenty of audio enhancers out there that can help clean it up a little and if you're familiar with DaVinci Resolve you could always use the fairlight table to tweak it even more
Just a tip…it’s going to copy your pitches so if you’re not in a tune you will get a not so desirable result. Try using audio files of some acapellas in key and then you’d get a better gauge of this tool in my opinion. Love this channel and always learning a lot. Peace and love guys 🙏
Yep - I saw the original video and I gotto say, it's a scam, because it makes all of the instruments sound perfect (I suspect they used real instruments for that 'demo'). This video shows what it really sounds like and I gotto say - it's not great, at all.
I do think, however, that it should be possible to use voice to dictate what an instrument sounds like - just obviously not this app - perhaps another developer will design something along those lines in the future.
@@realitycheck1086 the original video is not a scam, he adds a lot of audio fx and he can sing (unlike the guy in this video)
@@realitycheck1086
No.
The original guy did have an actual singing technic or at least knew how to sing in pitch and a key unlike this guy.
No hate on this guy.
i agree
Thanks Matt. You keep this old man in the loop. I can say we all appreciate your work.
I saw that video on YT long time ago (in AI world long time ago). The original video is from Emmet Ozcan, he is an incredible Artist, and he can do some crazy beatbox/singing. That is why that demo worked so good!
Food is good.
@@AcapellaFellaok
@@haffey2they're referring to your typo at end of the original comment
Yeah, I got the feeling that the original vocals/instrument has to sound close to the instrument you are converting it into in order for it to work properly.
you edited your comment but didnt fix the fact that you said "worked so food!"
It sounds like it is putting filter effects on the voice, rather than replacing the voice with instruments, because I believe I can hear your voice coming through.
The "master" seems to just apply a stronger filter.
Clearly that’s what it sounds like it’s happening when Matt does it but it’s more than that obviously: Unless the original demo faked it it’s not effects
He sang in pitch that the instrument he was trying to mimic could not play in.
Also, he needs to learn what the instrument he's trying to get atta this AI actually sounds like.
He said an acoustic guitar yet proceeded to sing sustained notes with bends, which just is not what acoustic guitars do.
If you can get that right, this actually does create convincing instrument sounds, just like in the original demo.
I was thinking that, singing through guitar effects would sound v.similar perhaps better in some cases, the demos seem to show it off at its best
Their samples/examples are 100% Fake. I purchased the expensive subscription and the quality is worse than any competitors I've tried 😢
Hello! What tool would you recommend? Hopefully with some free tokens to try it
Please suggest Free Ai Web site
Sheesh thank you for SHARING I was going try it 😂😂 man that's crazy tho
You need to take 2 points into account while you compare yourself to the original demo:
1. In the demo there are some post effects like reverb. It written there.
2. You need to have good rhythm and some musical talent to make the sound close to what you want like any voice conversion process
The demo is a scam. If you need to use another app to make the output sound like an actual instrument, then the demo is fraudulent at best.
@@realitycheck1086
lol
What an idiot.
Yeah seriously.
the original demo guy did make sounds that mimic the characteristics of the instrument he tries to get with his voice, where as this guy doesn't know how to sing, nor does he follow the characteristics of the sound of the instrument he tries to get.
You have to have that down. Then it creates pretty convincing instrument sounds just like in the original demo.
This was absolutely hilarious!
I've seen this guy's full video, and it's actually 10-15 minutes. He explains how he got the sound he did. Part of the reason you aren't getting the same quality as the guy in the demo video is that he was running the audio into a DAW to process the raw track (adding minor pitch correction), then adding effects in the DAW to the audio this tool produced. It sounded that good mainly because he's an actual audio engineer.
This tool is more meant for people who already know what they're doing in a studio environment and want to work out ideas they hear in their heads without having to hire session musicians. :)
Love your content man, and really appreciate the deep dives. Keep doing what you're doing!
Can you please provide the link to this guy's video?
@@meets2tarun th-cam.com/video/lI1LCfTx2lI/w-d-xo.htmlsi=759eQLTS8V9k0-Fz
@@meets2tarunDid you find it? If not the guy is Ummet Ozcan 😉
so you're saying the demo is a F*CKING SCAM, totally unrealistic to the average user
Why would a violin say "ne"? This is an absolutely amazing tool... for certain users! Thank you for showcasing it!
ni....ni......ni...
