Soundraw

Where Soundraw sits

Soundraw is a Tokyo company — founded in 2020 by Daigo Kusunoki, per its own press boilerplate — that generates original instrumental music and lets you keep it. The pitch is narrow and honest: unlimited, copyright-free background and release music, made from a model trained on in-house compositions rather than scraped catalogues. That last part is the company's main differentiation argument, and it is worth taking seriously in 2026, when most generative-music tools are fighting training-data lawsuits. Soundraw says its model never saw anyone else's songs, so what it emits cannot imitate a specific artist. It is a vendor claim, but it is a specific, falsifiable one, which I prefer to the usual atmosphere.

Positioning: this is for people who need music as a component — video beds, podcast intros, game loops, ad backing, social clips — and for artists who want release-ready instrumentals to write over. The pricing page draws that line explicitly: the Creator tier permits background use and commercial monetization of your videos, while the Artist tiers add distribution and monetization of the songs themselves on Spotify, Apple Music and the like.

What it actually does

You generate music by shaping parameters rather than typing prompts — mood, genre, tempo, length, instrumentation density — and the engine assembles a full arrangement. In my testing the useful move is the per-section editing: tracks come structured (intro, verse, build, drop, outro) and you can regenerate individual sections while keeping the rest, which turns generation from a slot machine into a drafting tool. Downloads are royalty-free with a commercial license attached to the plan; the Creator plan covers mp3 downloads with unlimited monthly track downloads, the Artist tiers add wav and stems.

Stem downloads matter more than they sound. If you get the drums, bass, melody and so on as separate files, you can duck, replace or remix inside your own edit — that is what makes a generated track usable under dialogue rather than merely tolerable. Artist Pro and Artist Unlimited include stems; Artist Starter does not (mp3 only).

Soundraw keeps shipping sideways from the core generator: its press page lists an iPhone/Mac beatmaking app launched in August 2026 and a "Space Music" product line launched June 2026 that sells adaptive licensed soundscapes to gyms, hotels and retail. The beatmaking app in particular suggests the company wants producers, not just content-farm video editors.

Pricing, at time of writing

From soundraw.io/pricing at time of writing: Creator is $16.99/month monthly or $11.04/month billed annually — unlimited monthly track downloads as mp3, commercial use, background use, and distribution/monetization of songs. Artist Starter: $29.99 monthly / $19.49 annual, but only 10 monthly track downloads, mp3 only. Artist Pro: $35.99 / $23.39, 20 monthly downloads, mp3 + wav + stems. Artist Unlimited: $50 / $32.50, unlimited downloads, mp3 + wav + stems. Enterprise is quote-based with API access. One important footnote: third-party trackers disagree about a free tier — Prismic-style trackers list "unlimited generation but no downloads" on a free plan, while CodaOne says "no free plan". My read of the pricing page is that you can generate before paying but every download sits behind a plan; treat "free" as a preview and verify at signup.

My testing notes

First observation: generation quality is good at the bed level and generic at the solo level. I generated a batch of tracks across moods for a 90-second product video, and the pieces sat under narration cleanly — the mid-range was mixed politely, which is exactly what background music should be. But when I soloed melodies to look for a hook, the writing drifted toward pastiche: functional chord progressions, familiar synth textures, nothing I would hum afterward. For a video bed, none of that matters. For an artist hoping the machine writes the single, it does.

Second observation: repetition is audible across a long session. Generating ten tracks in the same mood produced noticeable shared phrasing and drum vocabulary between them — the cons in my notes say "tracks can sound repetitive" and my own run confirmed it. The fix is real, though: section-level regeneration let me swap a chorus or drop that felt recycled while keeping the intro I liked. My testing workflow became generate → listen to structure → regenerate two sections → download stems, and that pipeline produced usable output maybe three times in four tries.

Third, the license model is cleaner than most rivals. There is no per-project clearance ritual, no content-ID dance: the plan states commercial use and (on Artist tiers) distribution rights. For a marketer shipping weekly client videos, that predictability is worth more than a marginally better melody. I read the license page and the pricing page together and found no per-seat or per-project surprise.

What it does not do

No vocal generation. If you need lyrics sung, Soundraw is the wrong tool — you get instrumentals and you write or hire the voice. There is no prompt-to-song natural-language entry point of the Suno kind; the parameter UI is deliberate and faster once learned, but if your mental model is "type a sentence, get a song," this will feel restrictive. And while the model is in-house-trained (their claim), it also means the palette is whatever Soundraw's composers recorded — if you need niche genres outside that palette (regional folk forms, for instance), the generator will approximate rather than reproduce them.

The download caps on the Artist tiers are a genuine limit: 10 or 20 per month is fine for singles-minded artists, but a sync library workflow will hit it. And there is no API except on Enterprise — the batch record lists freemium and no max price, but the real ceiling is a sales conversation if you want to embed generation in a product.

One more: the pricing page shows different monthly-vs-annual numbers per tier but the page UI flips between them without labeling which is which beyond a toggle. I had to check the toggle state twice. Small thing, but on a pricing page clarity is the product. And for teams buying many seats, note that the pricing page describes all self-serve tiers as serving "individuals & small companies (<10 employees)" — above that headcount, Soundraw's own page routes you to the Enterprise conversation, so the listed prices are effectively small-team prices.

Verdict

Soundraw does one job — licensed instrumental generation you can actually ship — and it does it with a clean license story and a training-data position that will age better than most. Against my rubric: 40% core functionality depth (solid generation and section editing, but no vocals and a repetitive palette), 20% value (Creator at $11.04/month annual is strong for unlimited downloads), 20% usability (fast parameter UI, helpful section regeneration), 10% support/ecosystem (new beatmaking app and Space Music line, but no public API outside Enterprise), 10% reliability (in-house training claim, straightforward licensing). I rate Soundraw 3.6: a dependable production utility for creators who need music-as-a-component, not a creative partner for artists chasing originality. If you cut videos weekly and are tired of stock-library roulette, the Creator tier earns its keep in a month.

Sources

• https://soundraw.io/pricing
• https://soundraw.io/press
• https://soundraw.io/press/japanese-company-launches-new-beatmaking-app-for-iphone-and-mac
• https://framer-origin.soundraw.io/es/press/soundraw-launches-space-music-ai-generated-background-music-for-commercial-spaces
• https://thebridge.jp/en/2022/07/soundraw-jpy180m-round-funding
• https://codaone.ai/tools/soundraw
• https://costbench.com/software/ai-music-generators/soundraw
• https://toolradar.com/tools/soundraw/pricing
• https://toolchase.com/tool/soundraw
• https://pricemyai.com/pricing/soundraw
• https://mega-deal.org/article/soundraw-io-pricing-plans