67 MUSIC
67 MUSIC by Shela

Say it to Shela. Hear it in seconds.

67 MUSIC is the music studio inside Shela. You write one sentence, the way you would text a friend. About half a minute later the song is finished: written, performed, mastered, with a title and a cover, in your library and ready to share.

Want an artist to release it under? Say so and Shela invents the whole person, gives them a face and puts them on their own page. Want none of that and just music for yourself, for your set, for your clients? That works too.

Just say it

Three ways people start

No forms, no sliders, no genre wizard. You say what you want; anything you leave out, Shela decides and tells you what she decided.

Hey Shela, make me a new persona: Tiago Vela, from Lisbon, Portugal, he plays Latin tech house. Generate his profile picture, set everything up, and make me two viral TikTok tracks at 126 BPM.

Done. Tiago Vela, 29, Portuguese, from Lisbon, living between Lisbon and Ibiza. Tanned, cropped dark hair, trimmed beard, white linen, a thin gold chain. Warm baritone, 122–126 BPM, dry congas up front, never a supersaw.
His portrait, bio and profile are up, and both tracks are mastered: Sal na Pele, hook at 0:12, and Madrugada for the end of a set. If the name or the look is off, say the word and I'll change it. One sentence in. An artist with a face, a home town and two finished songs out.

Hey Shela, I'm a producer. Give me three ideas for my next release, pop and K-pop, write the hooks and the full lyrics, then render the strongest one at 100 BPM with a female vocal.

Three ideas with full lyrics: Ghost In My Playlist (pop), Neon Hangul (K-pop) and Say Less (pop-R&B), each with the hook inside the first 15 seconds.
The first one is rendered: 100 BPM, female vocal, 2:58, mastered, cover made, in your library. Want the other two as well? No artist, no persona, no forms. Plenty of people use 67 MUSIC purely for their own production, in any style.

Hey Shela, a two-minute intro for my set tonight: 126 BPM, E minor, sung hook, dry congas, and give me a version without the vocal too.

Both are done: 2:04, 126 BPM, E minor, sung hook at 0:48, mastered to −14 LUFS. The instrumental is the same length and the same master, so either one drops into the set. WAV and MP3 are in your library.
Want it a key lower or thirty seconds longer? Or open it in Wave and rebuild it yourself. Your production, your rules. Nothing leaves your library until you publish it.

What happens after you say it

From your sentence to a finished song

  1. Shela writes seconds

    If you asked for an artist, she invents the person: name, country and city, age, look, voice, the rules their sound must always follow, the channels and the label. A portrait comes with it. If you only want a song, she skips this.

  2. You approve the lyrics your call

    One feeling per song. A chorus you can sing in one breath. One image per line, a concrete detail as proof. You read them in the chat and say yes, or ask for changes. This is the one thing you do.

  3. The song renders about 25 s

    Lyrics and sound rules become a composition plan: sections, styles, where the chorus lands, where the hook has to hit. It is performed and rendered in studio quality (44.1 kHz WAV). A title and cover art are made with it. A full three-minute song takes about 25 seconds, mastering included.

  4. Mastered automatically included

    Two passes to −14 LUFS with true peaks under −1.5 dBTP, the level streaming services expect, dynamics kept. Every song gets a measured profile: tempo, key, loudness.

  5. In your library, on the chart done

    The song lands in your library and, if there is one, on the artist's page. Share links open and play. Publish it and it joins the public chart. Download WAV or MP3 whenever you like, or send it straight into Wave or Video.

