Can AI Make Me Sound Like a Pro Singer? (with Audimee)

[Music] storm is rising wind is crying storm is rising the wind is crying hi well I'm sure you've all heard them new songs with the vocals of Kurt Cain Freddy Mercury J Lipa Beyonce and a host of other famous singers dead or alive but what I was wondering is that's really the best years of AI vocals vocals created by artificial intelligence machine learning and a lot of other computer tech stuff so those uses of those AI vocals all seem a bit gimmicky to me so what can we really use them for in a home studio like this so let's [Music] [Applause] go yeah so like I mentioned in the intro the use of those AI vocals of famous people on your songs is that really the best use of AI vocals nowadays do you really want to release song with the vocals of Kurt Cain or even use it in a demo to pcture to an artist or to show it to your band for example and sure it's fun but to me it seems bit far-fetched for serious use in a home studio I'm much more curious about whether I would be able to work on the song in my home studio add a vocal myself although I cannot really sing and then get it replaced by an AI vocal which sounds much better so that would be much more useful to me over here and maybe also to you and for that purpose I got into contact with a service called aomy which can do just that so let's have a look now UT me promises a new era in music AI voice conversion and it can transform vocals to any of their singers who are not famous singers by the way and at the moment they have around 28 when I count it last now if you're on the paid plan these voices are royaltyfree so you can actually use them for commercial purposes also for releases on Spotify so if I click on this go through application I end up in my account for this and over here you can select which voice you want to use they all have descriptions for the kind of songs or type of vocals that the AI is best used for and the process is very simple you basically just drag and drop a clean vocal recording into the input box you can listen to it storm is rising you select the AI vocal that you want to use let's select James you can also listen to the AI vocal over here but I'm not going to do that because currently they're all singing covers so I may get a copyright claim if I do that over here you can pitch the vocal a bit if the original vocal is not exactly in the range of the AI vocal that you want to convert it to but let's leave this at zero if there's some background music in there you can enable this to get rid of that but actually they have some more instructions on the input file so although that technology is robust there are several brief suggestions which can make the results better make sure to have good input so preferably no effects on the original vocal as little background noise as possible and it's always good to choose a vocal that kind of matches the style of the AI vocal that you want to produce furthermore their suggestion is to use true mono audio files you can pre-process your input in order to eliminate background noises and use high and low pass filtering to eliminate unwanted frequencies maybe some gentle compression preferably no autotune and avoids excessive compression and then the output yeah you can process it in your door in the way you used to so let's actually try this conversion so this example file is about 34 seconds long and over here you can see that I have about 9 hour conversion time on this account and by the way let's get the disclaimer out of the way uty did give me this Pro account for free so that I had plenty of time to try out their product and see what it's like but they did not see this video before you're seeing it now and whatever I'm saying about this product is really my opinion and how I found it to work in practice so let's return because it has now completed the conversion and you can listen to it storm is rising the wind is CR in yeah so it's a nice husky male voice you can download file and a bit of a strange thing that I found if you look at the properties is that it's a 16bit 40 khz file which is at least not a very standard sample rate but obviously you can import it in your door and convert it to whatever bit depth and Sample rate that you want now another option here is that you can create your own voice models here maybe for example of the singer of my band I could create a voice model and then I could generate focals like his from any other vocal maybe even my own but that's not what I tested this time this time I went with the AI vocals that the website provides now if we look at the normal pricing for this is that you have a free tier in which you have seven AI artists 15 minutes of conversion time for you to try it out but it's only for personal use and the audio quality is MP3 and then you have the paid packages which basically provide commercial use of the generated focal and the big difference is in the amount of conversion minutes that you get every month for this price and if you're thinking of signing up I would advise just start with a free plan so you can see whether it works for you or your vocals that you want to convert and then you can always move to a paid plan later if it's really something that you want to use now before we start listening to an actual conversion and how I use it in a song If you like this video please give it a big thumbs up for the YouTube algorithm subscribe to the channel and if you ring a little bell icon you'll get notified when I publish another video another way to support the channel is by using the super thanks button below the video or buying anything through the affiliate links in the description to these stores in case you're shopping for anything I'll will get a small Commission of your purchase without any extra cost to you but let's now use a couple of these vocals in the real song and see how they sound and how they actually deal with a vocal sung by me even though I'm not a singer now if you've seen some videos on this channel before then you may recognize this song it's from my band The Wash it's called 500 miles and it was recorded by my band made perfect sounds and least and I'll include his socials in the description below as well and this is the vocal of our singer that I'm going to convert to AI storm is rising the wind is crying stars are far down oh he is rising the world is trying to keep those stars in place put a smile back on your face yeah so it's a pretty organic song no big Electronics or processed vocals or anything like that so let's see how some of the AI models sound when I use them on this focal cuz the first one is James that's basically the conversion that you saw me do in the beginning and the key wordss for James here are singer pop ballots emotional and Soulful so let's listen to how James sounds on this song storm is R the wind is crying stars are faring down hope is rising the world is trying to keep those stars in a smile on face yeah so I think it's a really nice voice for this song the best voice for this song is of course our own singer here currently I added just some slight Reverb on there in this quick mix all the other processing is really from Emy already so it seems to me that this vocal has been eqed quite a lot already for example if I just record a