Hi, I'm Mark... aka mixPHANTOM! I've been mixing, mastering, producing and restoring audio for 30+ years. My plan for this channel is to make a repository of useful stuff to know... along with tests, analysis and audio production techniques. I have been making "video notes" for my own reference for years because I like to fully explore my tools and techniques - so I thought I'd "up" my production value and start a RU-vid channel! Thanks for checking it out!
Thank you soooo much for this clean demo. I struggle finding one not sounding creepy, even on official demos changing parameters each single seconds. Your video made me think that I definitely need this tool !
Great video, and with original ideas. From the Dreamtonics youtube video, I had the idea of recording myself with a big smile while singing, than run Natalie and perhaps injecting even a small smile on her singing (a vocal mode that does not exist), so my first idea was to have Natalie sing and possibly improve her expression. I didn't think of recording myself to improve MY voice and have it sing in tune as Saros does..
I'm sure every person out there is anxiously awaiting the amazing ability to transform their voice into their OWN voice again??? I mean, I doubt if there's anyone out there would like to have their voice transformed into a variety of female singers or different male voices in different octaves and different blends thereof. What a waste of time it would have been to show us how to do THAT. The only thing that's actually helpful here are the comments that say how difficult it is to get registered with these people and how it could be used to make deep fakes. Which you just did.
I was late to the party and others had already done the male to female demos so my thought was the best way to test the ability of the voice changer is to see if it would maintain my voice if I gave it my voice - in theory if it can't do that, it's hardly going to change my voice into Peter Gabriel's! 😎
Thank you!! This stuff was a little bit more intuitive with Vocaloid so I was having a hard time figuring out fries (outside of that 5-specific vocal fry overlay we had actual glottal stop phonemes and stuff), and you explained this all pretty well! I didn't even know pau existed!
Would be good to have an extra tool on the graphs to 'reduce' or smooth so when it's over-detailed in places but spot on in others we can adjust the troublesome bits and leave the rest.
Another well-timed video Mark, thanks! 👍😎 My earlier experimentation with SVS said that my sensitivity would be 75%, which is essentially the same thing, so that's cool. I tried most of the other recommendations apart from the last one, where the Transfer is off, will give that a go on some BVs later today and if it works on my next song. Cheers & Beers!
With "Rounding" off, Synth V places the midi note in 25 cent increments. The note is more centered around the pitch curve. But it is not in tune. If you adjust the note the pitch curve moves as well. So you preserve the details of the pitch curve but the whole note is now more in tune. great. But to do this for many notes is tedious. Have you found a way to do it in one step, like Variaudio can? Maybe a macro exists?
Interesting re the 25 cent. I'm sure it's possible to create a script if you have the skills... Variaudio is my solution too - I treat Synth V like a "real" singer and always tweak timings and pitch (and everything else!) in Cubase.
I watch this one year later. Would you say the software is now better? I want to create some trance tracks with female vocals on them, wonder if this would work.
Thank you for your tutorials. I'm trying to watch them all and they've been an excellent resource. I'm mind blown by this software, just purchased it two days ago! This video was really helpful! Amazing.
This company should hire you, most of us would not have gotten this far..I am grateful personally. I wish I had more control over pitch though, trying to edit the graphs are really a pain..Also more flexibility when it comes down to short cuts. I hope that this software improves although it's already incredible. Thanks again
A super cool piece of technology, which I would have purchased on day 1 to use on my own voice. But doing a “know your customer” process with some random stranger on the Internet is a non starter for me, when it’s just a matter of time until that information is exploited for evil by hackers, venture capitalists, investment companies and their ilk. - It’s not a question of “if” there will be tears in the future, but “when”.
@@mixphantom0101 yes, but if protecting Dreamtonics means exposing me and their other customers, it’s simply not a price worth paying for me. I’m just not willing to be a human shield in this war.
