As someone who does a lot of AI-powered lyric swapping, I've gotten pretty familiar with what's possible (and not possible) with AI music tools right now. So today we're diving into Lalals, an AI voice changer and music generator, to see how it stacks up. 🎤 Get 10% off Lalals with code MUSIC10.

• AI Voice Changer
• AI Music Generator
• Custom Voice Cloning
• Pricing
• Final Verdict

This is a paid review 💲 But you can count on me to give my honest thoughts, both to you and to the Lalals team. Let's get into it!

AI Voice Changer

First up, the main feature - the AI voice changer. You can choose from a curated selection of celebrity voices like Beyonce, Justin Bieber, Ed Sheeran, and many more. The big question is: how convincing are the vocal transformations? 🤔

In my tests, the results were generally solid. Many of the voices, like their Beyonce model, sounded pretty damn good.

But others, particularly the less popular ones, fell a bit flat or had some noticeable audio glitches.

👉 It can be hard to test models because a lot of factors come into play: the talent of the vocalist, the key of the song, etc.

My hunch is that Lalals is sourcing these voice models from various places online, rather than training them all in-house.

There's nothing wrong with that though! They have someone there curating models for you. Which is great! You don't have to go through 10 vocal models to find a decent one.

Some features they could add here to improve the product would be:

• The ability to pitch shift by semitone, instead of just octaves
• Batch processing audio files

Overall, I think Lalals is a solid starting point for AI vocal beginners because of their curated selection. Get 10% off Lalals with code MUSIC10.

AI Cover Songs

Out of all their features, I think this is where Lalals is even with the competition. 

The idea here is that you provide a specific song YouTube URL to a song and then Lalals changes the voice to the new singer.

And it works just as well as other tools on the market.

Keep in mind that AI cover tools usually don't handle group vocals very well. It confuses AI in the inference phase.

But for fun quick covers, particularly with solo vocals, it's a great place to start experimenting.

Some features they could add here to improve the product would be:

• Better processing for group vocals (which is difficult and no one has really done well... yet)
• The ability to select or trim to specific areas of the song

AI Music Generator

Okay, let's talk about the AI music generator. Lalals lets you type in a song description and lyrics. And it'll spit out an original song based on your prompt.

Cool concept, but how's the execution? 🎧

One feature they offer, that I haven't seen anywhere else, is the ability to use a specific AI singer in your original song generations. To do this, you must navigate to the singer page first, then use the "Lyrics To Music" option.

If you aren't trying to use a specific singer, you can get started on the "Music AI" page.

In my opinion, the music generator needs some work. The AI-generated instrumentals were OKAY - but they usually felt generic and contained AI noise. And the vocals had a lot of glitchy artifacts that made them sound fake. Definitely not something you'd want to use in an actual track.

I know the team is always trying to improve, but the Lalals generator just doesn't quite measure up when compared to other AI music generators. It's a fun toy to play with, but I wouldn't rely on it for any serious music-making.

Custom Voice Cloning

One neat feature Lalals offers is custom voice cloning. Upload 10+ minutes of voice recordings, and their AI will create a custom voice model that you can use in the app.

This is a great thing to offer if they want to compete with the other AI vocal tools on the market.

But having to provide a FULL 10 minutes of audio is kind of overkill. I've gotten solid results with other tools using just 3-4 minutes of voice data. 🗣️

Also, fair warning - Lalals currently makes you upload the voice clips as a zip file, which is a bit clunky compared to other platforms. Hopefully they streamline that process soon.

Unfortunately in my test, I wasn't able to use the custom voice I created. Even after waiting a day to make sure training was complete.

Pricing

Let's talk pricing. Lalals is charging $12/month for their basic plan, which only lets you clone 1 custom voice per month. Considering the hit-or-miss quality of some of their core features, that feels a bit steep to me. 

Competitor platforms are offering more bang for your buck, either with more voice credits or higher-quality results (or both). I think Lalals might need to rethink their pricing tiers to stay competitive, especially for music producers or anyone who needs to clone multiple voices regularly.

But hey, if you're just looking to dip your toes into AI voice tech and see what all the hype is about, their free plan could be a good low-stakes way to get started and have some fun messing around. 🆓 Use code MUSIC10 for 10% off Lalals.

Final Verdict

So where does that leave us with Lalals? Here's my honest take:

What They Are Doing Well:

👍 The AI voice changer has some solid celeb voices, curated so you don't have to go hunting for decent models.

👍 The idea of creating original music with specific vocals models is good - but it needs work.

👍 Good for beginners who want to experiment with AI voice tech.

What Needs Improvement:

👎 More serious AI music producers will find better quality & value elsewhere.

👎 Music generator needs work - instrumentals are generic, vocals glitchy, noticeable noise.

👎 Custom model was not usable in my test - like it didn't output any audio despite trying multiple times.

Bottom line: If you're just getting started with AI music tools, it could be a fun playground.  Give it a try! Use code MUSIC10 for 10% off Lalals.

Don't forget to sign up for more AI music tips. And if you need any custom vocals for your next track, hit me up - I'll take good care of you! 😉 Peace out. ✌️