AI DJ and Stem Separation — The 'Era of AI DJing' Begun in 2024

In 2024, rekordbox 7.0, djay Pro AI, and Serato Sample 3.0 introduced full-fledged AI features. Stem separation separates voice / drum / bass / other in real time; auto-mix has the AI automatically select tracks and sync BPM. Will AI replace DJs, or strengthen the human DJ? Traces the biggest technical shift in 100 years of DJ history.

djai-djstem-separationauto-mix2024future-of-dj

2024 — The Practical Emergence of AI DJs

In 2024, a major technical shift occurred in the DJ world:

  • rekordbox 7.0 implemented stem separation
  • djay Pro AI (Algoriddim) installed full AI DJ features
  • Serato Sample 3.0 for AI track analysis
  • Denon Engine DJ Prime OS supports AI stems
  • Deep-learning source separation runs at practical speed

The “era where AI assists DJ play” reached the practical stage. In parts of the DJ world, the debate that “AI replaces the DJ” has also begun.

What Is Stem Separation?

Stem separation is a technology that breaks down a finished track into its original elements (stems):

The Four Stems Separated

  • Vocals — main vocal, rap, muttering
  • Drums — kick, snare, hi-hat, cymbal
  • Bass — bass line, sub-bass
  • Other — synth, guitar, piano, pad

Previously

  • What DJs could use was the “finished track” — the 4 stems mixed together
  • Impossible to say “I just want the voice”
  • Producers needed to stem-export a multi-source

Realization of AI Stem Separation

  • Estimated separation of stems from a finished track by deep learning
  • Accuracy 80-95% (as of 2024)
  • Real-time operation possible (rapidly improved in 2023-2024)

This expands the technical range of DJ play fundamentally.

Practical Examples — DJ Play with AI Stem Separation

Example 1: Voice + Another Track’s Instrumental

  • Extract only the voice from track A
  • Play only the instrumental of track B
  • A’s voice + B’s instrumental creates a new mash-up

Example 2: Drum Swap

  • Remove the drums from track A
  • Replace with the drums of track B
  • A’s melody + B’s drums creates a new groove

Example 3: Bass Shift

  • Remove the bass from track A
  • Emphasize the bass in real time to fit the club sound system

These were impossible in 1990-2000s DJ play. Previously, the producer had to make the mash-up in advance. Now possible improvised during DJ play.

AI Features of rekordbox 7.0 (2024)

Pioneer / Alpha Theta’s rekordbox 7.0:

AI Stem Separation

  • 4 stems computed automatically during PC-side track analysis
  • On CDJ-3000 / OPUS-QUAD side, real-time stem separation
  • Advanced subscription plan required ($30/month)

AI Track Analysis

  • More accurate BPM detection
  • Key detection (musical key)
  • Energy level analysis — chart of a track’s energy over time
  • Mood analysis — emotion labels like happy, dark, energetic

Auxiliary Features

  • Compatible track suggestions — AI proposes “tracks that suit this one next”
  • Playlist auto-generation — specify a theme, AI creates a playlist
  • Set flow prediction — AI predicts the flow of DJ play

djay Pro AI — AI DJing on iPad

Algoriddim’s djay Pro AI (from around 2020) is an AI DJ app running on iPad / iPhone / Mac:

Features of djay Pro AI

  • DJ play possible on iPad alone
  • AI Automix — fully automatic track connection
  • Neural Mix (djay’s version of stem separation)
  • Direct integration with Spotify / Apple Music
  • $6-13/month (cheaper than rekordbox)

How to Use djay Pro AI

  • Connect headphones and speakers to an iPad
  • Press the Automix button and AI selects and mixes tracks
  • Hours of DJ play with no human intervention is possible

This is also called a “DJ-less” music playback system. For BGM use, it can fully replace the human DJ.

Actual AI DJing — Marshmello and Sound Machines Radio

Applications of AI DJing:

Marshmello’s AI Collaboration (2023)

  • EDM DJ Marshmello signed a protocol on AI music production
  • In studio sessions, AI generates melody and drums
  • Marshmello uses these in DJ sets

Sound Machines Radio (2024)

  • 24/7 AI-only radio on Spotify and YouTube
  • No human DJ, AI selects tracks fully automatically
  • Thousands of concurrent viewers
  • One alternative to “hire-a-human-DJ”
  • On Beatport streaming, AI recommends “tracks that suit you”
  • Easier for DJs to find new tracks they didn’t know
  • AI replaces the DJ’s source-digger function

Human DJ vs. AI DJ — Technical Comparison

Current (2024) AI DJ vs. human DJ:

Element AI DJ Human DJ
BPM matching Perfect (mathematical) Nearly perfect (skilled)
Beat mixing Perfect Skilled are good
Song selection Choice from database Feeling, audience judgment
Emotional flow Programmed flow Reads the mood of the venue
Improvisation Limited Strong
Stem separation Real-time Requires preparation
Cost $10-30/month subscription $500-100,000+ per set
Physical presence None Present at venue

AI DJ = efficiency, technique, low cost; human DJ = emotion, audience response, uniqueness. The mid-2020s is a stage of coexistence.

