Sonauto is the free AI music generator built on the Melodia latent diffusion model — text prompts, tags, or lyrics in, complete tracks with vocals and instrumentals streaming out in about 15 seconds. Unlimited runs. Full commercial rights included. Cornell + Y Combinator team.
The Sonauto loop is compressed to three operations: describe the song in plain language or tags, the Melodia latent diffusion model generates the full track in about 15 seconds, and everything you make lands in your account with full commercial rights attached.
Type a mood ("melancholic lo-fi with warm piano"), pick from thousands of style tags (harpsichord, slap house, post-grunge), drop in your own lyrics — or let Auto Tags suggest the vibe. Advanced mode adds Rhythm Assist, Style Strength and Duration Control sliders.
The proprietary latent diffusion model — Melodia — runs through 40 denoising steps to build the full track. Vocals plus instrumentation. First playback in about 15 seconds; keeps generating in the background while you listen. Two alt takes per prompt.
Full commercial rights on every tier — free or paid. Post to YouTube, release to Spotify and Apple Music, drop into a video game, license to a client. Section 8 of the TOS assigns output ownership to you, not Sonauto. No licensing fees, no extra clearances.
Every AI music tool generates full songs now. What most of them do not do is let you edit those songs like a producer: rewrite a single section, split into stems, load into your DAW. Sonauto ships all three, on the free tier, with commercial rights attached.
Love the verse, hate the chorus? Inpainting regenerates just the section you highlight — keeping everything else intact. Every other generator makes you regenerate the whole track and hope for a better roll. Sonauto does not.
Split any Sonauto-generated track into individual stems and export to WAV. Drop into Ableton, Logic, FL Studio — arrange around the vocal, replace the drums, mix cleaner. Full remix pipeline in one tool, from prompt to production-ready stems.
Section 8 of the Sonauto TOS assigns output ownership to you. Ship the track on Spotify with your name on it, sell to a client, put it in a video ad, use as a game soundtrack. Where Suno gates commercial rights behind Premier ($30/mo), Sonauto includes them on free.
Suno and Udio own the vocal-quality benchmark. Sonauto owns the producer workflow — edit, split, and ship, all on the free tier, with full commercial rights attached. That combination is what turns text-to-song from a novelty into a pipeline.
You need a soundtrack and cannot afford a composer. Free commercial rights + stem export mean you ship the game with music you own.
YouTube, TikTok, podcast intros. Instrumental toggle + fast generation + full rights = original music without royalty-free-library fatigue.
Generate a hook, split into stems, drop into Ableton, arrange around it. Inpainting + DAW export = full remix pipeline in one prompt.
Write lyrics, hear them sung back in 30+ languages, in any genre. Personalized birthday songs, wedding processionals, dog theme tunes.
Melodia is Sonauto's proprietary latent diffusion architecture — the same generative approach behind Stable Diffusion, applied to audio. 40 denoising steps take pure noise to a full track with vocals and instrumentation. v3 for quality, v2.2 stable for consistency, v3-Preview extends up to 4.5 minutes.
English, Spanish, Japanese, Korean, Hindi, French, German, Mandarin, Arabic... mix within one song.
Upload MIDI or audio clip — Melodia conditions on your chord progression or rhythmic pattern.
Harpsichord, slap house, post-grunge, talking blues. Auto Tags AI-suggests the vibe.
Community groups, curated playlists, staff picks, live radio events, remix pipeline.
REST API at api.sonauto.ai/v1/generations for programmatic access. Style Strength and Duration Control sliders on the client for fine-grained parameter tuning. Extend clips, replace sections, tweak BPM and key without leaving the app.
