Best Vocal Isolation & Instrumental Extraction 2021 [lalal.ai vs Spleeter]
HTML-код
- Опубликовано: 8 июн 2021
- A comparison review video for extracting instrumentals and vocals between Lalal.ai, Spleeter(splitter.ai), and Audacity!
Be sure to check lalal.ai out at bit.ly/2TU7BpH, a high-quality tool based on the world's #1 AI-powered technology, that lets you remove vocal and instrumental from any song!
Full versions of the comparisons
[K/DA - VILLIAN] • Music Separation AIs T...
[K/DA - MORE] • Music Separation AIs T...
[VJ Memes - Soul Searching] • Music Separation AIs T...
[jlbrock44 - That's All I Want To Do] • Music Separation AIs T...
This video is supported by the kind Patrons:
🙏Andrew Lescelius, Mazen Alotaibi, Sascha Henrichs, Jake Disco, Demilson Quintao, Martin Schmitt, Odium 2k, Tony Jimenez
Support me on Patreon if you hope to see more:
/ bycloud
or by becoming a member instead (same perks!):
/ @bycloudai
[Discord] / discord
[Twitter] / bycloudai
[Patreon] / bycloud
[Music 1] Steaminwaffles - Aquarium Boy
[Music 2] jlbrock44 - That's All I Want To Do
[Music 3] JV Memes - Soul Searching
[Music 4] K/DA - VILLIAN
[Music 5] K/DA - MORE
[Profile Art] / pygm7 
Наука
This is some of the coolest technology to come out of recent history in my opinion, thank you for this wonderful breakdown on how it works!
Am I the only one who loved that text filling progress effect while the music was playing?
Why are the names for AIs always so dumb? When they become sentient, they're gonna be pissed at what we called them lolol
Actually, who says that your designation as to what is "dumb" or not is even close to correct? And why would a sentient AI care about understanding things using the English language or verbal sound at all?
*lalal lol
This will be very useful for streamers that like to re-upload clips from their streams to their RUclips channel or twitch highlights, making it to where they don’t have to worry about the video getting copyrighted by music labels!
they already can. There is several ways to make the music sound in the stream but not in the vod
Reason why I'm here😂
More like the new age for sampling music
@@animeman3392 u make twitch vids??
Wow, just a week ago, I wanted to extract the vocals of a song but none of the programs/apps convinced me
This is very helpful, thank so much!!!
Your channel is the best!
Cool thing about spleeter is that it offers a feature where you can isolate individual instruments.
Same for lalal, it allows for bass, guitar, kick, etc separations...
Nothing comes close to UVR - Ultimate Vocal Remover, easily beats all competition so far and it's free and open source.
I’ve used AI for splitting vocals and instrumentals for some mashup stuff that doesn’t have official releases. Really cool stuff here and hopefully it keeps improving.
At 9:19 you say it's "much less muffled when spleeter has quite a lot of up-and-down".
8:22 Voice Removal from lalal sounds way worse than spleeter at 8:48 imo, especially comparing lalal's 8:31 with spleeter's 8:57. It's ironic that you say that spleeter has quite a lot of up-and-down, but I came to the exact opposite conclusion. 8:31 is way more of a rollercoaster than 8:57.
I do think that lalal generally did a better job in the comparisons you showed, but I think that it's probably for the best to compare both services on any audio if you want to get the highest quality result.
I want to take a moment to commend you for taking the time to compile these comparisons, cause you totally didn't have to do that and could've convinced most people that this is better than any predecessor by just saying it supposedly is. This kind of nuance is what I feel misses with Two Minute Papers. It's nice that you're still able to get that bag while showing potentially unsatisfactory results.
thanks for your kinds words and ya I think I didn't present it clear enough
at 9:19 I was referring to the lalal vocal isolation being less muffled compared to Spleeter. While it's the opposite as Spleeter has better vocal removal in that segment than lalal.
To simplify what I said in my video:
Vocal Isolation in that song
Lalal > Spleeter (because lalal conserved higher frequencies)
Vocal Removal in that song (Instrumental)
Lalal = Spleeter (both has up and downs)
@@bycloudAI 8:35 vs 9:01 indeed does totally line up with what you were saying in the video, thanks for the clarification and summarization!