Perhaps the violin demands a shrubbery!
Lol, Matt, this vid was hilarious! I was having dinner while watching and it almost made me spit my food out.
As a musician, I’ve got to point out that one of the main reasons why it didn’t convert your voice as great as you’d like it to was because most of the time, you were singing in the same pitch or within the same few pitches. Kind of like monotone. But it’d work great if you made a big difference between the notes/pitches you were trying to sing. Maybe give this tool a try again in the future and try your best to sing an actual existing song, maybe you will notice the huge difference. 😊 cheers!
I think its just the RVC software but just trained on instruments instead of voices
Jop
This is great, now I no longer have to regret that I never learned to play a instrument, I just have to sing instead.
Ummet Ozcan has great voice even with out AI. And he used some audio processing on his results. I think that is what made the difference.
not quite ready yet, but in another year, for sure. it may be a "post-vst" world pretty soon, which would be nuts.
it's ready if you can sing
You can't gaslight us into thinking those clips were good bro, they were bad 🤣
The percussion was the only exception 👍🏻
Lmaooo this right here , FACTSSS 7:41
This is such a good idea! With some refinement, I could really see this taking off! I'll have to check it out when it's a little more refined. Great video!
Definitely seems like either the demo they used were heavily cherry picked, or there is actually some fine-tuning going on with the voices of the people in the demos. What was shown here was substantially worse than what was floating around Twitter for sure.
Bro, remember that this thing transforms your voice into instruments, Matt put zero effort into making a proper melody, that's why the result is shit. This could be the best AI, but if you don't put any effort into making a sound that's coherent with the chosen instrument, it will sound much better. I'm impressed
No.
the original demo guy did make sounds that mimic the characteristics of the instrument he tries to get with his voice, where as this guy doesn't know how to sing, nor does he follow the characteristics of the sound of the instrument he tries to get.
You have to have that down. Then it creates pretty convincing instrument sounds just like in the original demo.
This sounds absolutely nothing like the ad they showed.
Actually, this sounds exactly like RVC but trained on dry indtruments. The voice always comes through. They charge you for this cr@p?
Only here for Matt’s singing!
Ps 6:49 burn it downnn
Super amazing, I am astonished
You can actually do this with Ableton Live now and I know some music producers they use this for their business productions
The problem is that in the demo, the guy was using Pro Tools Studio to master the instruments.
The way it works is that you will have to vocally imitate the actual instrument and the AI only changes your timber.
glad you guys like our software
we will be launching realtime soon
At the moment it's pretty much oriented to people with good capabilities to emulate the components of an instrument, not for the average joe until it's able to just translate the intention of the rough notes and smartly recalculate the whole thing to the actual instrument, like using midi sampled libraries.
This might be very heplful for me in my case. There's a song that I've had stuck in my head since childhood. I only remember the melody. With this tool, I can actually create the melody and hopefully use that as a way to find that song.
I tried with actual sung vocals and it couldn’t handle the volume. Came out weak and clipping. Would have been cool to have an instrument accompaniment to my vocals.
🎷😙 _saxamaphooone... saxamaphoooooone_ 🎶
chewbacca mating call :D ahahhaha 6:10
7:05 the wayy I laughed soo hard when the master quality was the same as before 😂
As a music producer for over 29 years and musician for over 40 years...this will be a game changer in the way i can express my artform.
The guitar AI demo at rhe beginning was able to interpret intonation and expression and that makes for some next level music form. I'm blowned away about the time in human history that I'm living in. 🤕
The AI demo you saw in the beginning is a scam - they used a real instrument to play that bit to sell an app, pretending that it was your voice converted to that instrument.
I do think it's possible to create an app which will be able to use voice to play any instrument, as is the case in the demo. The demo gave a good idea for developers - but the demo was also a scam, since we can see that the real end result is nothing like what they were stating it would sound in that demo.
@@realitycheck1086
lol Random blank accusation with no substance nor evidence.