Sonauto plays the free-tier + producer-workflow end of the AI music category. Against Suno's polished vocals and Udio's mastering quality, the trade-offs are real and worth being upfront about.
| Sonauto | Suno v4.5 | Udio | Riffusion | |
|---|---|---|---|---|
| Free-tier daily cap | Unlimited | 10 songs / day | 10 songs / day | Unlimited |
| Commercial rights (free) | Included | Personal only | Personal only | Attribution |
| Vocal quality | Good · improving | Category leader | Category leader | Basic |
| Inpainting (section edit) | Full featured | Basic replace | Full featured | None |
| Stem separation + export | WAV stems | Premier tier only | Standard tier | None |
| Extend to full length | 4.5 min · v3-Preview | Extend feature | Extend feature | Short clips |
| Multi-language vocals | 30+ languages | 40+ languages | Multi-language | Limited |
| Developer API | api.sonauto.ai | No public API | No public API | API access |
| MIDI / audio input | Both | Audio upload | Audio upload | Text only |
| Best for | Free producer workflow | Best-sounding vocals | Vocal + mastering | Developer API |
Sources: Sonauto.ai (rebranded to Treblo) 2026 landing page, Suno v4.5 product pages Q1 2026, Udio pricing docs, Riffusion.com developer docs, cross-referenced reviews on Tunesona, Tad AI, 5App.
"The inpainting feature is what sold me. Every other AI music tool makes you regenerate the whole song if you do not like one section. Sonauto lets me rewrite just the chorus while keeping the verse I already love."
"I generated my entire game soundtrack with Sonauto's free tier. Full commercial rights mean I can ship the game without worrying about music licensing. That saved me at least $2,000 in composer fees."
"Honest gripe — vocal quality is not quite at Suno's level yet, especially on complex melodic phrases. But the free tier + commercial rights + stem export combo is unbeatable for shipping projects."
Sonauto was founded in 2023 by Ryan Tremblay, who calls himself the Chief Machine Learning Alchemist, and Hayden Housen, the co-founder and CTO. Both Cornell alumni, both music technologists working on the intersection of AI research and creative tooling. The company was backed by Y Combinator and set out to do one thing at production quality: build a proprietary latent diffusion model — Melodia — that could take a text prompt and return a complete song with vocals and instrumentation in fifteen seconds.
The product thesis is the opposite of Suno's. Where Suno has bet on polished vocals and a subscription-first monetization funnel — free tier capped at ten songs per day, commercial rights behind Premier at $30/month — Sonauto ships unlimited free generations and full commercial rights on every tier. The free version is not a demo; it is the whole product. That decision was deliberate. Ryan and Hayden bet that opening the model would attract more creators than gating it, and the producer-workflow tooling — inpainting for section-level rewrites, stem separation with DAW export, MIDI and audio conditioning, developer API access — is what keeps them there. The 2026 rebrand to Treblo (same team, same songs, same account per their public FAQ) was a naming decision. The Melodia model, the workflow, the commercial-rights posture — all unchanged.
The honest trade-offs deserve naming. Vocal quality is not quite at Suno v4.5's level yet on complex melodic phrases — reviewers note occasional artifacts on sustained vocals and high registers. The AI can produce generic-feeling tracks when prompts are too simple; the Advanced mode and MIDI conditioning are there for a reason. Some advanced features have a learning curve — Rhythm Assist, section inpainting, and Style Strength calibration are not one-click. The rebrand is a real transition — App Store and Play Store listings may still show either name, and community pages are still moving over. Fine-grained musical control is limited vs a traditional DAW — Sonauto is a generation and edit tool, not a mixing/mastering suite. API pricing exists for developers running at scale, and terms of service specifically limit the free tier to personal, non-commercial or internal business use — commercial redistribution at scale needs a paid tier.
What you get in return is the AI music generator with the most producer-friendly free tier in the category. Melodia's latent diffusion model on the base plan, inpainting and stem export without a paywall, full commercial rights whether you paid $0 or $30 this month, and a developer API for anyone shipping generation into a product. If the workflow is "prompt → edit → ship," Sonauto is the tool built end-to-end for it.
latent diffusion model — the same generative architecture family behind Stable Diffusion for images, applied to audio. It works by learning to reverse a noising process: starting from pure noise, the model iteratively denoises across ~40 steps to produce a full track with vocals and instrumentation. v3 is the current quality engine, v3-Preview extends up to 4.5 minutes, v2.2 remains available as the stable engine for consistent output.api.sonauto.ai/v1/generations with a Bearer token header. POST prompt, tags or lyrics as JSON. Subscription plans exist for production usage; free-tier limits apply for evaluation. Useful for shipping generation into your own product — background music services, generative games, marketing tools.Unlimited on the free tier. Commercial rights included. Melodia v3 in your pocket. There is nothing to lose by trying it — no card, no daily cap, no catch.