The fact that you only really need one voice model, (maybe 2?, one female and one male), makes vocoflex addition really useful to create a whole range of vocal styles, however, the voices still tend to sound like someone who's native language is not english, trying to sing in english, (a bit like early Abba tunes when the girls sang the lead lines with quite a heavy accent), and i've yet to hear a convincing "Brit Pop" voice as all the demo's i've heard tent to sound Americana influenced. Its still relatively early days with this technology so as it develops i expect that will improve, i still see it as more useful for providing backing vocal groupings rather than lead vocals at this stage, but watching your, (and other guys), efforts with it is really cool and i can only see it improving as future software developments are made, keep it up, people are watching and following, nice.
Yes, Vocoflex is 100% dependent on the vocal performance it's changing - so the better the underlying performance, the more realistic the result! I became very aware that Saros pronounces with an accent. I messed around singing with my best "Cool For Cats" accent and Vocoflex does retain the accent - it's definitely more of a clever filter than an actual voice changer.
The id process is a nightmare been back and forward with them for a week and still no clearance, its absolute overkill if you ask me and very invasive as the software is spying on everything you do with it, they have so many areas IP email bank details etc to know who you are and on top of that cloning software is freely had of the internet, another case of the legit guy getting shafted trying to do the right thing when the software will be hacked in no time and passed around to anyone.
@@jesus4pincho Tell the hackers that they will abuse it not the paying customers who gave them their bank details, email and IP address let alone official identification that might end up on the dark web in 6mths.
@@jesus4pincho Its overkill, the customers that give them their bank details, email, IP address which are all traceable are not the ones that will abuse the software, its another case where legit customers are being persecuted for being legit while the bad guy will hack and abuse the software and it costs them nothing, Ive been 7 days and many emails jumping through hoops multiple times only to finally get cleared then bought and paid for the program and still no download link, its total BS.
It is overkill... but if it helps to know, the KYC is handled by a third party so Dreamtonics doesn't get copies of your data beyond your registered name.
@@mixphantom0101 And that is a worry, hackers are hacking banks, medicare, numerous data bases world wide everyday, how long before your personal details are on the dark web is my concern, I don't think they needed to go down this route with this program as Ive said my traceable ID is given to them without handing over personal official documents such as your pass port which they ask for.
It sounds like you have a lisp when singing. I'm not sure if it's the AI or you. Also, your singing voice seems so different compared to your speaking voice. For some people it is really like that, but are you sure you' re (' correctly' ) using your singing voice? A lot of people who dont know how to sing well, do not really know what their ' true singing voice is
Yes, I'm pretty sure that Vocoflex is more like a Vocoder than an A.I. Voice Replacer. It does struggle with certain sibilance, vocal fry, growling sounds etc. I'm sure the A.I. part of it's process will get refined by having a bigger database of analyzed voices.
It appears it requires Windows 11. Apparently, it's quite easy to work around a Windows 10 computer that "fails" the Microsoft spec for Window 11 upgrade.
Have you tried to set it up so you can get a soft whisper, to a normal, to tense, to a belting voice? You think Vocoflex can do that? Compared to RVC, besides how you convert VST vs offline, do you see any other benefits?
Yes, if my voice wasn't so croaky I'd have included it in this demo. A really clean and in tune vocal performance is needed for the best result - Vocoflex isn't generating a new voice, but is "shaping" the voice with resonant filters so capturing a soft, mid and belt performance as three voices and then animating the blends yourself works better than one capture where the singer goes through the range of tones in one clip.
(You probably knew this, but) The phoneme blocks are green in the piano roll because the Pitch Mode is "Sing" (green), not because of the track's own color.
I've been using Synth V for two years and I still "mess up" because all the tracks are the same colour! Dreamtonics needs to think of another way to identify "sing" modes.
New rabbit hole commensed. One of the best ideas I've seen on youtube in a long time. Really tells you how much can trust your real world sources. I should take a convolution of the alexas in my house cause I've always wondered what's actually going on with these. I've got 3 in my house cause the girls all want to listen to music in their rooms. The wheels are going now.. What have you started Mark, What have you started!! 🤣😅