Write the lyrics. Hear the vocal.

AI vocal hooks for electronic tracks. Describe the voice or bring a short reference, add your lyrics and download every take.

Create a vocal

Reference

pressure-hook.mp310.2 s · Saved reference

Positive styles

male vocals, deep husky male voice, rhythmic spoken-sung hook, tech house, dark, confident, short syncopated phrases, English lyrics

Negative styles

female vocals, backing vocals, choir, whispering, vocoder

Lyrics

[Hook] Lock the door, let the bassline run Nobody leaves till the night is done Eyes on me, keep the tension tight Say less, move, we own the night

Settings: 8 bars at 128 BPM in G minor, any voice. Generate: about 15 seconds of audio for 10 credits.

Activity 4

A real session: the reference is a 10-second cut of Pressure, one of our own takes, and every vocal in Activity was generated from it. A reference guides a new performance; it doesn’t clone the singer or isolate the original. References are part of Pro.

Three steps, one screen.

You describe the voice and write the lyrics. Vocal Lab asks for a vocal-only take, with no instruments, so you can try it in your track.

  1. Describe the voice

    In the Prompt field, say in English how the voice should sound: tone, gender, mood and delivery. Stuck? Click an example like Pressure and edit it. On Pro, + Reference lets you start from a 10–30 second clip you have permission to use.

    Prompt · Pressure example
    Vocal in English, deep and slightly husky. Tech house. Rhythmic spoken-sung hook with short syncopated phrases and pauses; softer breakdown from the same voice.
  2. Write the lyrics

    Any language, up to 2,000 characters. Mark the parts with [Hook] and [Breakdown], and aim for a 15-second hook, not a whole song.

    Lyrics
    [Hook]
    No small talk, let the pressure speak
    One step closer, move it underneath
    
    [Breakdown]
    Hold that look, let the silence drop
    Stay right here until the lights come up
  3. Generate, listen, download

    Pick 4 or 8 bars, the BPM, the key and the voice. Generate shows the length and credits before you start: 8 bars at 128 BPM is 15 seconds for 10 credits. The take lands in Activity, ready to play, rate and download.

Pricing · Private beta

Start from words.
Or from your sound.

Basic generates from your prompt and lyrics. Pro also takes a short audio reference to guide the voice.

Monthly prices in BRL, previewed for the beta. Access is set up by hand; there is no checkout or automatic subscription yet.

Basic

For writing from text.

R$49.90/ month

500 credits per monthly cycle

No audio reference Text prompt and lyrics only.

  • Describe the vocal and write your lyrics
  • Generate new vocal takes from text
  • Listen, compare and download your takes

Up to 50 takes of 15 seconds a month.

Request Basic access

Pro

With audio reference

For starting from a sound you already have.

R$99.90/ month

1,000 credits per monthly cycle

Audio reference included Upload a 10–30 s clip to guide your takes.

  • Everything in Basic
  • Styles suggested from your reference, ready to edit
  • Reuse saved references for more variations

For example: 1 new reference + 98 takes of 15 seconds with it.

Request Pro access

Upload once.
Keep exploring.

Vocal Lab credits cover new references and generation. Reusing a saved reference has no new upload fee.

New reference · up to 30 s
20 credits
Generate a 15-second take
10 credits
Listen or download again
0 credits

Generation uses 40 credits per requested minute, rounded up per take. Each new variation is a new generation. Planned allowances renew monthly without rollover. These credits are used only within Vocal Lab.

Use audio you have permission to upload. A reference guides a newly generated performance; it does not extract the original vocal or guarantee the same singer.

In your browser. Soon in your DAW.

In your browser

Nothing to install: sign in to write, listen and download your takes.

  • Web beta
  • Nothing to install

Vocal Lab VST

Vocal Lab is coming to your DAW as a VST plugin. Create vocals from your direction and lyrics, right where you make music.

  • VST plugin
  • Coming soon

Straight answers.

What does an audio reference do?

It guides a newly generated vocal take alongside your styles and lyrics, and the Studio suggests styles from it that you can edit. It does not isolate the original vocal or clone a singer exactly. In the Studio, use an authorized clip of 10 to 30 seconds.

Is it the same singer every time?

No. Each take is a new performance that matches your description. A fixed singer identity isn’t part of the beta.

Which languages can it sing?

Write the prompt and styles in English. Lyrics can be in other languages too.

Will the take be dry and in key?

Vocal Lab asks the model for an a cappella take in your key and tempo. It’s a request, not a guarantee: isolation can leave artifacts and timing can drift, so check every take by ear.

Does it work offline?

No. Generation happens in the cloud.

What makes the audio?

Vocal Lab sends your direction and lyrics to a cloud generation service from our server. Service credentials never reach your browser or the device.

How do I get in?

Vocal Lab is a private beta. Use Request beta access to ask for an account; password resets go through the same message.

Give your next track a voice.

Vocal Lab is in private beta. We open accounts by hand, one producer at a time.

Request beta accessHave an account? Sign inListen again