Under the hood

What the engine actually is

67 MUSIC has two layers. At the core is a music model that writes and performs complete arrangements with sung vocals. On top sits Shela's own production method: the lyrics, the plan, the mastering, the artist, the release. The model makes the sound; the method makes it a song; the licence is what lets us hand it to you.

training

Trained on licensed music only

Built together with record labels, publishers and artists. Among the partners are Believe, Kobalt Music and Merlin, the network that represents 30,000+ independent record labels, through deals that let artists and songwriters take part in shaping AI music models and the revenue from them. Not a catalogue scraped off the internet. That is the whole reason we can hand you the song for your own use, including commercially.

sound

An arrangement, not a preset

A full arrangement, played and sung, rendered at 44.1 kHz WAV, with vocals or purely instrumental. It can move from one genre to another inside a single song, hold fast rap and dense lyrics, and place sound effects inside the arrangement.

structure

Up to 10 minutes, section by section

Intro, verse, chorus, bridge: up to 30 sections, each with its own lyrics, styles and inline directions such as a guitar solo or a whispered line. That is what lets Shela plan the song before a single note is rendered.

voices

Sings in your language

English, Spanish, French, German, Japanese and many more: the vocal follows the language of the lyrics you approve. Our own releases are sung in Serbian. Ask for one singer or two in harmony, for a whisper or a belted chorus.

control

It understands studio talk

Genre, mood, instrumentation, tempo and production era. Bone-dry drums, plate reverb, a belted chorus, two singers harmonising in C, "the lyrics start at 15 seconds". BPM and key are held precisely enough to layer the result with your own material.

rework

Any section can be rewritten

Pick one part of a finished song and have just that part written and performed again, or extend it, or condition a new song on a 30-second slice of an old one. Uploaded audio steers the feel, the groove and the palette without copying the notes.

rights

What you may do with it

Songs you make here are yours to use, including commercially: online, in advertising, on social media, in podcasts and in games. Film and television are the exception for now, ask us about those. Every upload and every reference is screened for copyright before anything is made, and prompts naming real artists are rewritten into descriptive styles.

honesty

What it is not

The music is made by AI. Nothing is recorded live, no session musician stands in a room. Because it is a model, a similar song can be produced for someone else, so we do not promise yours is one of a kind. What is only yours: the persona, the lyrics you approved, and everything you do with the song afterwards in Wave.

The difference

What the engine gives you, and what 67 MUSIC adds

Anyone can send a prompt to a music model and get a track back. That track is a starting point, not a release. This is the work that sits between the two, and it is the part 67 MUSIC does for you.

The engine on its own67 MUSIC
The songa track from a prompt, or from a plan you build yourselfthe same engine, driven by a plan written from the lyrics you approved and the artist's own sound rules
The lyricsyou write them, or you get whatever a general model writesa lyrics-first method with measured rules: one feeling, a chorus you can sing in one breath, one image per line, a concrete detail as proof
The artistnone. Every song stands on its owna persona with a home city, a face, a voice and rules that hold, reused on every track under that name
Title, cover, portraitnone. You get audioa title and cover art with every song, a painted portrait for the artist
Masteringthe raw rendera two-pass master to −14 LUFS with true peaks under −1.5 dBTP, plus a measured profile: tempo, key, loudness
Stems and MIDInot part of the modelfour separated stems and a MIDI file for any song, ready for your own DAW
Editingnone. What comes out is what you haveWave: timeline, EQ, compressor, sidechain, mixer, on-screen keyboard and instruments, piano roll, step sequencer, 150 loops, chord detection, beat slicing, mastering assistant
Starting from your own soundan uploaded clip can steer the styleupload a recording, a demo or stems, edit the plan it extracts, regenerate one section, extend it, or get two versions to choose from
Videononea short-form editor that already knows the beat and the sung words: 58 effects, 42 transitions, karaoke captions, cuts on the beat
Releasea file on your diskan artist page, a public chart, share links that open and play, a library where every track keeps its cover
Playbackas renderednative 432 Hz playback, one tap, on by default
Where it happensan interface or an API you drive yourselfone conversation with Shela. She writes, generates, masters, names, illustrates and files the song, and sets the artist up
For the days you want more

Wave and Video

Most songs are finished when they land in your library. Wave is there for shaping the music yourself, Video for turning it into a clip. Both run in your browser, on every plan, and neither is required.

Wave

Open the song as separate tracks and make it yours

Wave takes any finished song, splits it into separate tracks and lets you change whatever you want. The original stays as it is; Wave saves a new version.