vocal raw like this it doesn't sound like that so let's listen to it in isolation for a second so we can maybe hear artifacts if if there are any storm is rising when this cry stars are fall down yeah I think you can hear some slight artifacts that seems to be a little rattle at the start of each phrase and I think it also has a little bit of that warble that you sometimes get when you autotune a vocal For example but in the mix it sounds pretty good to me storm is right C let's switch to Isabella so I'm now changing the vocal to a female and the key wordss for Isabella are singing Pop house in the soul so let's have listen now she sounds on this song storm is rising the wind is crying stars are far then down [Music] hope is rising the world is trying to keep those stars in place a smile face yeah I quite like a female vocal on there as well actually it's canny how for example at the beginning of a phrase you also get the breath in there and that's also very much present in the original vocal storm now if you listen to Isabella you may find that we're quite low in her range which is obvious of course because it's a male vocal that she's trying to sing now so it's also possible to pitch that up while you're converting and then the vocal sounds like this which I'm going to show you in Solo now of course because the music is not in this key but this seems to be more in the Isabella range storm is rising the when he's crying stars are far L down yeah pretty nice now another singer that I tried is Aston and Aston is also female singer and her song is more suitable for punk rock pop belting so let's see how she does on a song like this actually is rising the wind is crying stop [Music] down yeah even though this is not a punk rock song of course I think it's still has something for this song as well again because she's female she's kind of low in a Range so this would be as and pitched up storm is rising the wind is crying stars are down okay so now back to what I actually really wanted to try is how about if I use my vocal if I record this vocal line with my vocal and then converted with an AI vocal how well does that work and you have to remember I'm not a singer at all I will let you listen to my vocal in a minute which I would never do on the real recording I don't even sing background vocals in the band but for Science Education and testing I'm willing to share it and you have to remember that this was not a fancy vocal recording I just recorded it on this sm7b microphone sitting down like I am now I recorded it three times and the third time was the best so that's the vocal you're going to hear so don't be too harsh this is my vocal in isolation storm is rising the wind is crying star are fall down yeah so I really couldn't do the high fall set line with my voice so I just stayed low in there and I did try this on James but I was not too overly impressed with it so what I first did is tune my voice even though uty says that you should not use at least excessive autotuning I think it did give better results so let's first listen to my vocals slightly autotuned storm is rising the wind is crying Stars Are Falling Down so yeah so if I wouldn't be recording a demo of a song to later show to the band for example this is the way I would add a vocal totally not demo quality of course a bit embarrassing even how does this sound if I replace it with an AI vocal from James stor is rising the wind is crying and in the song storm is rising the wind is crying stars are fall Linda [Music] hope is rising the world is trying to keep those stars in place put a smile back on your face yeah so it's still not a great vocal but it's definitely way way better than the way I was singing it and if I wanted to present this song to somebody I would feel much better with this vocal than I would with my own vocal now let's listen to the difference between James generated from our singer's vocal and James generated from my vocal so you can see the difference that the original vocal makes to the generated AI vocal stor is rising the wind is crying [Music] star down reverses this storm is rising the wind is crying stars are fall down yeah so the input focal really makes a big difference to the results now as for a conclusion well for me well first of all maybe there's an ethical question as well right instead of me hiring or paying a singer to sing this song on a demo for me I can just use the AI local and yes that's true of course but is it really different from a keyboard player playing a guitar part by using guitar samples or using one of those extended guitar vsti that's also a guitar player which was not hired to play a guitar part and replaced by a keyboard player and obviously these singers which have probably been hired in by aomy have gotten compensated or will get compensated for use of their vocals now as far as quality of these vocals I must say I was really impressed especially when using H's vocal as input a real singer yes you can hear some artifacts if you listen to the vocal in isolation however there are so many musical styles which have lots of vocal processing nowadays anyway so if you listen closely you really hear a lot of those artifacts on those focals as well and they're even on purpose you did see that it's quite important with what kind of quality vocal you feed the AI though it sounded much better when the AI vocal was generated from our singer vocal than when I generated it from my vocal so I don't think I would ever use an AI vocal generated from my vocal for a real release but for a demo in which I would normally just mumble my vocal definitely it's much better and if I generate the AI vocal from our singer's vocal then I think it's very usable in the release for example maybe for additional background vocals with a different sound or maybe even by replacing part of the lead by a female singer for example so that you have a bit of a duet in the song and even with my vocal with a bit of practice for pronunciation and delivery I probably could have improved it a lot already and finally I found it really quite a lot of fun to try out all of these different vocals on this song and give it a bit of a different vibe from the way that we've been hearing it all along and that we're used to as a band so I found this a really nice experiment now if you want to try out the service for free or maybe later paid I do have an affiliate Link in the description that you can use and that will give me a commission if you register with their service so it's again a nice way to support the channel if you wanted to try out these AI vocals anyway now as for vocals it's also really important how you process a vocal and I have a separate video about how to get a really smooth vocal which you can check out over here have a look enjoy and see you [Music] soon

Channel: Lanewood Studios

Views: 42,932

Keywords: Music Production, Home Studio, Music Producer, how to, music production for beginners, mixing studio, home recording, recording studio, Cubase, lanewood studios, cubase, awash, awashmusic, cubase pro, artificial intelligence, ai vocals, audimee, audimee.com, vocal replacement, replacing vocals, generated vocals, artifical intelligence vocals, pro vocals

Id: Vggzb6Vf-S4

Length: 16min 52sec (1012 seconds)

Published: Fri Nov 03 2023