technology

Spotify DJ: Meet the voice behind the AI


When you click on the DJ button, it’s X’s voice that you hear (Picture: AFP)

Spotify jumped on the AI bandwagon this year with a new music personalisation feature called ‘DJ’ that started rolling out to Premium users across the UK and Ireland last week.

I used it and loved it. So, I caught up with the voice behind the AI — Xavier ‘X’ Jernigan.

When you click on the DJ button, it’s X’s voice that you hear. So, if you’re using Spotify in the UK, you might have heard him already. X tells me his experience of having Spotify’s AI voice modelled on him.

How does it feel to hear yourself as an AI?

‘It’s really exciting! The technology is so good that it sounds exactly like me. When I hear it, it’s like listening to myself on a podcast, which I’m used to. The trippy part is when it’s words in an order that I know I didn’t say. That’s when it becomes like, “What…?”

The coolest part is that the voice not only sounds like me, but the word choices are mine. It’s based on my personality, so it feels authentic. It’s like I injected my DNA into the project.

The voice captures my intonations, pauses, and natural breaths, and the word choices reflect my unique style. For example, I don’t say “tunes”; I say “songs,” “jams,” “bops,” “bangers,” and “hits.” When I hear that, it feels exciting and normal. And as a user, I get hyped when I listen to myself set up a segment, and it’s the perfect song at the right time. So I get really excited and give myself a hi5. I’m like, “Good job, X!” every time it happens.’

Spotify jumped on the AI bandwagon this year with a new music personalisation feature called ‘DJ’ (Picture: Spotify)

Can you tell us the process of putting your personality into this AI voice?

‘We spent a great deal of effort making sure the quality was there. The process involved recording my voice and training the voice model. Nothing that you hear are actual recordings. We’ve captured the way I talk, including things like the sound of a period, the length of pauses, where I take breaths, the intonation of a comma, and how I emphasise things naturally. We wanted to capture my unique speech pattern and capture a conversational tone.

Readers Also Like:  ADHD and the trend to self-diagnose by TikTok

The words and language you hear reflect my authentic self. Even the intro, “Hey, how you doing? I’m Xavier. My friends call me X.”—that’s exactly how I introduce myself.

Additionally, we have a writer’s room with music experts, data curators, and scriptwriters, where we discuss music and culture to provide storytelling around the artists. So, it’s a combination of the amazing text-to-speech technology through our Sonantic acquisition, my performance as a voice actor, and our knowledge of culture and music. All of these elements come together to create the voice you hear in the DJ.’

When did it hit you that ‘this sounds just like me’ when it wasn’t actually you talking?

‘When I was just told about the project, there wasn’t anything to show me at the time. I understood that I would be hosting but not actually speaking—the voice would be AI. It sounded exciting and futuristic, so I immediately wanted to be a part of it. But I still wondered how it would actually sound. Is this going to sound like me? Is it going to sound like a robot, or will it sound a little fake?

When they sent me a clip, I thought, “Okay, yeah, that sounds like me,” since I’m used to hearing my recorded voice. Then, about 30 seconds later, it hit me. “Wait, I never said these actual words; this is the AI voice talking.” Mind blown! It was such a surreal experience, and I just wanted to hear more.

The other piece was when I finally played it for my mom. She just thought it was something I had recorded. Then I knew it was spot on because it passed the mama test. Even though I thought it sounded like me, I knew we had something special once she said it.’

Readers Also Like:  Limiting global temperature is least of businesses’ concerns ahead of Cop28

Do you use the AI DJ, and how is that since it’s basically just you?

‘As someone with a background in music and curation, I didn’t think it would be a product for me. But once I used it and I had the context and storytelling around the songs and why they were being selected for me, it just took my listening to another level. This is now the primary way I personally use Spotify. I’m in the top one per cent of DJ users, and I love it.

It’s helped me discover more new music and artists in the past six months than in the previous six years. I enjoy the set-up, the discovery aspect, and how it supports artists by providing context around their songs and telling their stories. It’s the perfect thing I didn’t know I needed.’

How does it feel to potentially have your voice everywhere, like Alexa or Siri?

‘I tried to keep myself in a bubble, but it’s happening now in the US, where I’m starting to get recognised for my voice, and people are excited to meet me. With this product, I like it because it’s not about being your assistant; it’s about being that friend that’s there to help you enjoy life a bit more by giving you the music that you need at the right time.

I love it best in this context because that’s who I am — a creator and a curator. So it’s really humbling that people are connecting with me in that way because I know they’re also then connecting with artists and their music and becoming fans.’

AI is a concern for music artists. As someone who has worked firsthand with this technology, how do you think it will affect artists?

‘AI is changing at a rapid pace. We’ve never seen technology in general move so fast. It’s changing every day. So we have teams that are looking at the ethics behind it, and we’ve factored that into how we’re using AI specifically with DJ and other things to come.

Readers Also Like:  Microsoft re-launches ‘privacy nightmare’ AI screenshot tool

Then there’s how other companies and people are using AI in general. When it comes to Spotify, we do not stand for anything that encroaches on creators’ creative expression and takes their Intellectual Property (IP) without their permission. What we do stand for is the use of AI to enhance creative expression with human input.

I don’t think anything will ever replace humans, and our use of it isn’t looking to replace humans; it’s looking to enhance the potential of human creativity, which is the most beautiful thing on this earth.’

What can we expect from Spotify DJ as we move forward from the beta version?

‘You can expect us to continue innovating and staying fresh. We won’t let it get stale. It’s going to keep learning you, and the quality of what DJ is saying will continue to get better. So, expect it to keep blowing your mind. Just stay tuned, keep using it, and keep rocking with me as your DJ. I promise you won’t be disappointed. It’s only going to get better.’


MORE : How to connect Spotify to BeReal – easy guide to pair your music


MORE : Spotify’s AI with ‘stunningly realistic voice’ will tell you what to listen to





READ SOURCE

This website uses cookies. By continuing to use this site, you accept our use of cookies.