@@Eric-dd8bk th-cam.com/video/lI1LCfTx2lI/w-d-xo.html
So, according to the demo dude, he trained his own instruments - he did not use the original app sample - AND he used a crapload of additional effects to make it sound passable.
Curiously enough, even that explanation video could be BS, though I'd love to see someone try to copy his steps to see if it's legit.
Clearly needs quality input, in time. Has the potential for sure
Bro I am CRYING 🤣🤣🤣
"Never said I had a musical ear."
Dude, I have 40 years in music - You're singing out of pitch and changing pitch (drifting) all the time, thats whay it can't track you! But thats good as it shows both it's microtonal capbilities and that the Charles Babbage rule of GIGO is still true! Nothing personal ;o)
Oh lol that's my guitar model from like june. I've noticed that human voices work best with the electric guitar models. Otherwise, I think they just put all the instrumental models in one place, it's just plain ol RVC :)
I think this app seems like it's for musicians or attuned users to get the most out of. The original demo guy was a professional composer. While I love your videos Matt and will continue to enjoy them, maybe the quality of the output reflects the input? It gave me a chuckle anyway! 😂
So what if you generate music with one AI, then use this AI to clean up the audio and make it sound more like a specific sound?
*I TURNED MY SELF INTO A 🎻 MORTY!*
Bro, this review had me in tears. The results and your reactions 😂 Very informative and very funny bro
You can make it also with rvc open sourced. It's the same technology like human voice convert. It all matter of what model you trained
You mean that I can, for example, train an rvc model with samples of trumpet guitar solos or something else, and use it as a model in the Real Time Voice Changer software? 😲
errmm.....not quite. which is good, because it means we still need to have skill.
Imagine having no context and tuning into the middle of this video
that option in musicfy is now not there can any one check please
AI changes vocal scales and adapts according to the instrumental. (uploading a vocal as an example to analyze would be even better)
5:57 the mosquito in my ear at night
Your efforts are amazing, but the result is the same mean than your singing. But I think that if someone is good in matching the notes, it will work. Love it.
Good video. Thanks for trying out the premium for us... so we don't have to. HAHA. I think their examples in the demo might be tough to replicate actually using the software, especially as an untrained musician. Still early days with this technology, though... interesting to see how this particular type of software will advance over the following years. I imagine by leaps and bounds.
The video mentions the added effects on top of the instruments to make the sounds, you need a daw to add these effects.
Thanks for covering this. Honestly though, this is... not great. Way too much of the voice comes through and just a hint of the instrument. Like someone else said, sounds like a audio filter. Not at all what the promo vid showed.
Heavenly Yeah! I am want to try it. Thank Big Matt ✨✨😁✨✨
Kinda seems like that trailer was a tad misleading. I mean, it's not completely useless, but it's much more limited than they made it seem.
Wow, how festive Matt!
Music producer here. Omg!!! 🤣🤣🤣😭😭😭 I laughed so hard on the violin part. Let’s have a whole live stream doing this. So much fun!! But I would not pay for it quite yet.
I was thinking of buying a subscription for a few months but with these results I will wait for it to improve a little more haha
The demo in the start was made using a software called VOCHLEA DUBLER 2 not Musicfy
I know you did your best but i couldn't stop laughing the whole time 😂
Is it just me or does this sound like rvc with effects?
the more I watch the more it feels like this company is charging for a free open source ai
Exactly
Well, to be honest you are not giving the program a good source material to work with :D It is not about you singing in/out of tune or anything like that. But you really have to sing with the intended instrument's capabilities and character in mind. Not really sure what they were trying to achieve with the "cinematic" one, but i.e with your first one - the acoustic guitar, you were humming long sliding notes that are not really what an acoustic guitar is capable of doing ;) It sure is still in the uncanny valley territory, but I was able to squeeze some pretty decent saxophone performances out of kits-ai. Definitely better than what most expensive VST (virtual synths) can achieve with the brass / woodwind instruments! The reason being the opposite - you can not have the same amount of control over the timbre and vibrato's with the keyboard, but with the voice you can, and it translates really well.
Beeeeest video to date. Really fun work, bro.😂☮️
lol I'm that guy making those noises ALL DAY. This was made for me.