Where AI DJs Excel

1. BGM Use

  • BGM at restaurants, cafes, bars
  • Average hours-long dance music
  • The 50,000-yen cost of a human DJ is unnecessary
  • Some stores already run AI DJ + streaming

2. Fitness / Gym

  • Automated workout playlist
  • Selects tracks matching user BPM / heart rate
  • Fully automated personal DJ

3. Personal Streaming

  • Alexa or Google Home: say “party music” → AI DJ plays automatically
  • Home DJs become AI

Where AI DJs Struggle

1. Club / Festival Main Sets

  • AI still lacks the ability to read the audience’s reaction
  • Difficult to judge “what should be played now”
  • Cannot read the mood of the time

2. Artistic DJ Play

  • Selection that creates surprise and unexpectedness
  • Unexpected mixes of classical tracks
  • Break switches that b-boys expect

3. Community Building

  • DJ play with cultural significance — sets like Berghain, Bushbaby, DJ Nobu
  • Creates a social phenomenon beyond music

These are areas where AI struggles at the stage of mid-2020s. But many may become possible around 2028-2030.

Ethical Issues of AI DJing

Ethical issues that come with AI DJing:

1. Artists’ Rights

  • Is it possible for AI to “mix in the style of Deadmau5”?
  • Whose style is allowed to be borrowed?
  • Complication of copyright and personality rights

2. Labor Issues

  • BGM DJs and cafe DJs are replaced by AI
  • Falling DJ income at small venues
  • Risk of replacing Wedding DJs and Corporate event DJs

3. Music Homogenization

  • What AI learns is existing sources
  • Possibility that new music and styles will be harder to emerge
  • The value of DJ play that does not presuppose AI (= the rarity of the human DJ)

Modern DJs’ Responses

Modern pro DJs’ responses to AI:

Active Adopters

  • Deadmau5: introduced AI music production early
  • Skrillex: uses AI stem separation live
  • Marshmello: uses AI as a “creative partner”
  • Nina Kraviz: uses AI-generated sources as stem materials

Cautious

  • Peggy Gou: “AI is an auxiliary tool; the center is the human ear”
  • Ricardo Villalobos: “Value human improvisation”
  • DJ Nobu: “Relying on AI erases individuality”

The modern DJ world is split into “use AI vs. keep distance.” Complete rejection and complete dependence are both minorities; most take an intermediate position of “auxiliary tool.”

Fusion with Ableton Live

In addition to AI DJing, the fusion of Live production also proceeds:

Ableton Live Used in DJing

  • Ableton Link syncs with CDJ
  • Bring Live’s loops, samples, effects into DJ play
  • CDJ + Ableton = hybrid DJ

Representative DJs

  • Deadmau5: live performance with Ableton Live
  • Richie Hawtin: DJ set with Ableton Live + CDJ
  • Amelie Lens: new expression with Live + AI

This is the fusion of “DJ = one who selects tracks” to “DJ = one who makes tracks.” The boundary between producer and DJ is disappearing.

The Image of the DJ Around 2030 — Prediction

Prediction of the DJ 6 years from now (around 2030):

1. AI Becomes a “Basic Feature”

  • Stem separation is standard in all DJ software
  • BPM matching and key mixing are fully automated
  • The human role is track selection, reading the mood, and staging

2. Hybrid AI + Human DJ

  • Peggy Gou, Amelie Lens, and other top DJs will be human-led with AI support
  • Mass BGM becomes AI-only
  • Middle DJs (mid-scale venues) mix AI + human

3. VR / AR DJ

  • Metaverse clubs like VRChat
  • Audiences worldwide experience DJ play in VR
  • Overlay DJ set info in real clubs with AR glasses

4. Streaming Subscribers Become Main Revenue

  • DJ music revenue via monthly subscription
  • Beatport LINK, TIDAL, etc. become mainstream
  • Music purchase model becomes minority

Philosophical Questions of the AI DJ Era

Philosophical questions in the AI DJ era:

“Who Is the DJ?”

  • 1935: the person who talks on radio
  • 1970s: the person who makes people dance in clubs
  • 1980s: the person who plays instruments
  • 2000s: the person who discovers and curates music
  • 2020s: the person who makes a space of unity with the audience
  • 2030s?: A music curator co-working with AI?

“What AI DJs Lack”

  • Empathy — feeling the hearts of the audience
  • Serendipity — unexpected selections
  • Uniqueness — the reason “because it’s that DJ”
  • Narrative — a story through a night-long long set

These may not be fully replaceable by AI. But whether AI can “prove it cannot replace” is still in the experimental stage.

The DJ Lineage of 100 Years and AI

Through 100 years of DJ history (1920s radio → 2020s AI):

Era Technique Music Place
1920-1960s Radio, records Jazz, rock and roll Home
1970s Turntable × 2 Disco Club
1980s Roland 909/303 House, techno Huge club
1990s CDJ series House, techno, trance Mega club
2000s rekordbox, PC EDM Festival
2010s Streaming, SNS Multi-genre Boiler Room
2020s AI, stem Genre-fluid Hybrid

Technology kept changing, but the core of moving people through music has not changed. Even in the AI era, the human ear, emotion, and reading of mood should remain at the DJ’s center — that is the hope of the DJ world in 2024-2026.

From This Article to the Next (to the Final Article)

We traced 100 years of DJ lineage, but there was little chance to touch on Japan’s DJ culture. As the last article of the dj series, we trace the 30 years of the Tokyo club scene (1990-2024). Tokyo club history — WOMB, ageHa, Vent, Contact, etc. — the international activity of Japanese DJs — Takkyu Ishino, DJ Nobu, Wata Igarashi, and others — and considerations on the “DJ culture unique to Japan.”

← Back to 100 Years of DJ