Show HN: Sonauto API – Generative music for developers

Pricing and Hobbyist Plans

• Discussion on how to offer a flat-rate “unlimited” hobbyist plan without abuse by larger customers.
• Suggestions: fast/slow tiered queues with limited high-priority requests and unlimited low-priority ones; leveraging natural limits (people can only listen so fast) and signals like play/pause or skip to prioritize computation.
• Concern about separating “prosumer” users who need fast iteration from background/low-priority use.

Model Quality, Architecture, and Positioning

• Sonauto is contrasted with Suno: Suno’s token-based LM is described as more consistent and “radio-safe” but less diverse; Sonauto’s diffusion model is said to be more varied and realistic, especially vocals, but less consistent per generation.
• Some users say Sonauto sounds more “real” and varied; others find certain outputs derivative, low-quality, or “supermarket background music.”
• Technical notes: audio models share structure with image diffusion but need far more compression and precise rhythm/lyric placement. Links to Stable Audio and MusicGen papers are shared.

Use Cases and Workflow Integration

• Proposed uses: background music for games/videos, playlist transitions, infinite personalized work music streams, “Weird Al”-style parody covers, generating riffs or transitions, or inspiration material for human musicians.
• Musicians express strong interest in additive tools: AI drums over existing tracks, accompaniment generation, style-preserving vocal/lyric additions, DAW-friendly plugins.
• Some argue the value is in quick, personal, “gift” songs or experimentation; critics say the lack of effort makes such gifts “meaningless.”

Ethical, Legal, and Copyright Debates

• Fierce disagreement over training on copyrighted music:
  • One side frames it as analogous to human influence, a way to “bring musical knowledge to everyone,” and rejects major-label control of musical history.
  • The other side calls it piracy/IP laundering, exploitation of unconsenting artists, and “slop” that commoditizes human creativity; emphasizes scale differences and the model’s total dependence on others’ work.
• Questions about whether AI outputs should be copyrightable; Sonauto’s TOS assigns any rights in outputs to users but also reserves broad reuse rights for the company.

Impact on Musicians and Culture

• Concerns: devaluation of music, displacement of teaching and session work, enshitification of streaming catalogs with cheap AI filler, and loss of incentives for indie artists.
• Counterarguments: music tech has always changed practice (player pianos, drum machines), AI can resurrect/merge genres and lower entry barriers, and art can have value even if only the creator hears it.