A strange mix of your own voice and the intended instrument, it needs work to enable it to be more usable by the masses. Also similar apps have been available on iOS since about 2010, "Voice Band" for instance.
BTW violin indeed sound close to human voice that's why the conversion is easier
As a beatboxer im pretty exicited for this
Finally, we can make mayonnaise an instrument.
I feel like if you have negative -10 music ability in your skill tree, maybe this isn't the AI for you.
expectations of the promo video didn't turn out well. 😉
can't wait to try out Google's Lyria .. this could be a blast.
In voicemod they have AI Voices in real-time and it use the same principle of voice conversion. So I assume it possible to make it real time
🤣 To the comments saying this is good... COME ON!!! This is terrible! Are we watching the same thing!? 🤣 It's just noise!!!
Tested it and i must say, even after serveral tries it just sucked. Either the trailer is highly cherry picked or the audio in there is straight up fake. Wont get my money with that
Love your content Mattvidpro
THis is amazing cause you now turn acapellas into your favorite instrument
This is game changing, yesterday musician must pay a lot for musical instrument. Now the creativity is unlimited!
Let us know when it gets better 👌
I quite like it the way it is, because it still needs human skill to master that.
I suspect it would help a lot if you cut out the heavy vocal fry. It can't figure out a fundamental from that.
Dude, that was totally the funniest video you've made so far lol
12:56 -13:20 Those facial reactions are such a memeable moment 😂
I just feel like some how "Primus" (Les Claypool) in in the a.i. that's creating this? Who's with me?
Thanks Matt!
This is just RVC V2 btw. An opensource software you can download for free.
Sir, you are not musically inclined are you? lmfao, hilarious video thanks.
No the heck did the original demo guy do so well?
Who can tell me what kind of melody it was ? 10:49
That was educative and hilarious. Thanks man
They're probably using RVC so it's possible to make those models yourself
Exactly
Do they already have instrument model weights to download?? I wasn't aware of that yet, I've only been using vocal models.
with Realtime voice changer client, there are some guitar models you can do this with and they sound fine. this version is more complex on how it picks up sounds.
(btw the guitar one is a model ive used for it)
Yeah I suspect Musicfy is pretty much just RVC trained on instruments. You can still hear Matt's voice in all the results, so all it's doing is applying a filter, not actually generating instrumental tracks. It's very crude, not like what comes out of most generative audio AI.
Yea I tried this one and paid for it and regretted it immediately 😢. But let's hope it improves like SunoAI has
sorry but I laughed hard at 6:17 you hoping it converts better and having high hopes then you give this face lol
Same here😂😂😂😂
Music but easier!!!!! Amazing stuff!!!
Love it. That's so funny and awesome at the same time.
Is there a way to clone your voice for singing? I want to hear what I would sound like if I knew how to sing.
That's what autotune is for, and it's an industry standard
@@drillerdev4624are you new to AI? Ai singing has nothing to do with auto tune . With Ai you can make anyone “sing”.
@@Edbrad he was asking for his own voice to sound ok singing. You don't need AI for that, just autotune.
you get paid how much to do this?
Bro Im sorry but the results made me laugh 😭😭
Sorry I have missed what the website is for make music from my speech!😊
All dependant on a library of LORA instruments.
Impressive as hell. Not quite the substitute for conventional instruments (yet, perhaps). I imagine it can definitely be improved upon. But, it is cool to make music with these sounds.
I'm a millennial , but I imagine electronic keyboards and other electronic instruments got a lot of criticism at first. But musicians made a lot of amazing music with it.
Likewise, these ai instruments might get a lot of criticism, but I can definitely see a lot of cool music being made with this.
Impressive? You can still hear Matt's voice in most of them. These results are worse than what can be done manually with some DAW software. There's plenty of TH-cam videos showing how to convert your voice into music, and the results are legitimately good, and sound nothing like a human talking into a mic. It's not accessible to the average person, but neither are the doctored results from Musicfy's demo video.
Seems likely this is just Google Magenta from 2017, with some fx slapped on...
Ummet OZCAN.... that guy is a legend
I’m a producer and I welcome it!