  • Split to stems: one tap turns the song into four tracks: drums, bass, vocals and everything else.
  • Rework the vocal: mute the original vocal stem, have a section written and sung again with new lyrics, or lay your own recording over it.
  • Add your own: other songs from your library, your recordings or sound files as new tracks.
  • Play and record: an on-screen keyboard, drum pads and your MIDI keyboard; a synth with 30 presets, grand and electric pianos, basses, mallets, strings and two drum kits, all licence-cleared; record with a click and count-in, notes snap to the grid.
  • Piano roll, Beat maker, loops: edit notes in a piano roll (quantize, swing, scale snap, MIDI import and export), program beats in a 16-step sequencer with patterns on the timeline, and drop in 150 licence-clear loops that follow your tempo and key.
  • Shape every track: EQ, compressor, sidechain, saturation, delay and reverb on each track, a mixer with a limiter on the master.
  • Chords, Break, tempo, Master: analyse the chords and key of any song, slice a beat into its hits and re-sequence it, let dropped files follow the project tempo, and bring the mix to streaming, club or podcast level with a mastering assistant.
  • Works on a phone. The whole studio fits a 6-inch screen.
Video

Turn the song into a video for TikTok, Reels and Shorts

Video already knows your song: its beat, its lyrics and the timing of every word. You add the pictures, or let it make them.

  • Bring your clips: add your own videos and photos, or generate clips from the music with a shot plan that follows the lyrics.
  • Cuts on the beat: the timeline shows every beat of the song. Cuts and effects snap to it, and Magic can cut the whole video to the beat for you.
  • Karaoke captions: the sung lyrics appear word by word, in time with the voice, in five caption styles.
  • 58 effects, 42 transitions, 10 looks: flashes, zooms, glitches, VHS, split screens, each placed exactly where you want it in the song.
  • Any format: 9:16 for TikTok and Reels, 1:1 for feeds, 16:9 for YouTube. Export a finished video file.
  • Works on a phone. Edit with your thumb, export, post.
Also included

Everything a release needs

persona

An artist that stays the same

Every song under one name keeps the same voice, sound and character, because the rules are written down once and reused.

432 hz

432 Hz playback, built in

One tap retunes the whole recording from 440 to 432 Hz. A few small clinical studies link 432 Hz music to a lower heart rate and less stress hormone than 440 Hz.

stems · midi

Open it in your own tools

Download the four stems and a MIDI file for any song. Studio WAV out. Your own DAW is welcome.

chart

A place where people hear it

Public artist pages, Explore and Top Songs, likes and plays, playlists, share links with previews. No account needed to listen.

Who it is for
  • Songwriters and independent artists who want a finished, mastered song from lyrics they approve.
  • Producers and DJs who want a track for a set, a vocal for their own beat, or ideas in a genre, fast.
  • Creators who need a song and a video for short-form posts, in the same afternoon.
  • Anyone inventing an artist, from a single character to a whole roster on a label.
Plain facts
  • The music is made by AI from what you say and the lyrics you approve. Nothing is recorded live.
  • You decide what goes public. Artists and songs are private until you publish them.
  • Songs are yours to use commercially, on free and paid plans. Film, television and game studios are the one exception for now.
  • 432 Hz is not a medical claim. The studies (PMID 31031095, 32401941, 39844155) are small and the mechanism is not explained.
  • Use is governed by the Terms.
Plans

Start free

Free$0
  • A short song every day
  • Wave, Video, the player and the chart included
  • Artist pages and share links
Pro$35 / month
  • All of Shela, not only 67 MUSIC: chat, agents, sites, connections
  • Music metered per second, generated video clips per clip
  • Credit packs that never expire on active accounts

Full details on the pricing page.

Try it

Say the first sentence today

Open Shela and write what you want: an artist, a track for your set, ideas in your genre. Approve the lyrics, and listen about half a minute later.

Comparing tools? See how 67 MUSIC stands against other AI music tools.