@@bycloudAI in my opinion, vocal removal for that song spleeter is better haha. Like for that song lalal has a more noticeable volume decrease in parts while spleeter doesn't. Of course, this can easily be fixed with volume normalization in audacity and similar programs, but still I'd say spleeter won the vocal removal for more.
Great video as always! Super interesting subject, I'd love to see more papers related to music.
Absolutely awesome thing, from a thing that it was a nightmare to do of isolating the hole track this aliviates the need of messing around with the settings.
No surprise an isolator has to struggle with highly stylized vocal layers like in the K/DA examples, where obviously vocoders were used to model the special way it sounds. The neural networks are probably trained on rather naturally sounding vocals.
Ive been searching for stuff like this for years
Oh my god, I wanted something like this in ages, all the other options are not reliable except spleeter but I didn't know it existed before
Great vid my man!
La la killing The music isolation game right now no other Music isolation can compete
i think that the best solution is to use both services, and combine the best parts manually.
nice vid, nice channneeel !
This site really is mind blowing. Not “perfect” but IMO 99% perfect with rational expectations.
Great channel!
Audacity with the results of voice isolation is powerless
I remember when this website was in beta, I requested an API integration for a VST I made. It’s quite amazing what it can do even through the VST
idk why ive never thought of one of these as a vst, probably several out there but that can be very useful tbh
Subscribed. Gr8 vid brodie
time to instrumentalise my beloved tracks!
Great review of the three and excellent corresponding visuals to see what things actually looks like. I've been using lalal for breakdown of some old band tracks from 20-30 years ago and its been very helpful pulling out vocals and other instruments so as to be remixed at more suitable levels. The big bonus of Spleeter is its free once one gets over the initial installation and configuration which for me has been a technical barrier. Not impossible, but haven't spent the 2-3 hours that I envision figuring it out. That said, you've proven to me that it does a different but fairly comparable job and though lalal is very affordable / reasonable and super easy to use, it really comes down to the amount of audio one has to edit. If your total needs are less than 6 hours of music lalal is a no brainer, but if it gets into hundreds of hours of music, Spleeter may be the best option or at least the most affordable.
after seeing ur vid i went test lalal for myself, i wouldn't say it's the best on the market, neither the price nor the quality. however it could be of best quality within a certain genre. for me i prefer dango for its traning data was based on asian pop and female vocal, i guess to get the best isolation for your target you just have to know what each ai does best very well. BIG THX FOR THE VID
Great A.I. review video.
I hope this comment boosts your channel, in the RUclips algorithm.
thirsty
so helpful - cheers
I actually plan on using one of these AIs to get seperate bass and vocal tracks to create a 5.1 master of my favorite songs
Stayed for the memes 😎
The Audacity effect should not be regarded as vocal separation. It is centre channel separation.
loll
nice video, thanks a lot
Ultimate Vocal Removes wipes the floor with both, and it's free
I think of this as Audio Segmentation and it's a tech I've been waiting for since the 80s! I wonder why Adobe doesn't appear to be contributing to research here?
excellent
I have both the instrumental and the original song, is it possible that I could use both combined to extract the vocals?
:o that is really awesome
cool thanks
also that K/DA acapella is fire i wish you could share that file but that would probably go againt lalal terms and the nature of your sponsorship
In TV and radio debates, overlapping voices create confusion and hinder understanding. Our solution is the BoondockEcho device, which separates and identifies individual voices for clear transcriptions and reports. This enhances the quality and comprehension of debates, benefiting broadcasters, journalists, researchers, and viewers alike.
The Audacity Vocal Isolation sounds kinda fire
Came from discord ✌️
I think it's pronounced Lalalai.
Please make a video comparing with the new Splitter.ai 2 Stem Pro
I wonder how this would do with the song Paradisus-Paradoxum by MYTH & ROID. I feel like the AI could struggle with it, but I guess there's only one way to find out.
It would be nice to hear a comparison to rx8 😯
I finally did it ma...I finally did it.
Cool!
