Treblo

Yes. On June 3, 2026 the team announced that Sonauto had become Treblo and moved to treblo.com. Accounts, logins, songs and shared links carried over, and sonauto.ai links redirect to the new domain. The terms and API terms still name the company as Sonauto, Inc., and the Android listing still shows Sonauto as the developer.

Developers need one change: the API host is now api.treblo.com. The announcement says the old api.sonauto.ai host still resolves for now but asks integrations to migrate, so a hard-coded old host is a future outage risk rather than an immediate one.

The homepage describes the free tier as unlimited songs on the latest model, Melodia v3, with no daily limits and no first-three-free trial. Its comparison table lists generation, inpainting, extending and stem separation on free accounts, plus downloads in OGG, MP3, WAV, FLAC and M4A. That table is the provider’s own summary of competitors, dated September 25, 2026.

No paid consumer plan appears on the public pages, but the terms keep a section on subscriptions that renew automatically. If a paid option appears in the app, read its renewal terms before accepting it. The About page describes a team of four people, and the rename post says they still answer their own support email.

Section 8 of the terms says you own whatever rights may exist in the music you generate, and Sonauto, Inc. assigns its own rights in those outputs to you. The commercial-use restrictions elsewhere in the terms do not limit your use of your outputs. The homepage FAQ adds that songs can go into videos and posts, as long as what you put into them is yours.

• Outputs may not be unique; another user can receive a similar song, and the assignment does not cover theirs.
• The provider makes no promise that a song is copyrightable or lawful to use.
• You are responsible for checking that your use of a song complies with the law and the terms.

The same section grants Sonauto, Inc. a perpetual, irrevocable license to your outputs and to the prompts, lyrics and other content you submit to generate them. The stated purposes include developing and improving the service and its AI and machine learning models. There is no opt-out described for consumer accounts.

The API terms take the opposite position for developer traffic: API inputs and outputs are not used to train or fine-tune models. If unreleased lyrics matter, the API route is the one with a written training exclusion. Publishing a song also counts as a Contribution, which grants a separate license to use it, including your name and voice, for commercial and advertising purposes.

The terms prohibit uploading a recording of an identifiable person’s voice, image or likeness unless you are that person or have their consent. They also prohibit claiming that a song was created, performed or endorsed by an identifiable artist, or titling and promoting it as by, inspired by or sounding like a named artist without consent.

Uploads that contain third-party copyrighted material need permission from the rights holder. For covers or remixes, start from your own recordings or audio you are licensed to use, and describe the style with genre tags rather than artist names.

The developer pricing page charges 100 credits per song, and a request with num_songs set to 2 costs 150 credits. Plans are monthly, from Starter to Enterprise, with overage billed pay-as-you-go. The API terms allow commercial use in your own products but forbid using API outputs to train music-generation models.

• Free trial: 3,000 credits after verifying a card.
• Starter: $11 a month for 20,000 credits.
• Pro: $88 a month for 160,000 credits.
• Scale: $330 a month for 660,000 credits, with cheaper overage.
• Every user-facing integration must show a visible Treblo attribution unless a separate agreement waives it.

1. Switch lyrics to instrumental and choose three tags, such as lo-fi, 2020s and guitar.
2. Generate two versions and keep the stronger one.
3. Extend or regenerate a section until the track passes 60 seconds.
4. Download WAV and MP3 and open both in an audio editor.
5. Check duration, clipping, stray vocals and whether the ending can be faded cleanly.

Accept a downloadable file of at least 60 seconds with no vocals and no audible artifacts at the edit point. Reject a track that includes sung words, stops abruptly or cannot be downloaded. No audio result is reported until the task is run.

To clean up or master recorded audio rather than generate it, see Auphonic or Adobe Podcast. For spoken voiceover instead of songs, compare ElevenLabs.