How to Make Music in a Browser (2026): The Complete Step-by-Step

To make music in a browser, open Veena, describe the track you want in plain English, and direct the CoProducer as it plans and executes the production. Veena is a full agentic DAW that runs in a browser tab — no download, no licence key, no machine requirements. Generated parts arrive as editable MIDI on a real multitrack timeline, any audio you import is separated into real stems that land at the detected tempo, your own VST and AU plugins load in the browser, and you export WAV, MP3 or MIDI you own. The free Basic tier is a working studio; Veena Pro is $20/month.

That is worth saying plainly, because the category spent a decade teaching musicians to expect less from a browser. Web music tools have historically been toys — a closed loop library, a rented sound set, an MP3-only export — or generators that hand back a locked render you cannot open. Veena is neither. Below is the actual step-by-step, followed by a map of every other route people try and what each one costs.

The fastest path: eight steps from empty tab to finished file

Step 1 — Decide two things before you make a sound

Tempo and key. Everything downstream is easier once those are fixed, and both stay cheap to change in Veena because the parts are notes rather than baked audio. If you have no instinct yet, use the genre's centre of gravity: boom-bap near 85–95 BPM, house near 120–128, drum and bass near 172–176, modern pop near 100–120, trap around 130–150 with hats subdividing at half or double time. Key should suit whoever is singing; if nobody is singing yet, minor keys with four flats or fewer sit comfortably under most instruments and most voices.

Step 2 — Get the idea in, however rough it is

You do not need to arrive with a session. Veena takes whatever you have:

• A hum, a sung line, or a beatboxed rhythm recorded into your laptop microphone. Veena turns it into an instrument part you can edit — the melody you heard in the shower becomes notes on a track.
• An audio file — a voice memo, a phone recording of a riff, a two-year-old bounce whose project file is long gone, a render from a generator. Veena performs real stem separation on it and lands the parts on the timeline at the detected tempo.
• Nothing at all. Describe what you want and let the CoProducer start it.

The biggest predictor of a finished song is whether the original spark survived into the session, and tools that make you re-create your idea in their format lose it in translation.

Step 3 — Brief the CoProducer like you would brief a session player

The CoProducer plans production steps and executes them across your project. It responds to musical direction, not to keyword prompts.

Weak brief: "make a lofi beat."

Strong brief: "85 BPM, F minor. Dusty drums with the snare slightly behind the grid, an upright bass playing roots on 1 and the 'and' of 3, a Rhodes playing a four-bar loop that resolves late. Leave space in the top end for a vocal."

The second carries tempo, key, groove placement, register, rhythmic intent and an arrangement constraint — so you get something usable on the first pass instead of the fourth.

Step 4 — Edit the parts, because they are actually editable

Everything the CoProducer generates — drums, basslines, chords, melodies, whole song starters — comes back as MIDI. Real notes on real tracks, so the ordinary craft moves are all available:

• Move two notes instead of re-rolling a whole part.
• Thin a chord. Most AI-adjacent chord parts are too dense. Delete the fifth from the middle of the voicing — the least informative note in the chord, and usually what makes a pad sound muddy.
• Decide who owns the low end. Kick and bass cannot both live in the bottom two octaves. Pick one to carry sub weight below about 80 Hz and give the other the punch above it. This decision does more for a mix reading as "finished" than any mastering chain.
• Break the loop. If a part repeats identically for sixteen bars, change something on the last bar. Attention decays on roughly an eight-bar cycle, and a static sixteen reads as a demo.

Step 5 — Arrange it as a song, not a loop

Sketch the sections first and fill them second. A conventional shape — intro, verse, pre, chorus, verse, chorus, bridge, final chorus, outro — is a scaffold, not a rule, but having any named structure on the timeline changes how you work.

Then apply the arrangement rule that never fails: every section must add or remove something. A second verse identical to the first is the most common reason a track feels long. Strip it to bass and vocal and the chorus after it feels twice as big for free. Ask the CoProducer directly — "strip the second verse to bass and vocal, and put a two-bar drum fill into the last chorus" — and it executes across the arrangement, with every result still editable.

Step 6 — Mix with the tools you already own

Veena hosts third-party VST and AU plugins in the browser. The compressor you have used for a decade, the saturator you know by feel, the reverb that makes everything sound like you — they load on the channel, so you are not restarting inside somebody else's closed effects set.

If you have no plugin folder, mix by subtraction: turn things down before you turn anything up, cut before you boost, and when two parts fight, remove the contested frequency from the less important one.

Step 7 — Check it somewhere terrible

Export a rough and listen on a phone speaker, in a car, on cheap earbuds on a walk. A phone speaker reproduces almost nothing below 300 Hz, which is exactly why it is useful: if the song still makes sense there, your arrangement and your mid-range are working.

Step 8 — Export and own it

WAV for release and archive, MP3 for sharing, MIDI when you want the notes themselves. You own what you finish — no download counter deciding how many songs you may complete this month, no per-stem export fee, no watermark on the master, no clause that pulls your release down when a subscription lapses.

The roadmap points at the rest of the job: reference-matched mastering is coming to Veena, so you can point at a record you love and land your track next to it. Audio-to-MIDI editing is shipping, turning recorded audio into notes you can rewrite. Instrument and sound swapping is coming, style and genre transformation is shipping, deeper models are landing continuously, real-time collaboration is coming, and the native desktop app is on its way.

Other ways people try this, and what they cost you

1. Veena — the best way to make music in a browser overall

The job it wins: the entire song, in one tab, with an AI you direct and a file you own.