I'm stll having hard times to believe that such improvements made on technology. I mean how can you tell a program which waves are vocals? Well of course we couldn't answer this question back in 2010 but now I can safely say that the program learn it from experience!! It listens to the other songs and decide the similar waves that could be modal. It learns from thousands of songs and have a common vocal track, and then uses phase reversion to remove the actual vocal waves from the song.
Welcome to the machine learning, folks. Welcome to the future.
Hi everyone, I'm new to this whole thing. I have been wanted to make some covers of some songs and if I use 'Spleeter', I will still have to get a music license, right?
I have a video, the person is talking in the video. The video also has music in the background as the person is talking. The background music is rather loud and it is hard to concentrate on what the person is saying. I like to remove the background music in the video but keep the audio of the person that is talking. Is there any free programs, that I can use that will remove the the background music, but keep the audio of the person talking? Thank you.
What about Ultimate Vocal Remover v5? Is it old now?
tested lalalai. pretty impressive results! i gues - youtube need to take things like that for videos with copyright music. replace thing that they use now
sounds like if you find a way to combine spleeter and lalal -- you'd probably have something pretty solid -- but I'm just a pretty dumb layman.
This is fucking good
that's what she said
Only works with music right, the only "voice" I've ever had to remove were an out of breath guy holding the camera. When would you ever need to remove the voice in a song.
karaoke?
I love you for putting K/DA
Lalal definitely seems to be the best, but you do have to pay for it. Whereas, Spleeter is open source and therefore costs nothing.
Pretty awesome to see how far this technology has come over the years.
What about iZotope RX8? It wourks out pretty damn good.
it's one of the best, only thing it could really do is utilize other hardware for the rendering to be faster, my gpu is just chillin
Great job thanks
I just want to extract tracks from shows that don't have a soundtrack
Spleeter iz Betta.
Ya Betta belive me
League of legends songs are copyright free?
Now for an ai that separates each instrument
Available to some extent at least for one year now. Upload your file to ezstems.com and select 5 stems. You get vocals, drums, bass, piano, and other.
More interesting would be splitting stems into pure vocals/instruments plus the parameters for the sound effects.
Spleeter can actually already do that.
Rx8 Music Rebalance works way better than all this
Basically, those who's only trying to mess around and sht, without intention of making money out of any of these, are locked out of this tech.
El servicio online que brinda está pagina es muy bueno, pero gratis ?.........
Pero como no todo en esta vida es gratis, esta página solo te permite procesar dos canciones.......
.
lalai.ai looks great, especially with the additional instruments it features. BUT, it appears to be only an online upload option. No actual software to run on your own computer. No thanks.
wait, i never knew league of legends music was such a jammer!
Need me more exper-tize-ing.
Lalalai: 😴
UVR: 😎
No
Shouldn't it be pronounced LA LA LAI?
THIS
get voice samples
get music
put randomly filtered voices over music
train AI to generate music and voice streams with same network
I though Spleeter sounded best
120th liker
They use the same software lol all of em do .at least all free ones do lol
I mean u can find 50 free online that will isolate vocals n the rest ..but the software all the same and it's not very good
vocalremover sucks process just slows down at 80%
why is this a 12 minute video
Moneeey
what a weird name
First
Yes. Yes, you were first. 🙁
Izotope RX 8 is better lmao
No it isn't. I own it
@@SeanyKrabs i own that too
man, your channel gives me anxiety
Wow audacity sucks
finally I will be able to remove all those tough-acting offended snowflake vocals from my deathmetal music collection and enjoy music only, yay!!
IRONY... putting annoying incessant background music over your narration, which i would need to remove before your video is ok to listen to.
Facepalm.
i also use xtrax stems 2 and phonicmind.com, both use AI as well
Don't forget there is also this script too! I can't run it locally on my system but i use colab to use it. It provides me some good enough results might work better tho if you use isolation on an instrumental and the normal song to phase some of it out.
also use wav or flac, but mp3 might give some better results sometimes but at lower quality like with yt rips, no idea why.
github.com/tsurumeso/vocal-remover
there is also some user made ai models for it as well as a gui version. although those i can't really use x3