LALAL.AI Review: A Potent Audio-Editing Tool

Background: What Is LALAL.AI?

Even with so much advanced technology involved in everything, musicians still want more. And it’s not really a surprise. For instance, it’s been quite a challenge to do mastering or additional mixing on an already finished song.

Or, even worse, it’s been almost impossible to extract individual audio tracks from a finalized file. If you do any sort of music recording, producing, or mixing, you’ve tried extracting a vocal or instrument track from a file. And it’s an almost impossible task.

Well, if you’ve been looking for a solution for this, you’re in luck. There’s actually a pretty decent simple tool for doing so. The tool in question is called LALAL.AI and it’s just incredibly simple to use.

We’re looking at an independently-made browser-based app. It’s capable of analyzing files and extracting individual tracks. According to the official info available on their website, they’re a team of specialists who devised their own neural network called Rocknet and its superior version called Cassiopeia.

I’m not entirely sure how it works as I’m only interested in the finalized files. However, they “trained” the software to do this using 20 terabytes of data.

How Does It Work?

Essentially, the app uses the neural network to analyze and “extract” individual instruments from the full mix. It’s not something that would interest an average user. But if you’re more into the whole process, I suggest that you check out this blog post about Cassiopeia.

Basic Features and Interface

To start working with audio files, all you have to do is open the website’s home page. On there, you’ll find the upload button. Or, you can just drag and drop the file over it.

Above the “Select File” button, you’ll find the “Stem separation type” dropdown menu. Essentially, here you select what you want to separate. There are seven options:

So if you select the first option, you get a separated vocal track in one file and an instrumental in the other file. If you select drums, you get the isolated drum track as one file and everything else without drums as the other file.

After uploading the file, the app takes some time to process it. Once it’s uploaded, you get to listen to short previews of these files. If you’re satisfied with what you hear, you can select the option to finalize these two files and download them.

And that’s pretty much it as far as the features go. And that is pretty much all you need with a tool like this one. Features are pretty simple and to the point, just the way that they should be.

Is It Good?

Now, to be perfectly honest, this is one of the most interesting track extraction tools I’ve seen in a while. Two things made this one so impressive. First, we have its super-simple interface. It’s very intuitive (in fact, it may feel too intuitive to what we’re used to) and you can just drag and drop a song that you want to process.

Secondly, the tool actually works pretty well. Of course, it can’t do magic and you can notice that something has been altered and extracted. And it works much better with higher-quality audio files. Nonetheless, it’s super-useful and will save you a lot of time if you need a vocal track extracted.

But I’ll get into more details in the section below.

Extracting Vocals and Instruments

I explored the app with four individual songs. I tried to make it as diverse as possible and try out different things. These were the results.

Song 1

I first attempted to extract vocals from an mp3 file of a full song. This was a 1980s metal track with distorted guitars and loud cymbals, so it felt like a tough one. To my surprise, I got a pretty decent vocal file out of it.

Now you could notice that both instrumental and vocal tracks are extracted. Of course, tracks weren’t super clean and you could notice where the app had to completely change the EQ.

I can say the same thing when I extracted the guitar from it. On the other hand, the track without the guitar sounded pretty great.
However, the song I chose was tough. It was drenched in delay and reverb and it featured dynamic compression. Not to mention that it was a low-res mp3 file (128 kb/s).

Song 2

But then I tried another song as a WAV file. Here we had a metal song with more modern production. A tough one as well, mostly due to high-gain guitars. Nonetheless, the mix was clearer and the file sounded much clearer.

As expected, LALAL.AI did a much better job on this one. While not super-perfect, I was able to make out each note in all of the riffs and lead parts. Just to note, we’re talking about super-distorted and downtuned guitars. So this was pretty impressive.
The EQ was, once again, changing throughout the song. However, it wasn’t as “glitchy” sounding as with the previous low-res file.

Song 3

My next attempt was with a softer rock song, featuring acoustic guitars, electric guitars, and a lot of vocal overdubs. Acoustic guitars were extracted with ease. I can say the same about most of the vocals as they all sounded clean.

There were a few mild glitches in the overdubbed vocal tracks. However, we’re practically talking about a choir so it’s still pretty impressive how clean it turned out to be.

Song 4

For my fourth and final test, I used the best-sounding file that I had with me. It was a FLAC, featuring a 32-bit resolution and a 48 kHz sample rate. It’s also a well-produced and well-mixed pop rock song from the 1980s. So I’d say that it doesn’t get better than this.

The results were, obviously, the best with this quality of audio. While there were a few spots where you could notice that the vocal track was extracted, it still worked incredibly well.

But when I extracted the electric guitar, I was amazed with how super-clean it was. We’re talking about a twangy-sounding slightly overdriven Fender Strat. It cut through the mix pretty well and I manage to make out every single note from the extracted track.

Are There Any Flaws?

Overall, I can’t seem to find any flaws with it. If you want me to be really nitpicky, then I’d like to see some improvements with the design and user experience. But that’s not much of an issue.

As for the “glitchy” sound of some tracks, I can’t really complain about it. I’ve fed LALAL.AI with some tough nuts to crack, especially with the first example. If the file is high-resolution and we’re looking at a mostly acoustic and polished song, you’ll have no trouble at all.

Who’s It For?

LALAL.AI is for pretty much anyone who wants to extract vocal or instrument files. DJs who need individual tracks for remixes or samples, instrumentalists trying to learn songs by ear, instrumentalists looking for backing tracks, vocalists looking for backing tracks – everyone is welcome. In particular, you can get the cleanest results from acoustic tracks.

I find it super-useful for musical analysis. If you want to make out individual parts that aren’t easy to figure out in the full mix, the app certainly helps.

However, objectively, DJs and anyone doing covers will find the best use for it. Even if the isolated track doesn’t sound super-clean, they’ll find a great use for it.

Free Demo

Another thing worth mentioning is that they offer a free demo. You get the chance to try out three songs before making a purchase. Honestly, it’s more than a great deal. I tested it with four songs just to make sure. However, I think three would be more than enough to see how it works and what the app is capable of. After that, you can make a decision and decide whether you want to purchase it.

Conclusion

Overall, this is a super-useful tool for anyone looking to do remixes and covers of famous songs. As I mentioned, don’t expect it to do magic. There are plenty of musicians out there who would like to extract individual tracks from low-res MP3 files.

But you can’t expect to get high-quality results this way. I could say the same thing with some very specific songs or odd mixes, but that’s not a surprise. The rule is simple – if you use a high-resolution FLAC file and a song that’s mixed well, you’ll get the best results.

Nonetheless, LALAL.AI is an impressive tool. Above all, I was surprised by how easy it was to use. Just put your file into it, select what you want to do with the file, and you’re all set. In addition to this, the app isn’t even that expensive, especially because it’s one of the best options on the market today.