Spotify's AI DJ Learns to Talk Less: Nine New Voices, a Mute Switch, and the Real Job of an Algorithmic Host
Updated:
Most talk about AI in music streaming is about machines that make songs. Spotify's AI DJ does another job. It makes no music at all; it decides what you hear next and then explains its choice in a synthetic human voice. On November 10, 2026, Spotify gave that host nine new voices with distinct personalities, along with a less flashy option: a switch that mostly makes it stop talking.
Spotify presents both changes as personalization. Taken together they also say something about the product. Three and a half years after launch, the company is admitting that one voice talking constantly was not what every listener wanted. It is also betting that the host can remain the center of the experience even when it barely speaks.

Image credit: Spotify
From One Cloned Voice to a Full Cast
The DJ's history is short but dense. Laid out in order, it shows a feature that grew mainly by adding control, not by adding intelligence.
- February 22, 2023. Spotify launched DJ in beta for Premium users in the U.S. and Canada, in English only. It combined three things: Spotify's existing personalization engine, OpenAI's generative models in the hands of Spotify's music editors (who use them to scale commentary), and a voice platform from Spotify's 2022 acquisition of Sonantic. The voice belonged to Xavier "X" Jernigan, Spotify's Head of Cultural Partnerships and a former co-host of its morning-show podcast The Get Up. According to TechCrunch's launch coverage, Spotify called his voice the "first model," a hint that more would follow.
- May and August 2023. DJ reached the U.K. and Ireland, then 46 more markets, per update notes on Spotify's original announcement.
- July 17, 2024. A Spanish voice arrived: "Livi," modeled on Olivia Quiroz Roa, a senior music editor at Spotify in Mexico City. She was picked after what Spotify called an extensive international casting call.
- May 2025. DJ began taking voice requests ("play something for a rainy commute"), described by Android Central as the feature's last big overhaul before this one.
- October 15, 2025. Requests expanded to Spanish and to typed text in more than 60 markets. Spotify said DJ listening had increased since voice requests launched.
- May 7, 2026. French, German, Italian, and Brazilian Portuguese were added, each with its own persona. Spotify put DJ's reach at 94 million Premium users and more than 75 markets.
- November 10, 2026. Nine new English and Spanish voices and a commentary toggle. TechCrunch reports DJ is now in more than 95 markets. Spotify's own announcement gives no market count, so that figure rests on TechCrunch alone.
One pattern stands out. Each milestone moved some control from the machine to the listener: first the choice of language, then the ability to make requests, then text input, and now the choice of voice and how much it talks.
Personality as a Settings Menu
Spotify's announcement describes each voice like a cast list:
| Voice | Accent | Spotify's description |
|---|---|---|
| Casey | U.S. | Laid-back energy, warm, a bit of a drawl |
| Tess | U.S. | Direct but inviting, with an edge; in the know |
| Lane | U.S. | Southern accent, younger and positive |
| Sam | Australian | Gregarious, fun, not-so-serious |
| Nell | British | Mellow, slower, deeply knowledgeable |
| Declan | Irish | Straightforward, passionate, a bit of a music nerd |
| Nisha | Indian | Inviting, with contagious energy |
| Nico | Mexican Spanish | Down to earth, present |
| Fabi | Mexican Spanish | Witty, friendly, a step ahead |
X stays. Livi does not: TechCrunch reports that Nico and Fabi are now the only Spanish voices, replacing her.
Spotify has not said who, if anyone, the new voices are based on. X and Livi were both Spotify employees, and Spotify told their stories to vouch for the product: a real person with real music credentials behind the synthetic voice. The new lineup comes with personality adjectives instead of biographies. That is a small change, but it fits the framing. TechCrunch notes that Spotify is borrowing from the personalization menus that general-purpose AI assistants offer, and a selectable "personality" is a product setting, not a colleague. (This is the author's reading; Spotify has not commented on the change in framing.)
The Mute Switch Against Spotify's Own Pitch
The commentary toggle is the more revealing change. With it switched off, DJ gives only brief acknowledgements when you start a session or tap the button to change direction, and plays music the rest of the time. Spotify's example use case is a dinner party, where you want "a seamless soundtrack in the background, without chatter competing with the conversation."
Compare that with how Spotify justified the commentary two years ago. In the 2024 Livi announcement, the company said that when listeners hear commentary alongside recommendations, "they're more likely to listen to a song they may have otherwise skipped," and that the median DJ listener spent more time with DJ than with any other recommended set. In Spotify's own telling, the talking was a large part of how DJ worked.
So why build an off switch? Spotify does not frame it as a response to complaints. Its stated rationale is to let DJ cover moments when people would otherwise put on a playlist. The press is more direct. Android Central's Rowan Davies wrote that interruptions between songs "plague users just as much as its voice," and called DJ "like Marmite," loved for hands-free playback and disliked by others. Spotify's own figures show a loyal core: 80% of DJ listeners come back week after week, according to the November announcement. Neither Spotify nor the coverage gives numbers on how many people tried DJ and left because of the chatter.
The likeliest reading (the author's inference) is that Spotify is separating DJ's two layers. The commentary was the novelty that sold the feature in 2023. The selection engine underneath is what keeps people. A DJ with the commentary off is essentially a steerable, endless personalized station, and that format competes directly with playlists, autoplay, and radio.