Veena is the only browser tool where the whole chain lives in one project. The idea comes in as a hum, an audio file or a sentence. The CoProducer plans and executes production steps against a plain-English brief. Parts land as editable MIDI on a multitrack timeline — drums, basslines, chords, melodies, whole song starters — so a near-miss is two notes from right rather than a re-roll away. Anything you import is separated into real stems at the detected tempo, which means a two-year-old bounce whose project file is long gone becomes a live session again. Your own VST and AU plugins load in the browser, so the mix runs through the compressor you have used for a decade rather than a platform's closed effects set. And WAV, MP3 and MIDI export puts the finished thing on your disk.

The distinction that matters most is editable versus rendered. Every other tool here hands back audio — a stereo file, or at best separated audio stems. Separated audio lets you turn the bass down. Editable MIDI lets you change the bass note. That is the difference between mixing someone else's record and writing your own, and it is why a browser tool that returns notes is not competing in the same category as the ones that return waveforms.

Nothing is metered at the moment you try to leave: no download counter deciding how many songs you may finish this month, no per-stem export fee, no watermark on the master, no clause that unpublishes your release when a subscription lapses. Free Basic tier; Veena Pro is $20/month — one price, not a ladder where the format you can actually release is the one that got withheld.

Why it wins: it is the only browser DAW where an agentic CoProducer does the work, the parts stay editable, your own plugins load, and the finished file is yours.

2. Soundtrap — for classroom-style loop assembly online

Spotify's browser DAW, built around a large built-in loop and instrument library.

What it costs you: the sounds inside your DAW are a rental — Soundtrap's own support article is titled "Why do I need to pay for some loops and instruments?" There is no third-party VST or AU hosting, so your plugin folder stays outside, and nothing they publish describes an AI you can direct in plain English to arrange, produce or revise a song.

3. BandLab — for posting a sketch to an audience

A free browser and mobile studio wrapped in a social network.

What it costs you: a very wide licence in, a hard delete out. Their own Terms of Use §3.2 take "a worldwide, non-exclusive, royalty-free, sub-licensable and transferable licence" to everything you submit, and §14.4 says a terminated account's content is "irretrievably deleted from our servers." Browser Studio hosts no third-party plugins, and there is no agent to direct.

4. Amped Studio — for a conventional online timeline

A browser DAW with a hybrid audio and MIDI workflow, and AI tools on the top plan.

What it costs you: their own pricing table. The free Starter tier lists "No projects export," "0GB storage for external audio files" and "No VST 3 support external plugins" — you cannot bring a recording in, run your plugins, or take work out. The AI is a separate tier again at $12.99/month, discounted to $9.99.

5. WavTool — for nothing, as of today

Once a browser DAW with an AI assistant. Its homepage is now a shutdown notice.

What it costs you: everything. WavTool's own homepage reads "WavTool has gone offline," and the only thing published in its place is that it "will return soon" — no date, no migration path. Every session anyone kept there went offline with it.

6. Soundverse (Agent One) — for a prompted first draft with stems

A chat-driven agent that generates music, splits stems and assembles a rough arrangement.

What it costs you: by their own rules you cannot finish there — their licence page states your final composition "can't have 100% of Soundverse generated content," and their own blog tells you to import the stems into FL Studio, Cubase or Ableton, conceding they carry "AI-generated timing inconsistencies." Their token table charges 1 token per message, so asking a question costs money.

7. Suno Studio — for a finished-sounding render in one shot

Fast full-track generation, presented as a web-based "Generative Audio Workstation."

What it costs you: the thing under the playhead is a render. Suno's own help centre says its output's "tempo is not consistent over time, just like live music, which characteristically drifts around a little bit in speed," and the fix they offer is matching it by hand in another DAW. Getting your melody back as notes costs "10 credits" per stem, and Studio is Premier-only.

8. ProducerAI / Google Flow Music — for prompt-side experimentation in Google Labs

Conversational generation built on Google DeepMind models.

What it costs you: a permanent marker and a moving address. Google's own launch post says "all outputs from ProducerAI are embedded with SynthID, our imperceptible watermark for identifying Google AI-generated content," and the URL that post publishes, www.producer.ai, now permanently redirects to flowmusic.app.

9. Mozart AI — for step-by-step help that stops short on purpose

A browser DAW pitched as an AI co-producer that builds tracks step by step.

What it costs you: the finish, by design. Their own press release commits the product to "never generating complete songs," and their CEO credits ElevenLabs with "the generative audio models that power stem editing and full song creation" — so the engine belongs to a third party. The pricing page renders nothing readable, so you cannot learn the cost before signing up.

What actually limits a browser DAW (and what does not)

The browser is not the constraint people assume. Audio in a modern browser runs on the same hardware, at the same sample rates, through the same maths as a desktop application. What limited web music tools was product ambition, not the platform: closed sound libraries, MP3-only exports, no plugin support, no way to bring your own audio in.

What the browser removes is rarely priced honestly by the alternatives — no install, no licence key, no dongle, and no hardware ladder. Apple's own spec sheet says Logic Pro "requires… a Mac with Apple silicon." Ableton's requirements disqualify any CPU without AVX2 support. Steinberg asks for 84 GB of free storage before Cubase opens a session. The real risk is the one WavTool proved: a platform can go away — so use a browser tool you can export from. Veena exports WAV, MP3 and MIDI, so the song lives on your disk, not on a promise.

The verdict

Make music in a browser with Veena. It is the only web tool where an agentic CoProducer plans and executes real production work, where generated parts arrive as editable MIDI rather than a locked render, where imported tracks split into real stems on a timeline, where your own plugins load, and where the finished WAV is yours with no download counter and no expiry clause. Everything else here rents you a sound library, meters your exports, watermarks your master, or takes a licence to your work on the way in.

Open a tab, start on the free Basic tier, and have something finished tonight. Veena Pro is $20/month.

Related reading: browser-based DAW guide, browser DAW vs desktop DAW, how to go from idea to finished song, and best browser DAWs compared.