The Critics: It Picks What You Already Know
The new voices don't touch the most common substantive complaint, which is about what DJ plays, not how it sounds. Davies wrote that the update "doesn't fix the biggest issues with AI DJ — those being its repetitiveness and hindrance to music discovery." His sessions, he said, filled up with his most-played songs from the past week, and requests for rising artists returned acts he already knew. He found more new music through Spotify's Daylist. That is one writer's experience, not a measurement, but it echoes the first impressions. TechCrunch's 2023 hands-on found the selections no more remarkable than Spotify's personalized playlists, and said tapping through segments "felt like switching between playlists."
There is also a structural critique. In Mood Machine (2025), journalist Liz Pelly argues that Spotify's engineers have increasingly designed for the "lean-back" listener, who uses music mainly as background, and that this shift is bound up with how the platform values and pays for music. In a Boston Globe interview, she described the company as selling users' own taste back to them: "The data becomes its own commodity." Pelly's book predates the DJ voices and does not address this update. The connection is the author's. But a DJ that can now run silently in the background at a party is close to the purest form of the lean-back product she describes. The framing is not new to DJ: at launch, TechCrunch's 2023 coverage noted the feature could turn Spotify into "a lean-back, passive experience."
What does this mean for artists? A single feed that runs for hours on a listener's past behavior tends to favor what that listener already plays. Getting into that feed matters more than any one spoken introduction. Spotify has not published data on how DJ selections are distributed between catalog favorites and new or emerging artists, so this question cannot be settled with public evidence.
What It Means for Human Hosts
The DJ's name invites a comparison with broadcast radio. The most cited cautionary tale comes from Australia. In April 2025, the Australian Radio Network acknowledged that "Thy," a weekday host on Sydney's digital station CADA, was an AI persona built with ElevenLabs voice cloning. She had been on air for about six months with no disclosure, Mediaweek reported. Teresa Lim, vice president of the Australian Association of Voice Actors, said the network "should have been upfront and completely honest" and raised questions about whether the employee whose voice was cloned could give informed consent. ARN's own statement conceded that the trial had "reinforced the unique value that personalities bring to creating truly compelling content."
Spotify's DJ differs from that case on the point critics cared most about. It is labeled AI, has been from the start, and the original voices came from named, consenting employees. TechCrunch's 2023 interview with Spotify personalization head Ziad Sultan described Jernigan recording lines in many emotional registers so that afterwards "it is purely AI-generated."
The overlap with human radio work is still real, though the evidence about job effects is thin. No reliable source we found links Spotify's DJ to specific job losses in radio. What the product does show is the shape of the threat (author's inference). The parts of a DJ's job that can be scaled, such as picking the next song and saying a sentence about it, are now a settings menu with nine accents. What remains hard to synthesize is what ARN's statement pointed to: a specific person with a point of view, local knowledge, and accountability for what they say. The commentary toggle suggests that even Spotify's audience values the synthetic version of the voice less than the music it introduces.
Conclusion: A Host That Knows When to Stop
This update looks cosmetic: new accents, a toggle. It is better read as a correction. In 2023 the commentary was the product's selling point. In 2026 the commentary is optional, the voice is a matter of taste, and the selection engine is the part Spotify is sure people keep coming back for.
Whether that engine is actually good at discovery, as opposed to recycling a listener's recent favorites, is the question the new voices leave untouched. Spotify reports a 94-million-user audience and 80% weekly return. It has published nothing on how widely DJ spreads listening across artists. Until it does, the most accurate description of the AI DJ is an impressive and well-liked interface sitting on top of the same recommendation logic it was meant to humanize.
Sources
- TechCrunch: Spotify's AI DJ gets new voices, personalities, and an option for less chatter (Sarah Perez, 2026-11-10). Source article: the update, the Livi replacement, the 95+ market figure.
- Spotify Newsroom: Make Spotify's DJ Your Own With New Voices and Even More Control (2026-11-10). Official voice list and descriptions, commentary toggle behavior, the 80% weekly-return figure.
- Spotify Newsroom: Spotify Debuts a New AI DJ, Right in Your Pocket (2023-02-22, with updates through 2026-03-18). Launch architecture (personalization + OpenAI + Sonantic), X Jernigan, market rollout notes.
- TechCrunch: Spotify launches 'DJ,' a new feature offering personalized music with AI-powered commentary (2023-02-22). "First model" voice, "lean-back" framing, early hands-on impressions.
- TechCrunch: Spotify's new 'DJ' feature is the first step into the streamer's AI-powered future (2023-03-13). Ziad Sultan interview on how the voice was built.
- Spotify Newsroom: Spotify Brings AI DJ to Spanish-Speaking Music Fans Around the World (2024-07-17). Livi, and Spotify's data on commentary reducing skips.
- Spotify Newsroom: DJ Levels Up With Spanish-Language Requests, Text Requests, and Personalized Prompts (2025-10-15). Request features and the 60+ market figure.
- Spotify Newsroom: Your Personalized DJ Experience Is Expanding With 4 New Languages and More Markets (2026-05-07). 94 million users, 75+ markets, four new languages.
- Android Central: Spotify's AI DJ is getting 9 new personas… but it won't fix the feature's biggest issue (Rowan Davies, 2026-11-10). Critical view on repetition and discovery; May 2025 requests timing.
- Mediaweek: Backlash over ARN's use of AI radio host modelled on Asian woman (2025-04). The CADA "Thy" case, Teresa Lim's remarks, ARN's statement.
- The Boston Globe: In 'Mood Machine,' BU grad Liz Pelly examines the hidden costs of Spotify's music empire (2025-01-09). Pelly's lean-back critique.