Launch HN: Soundry AI (YC W24) – Music sample generator for music creators

Original Article ↗ Hacker News Discussion ↗

Product concept and roadmap

  • Tool generates AI-based music samples (loops and one-shots), positioned as “ingredients” rather than full tracks.
  • Roadmap: move from samples → stems (separate instrument tracks) → full stem generation → distribution platform.
  • Next major feature is stem separation/generation and editable stems. MIDI generation is under consideration due to user demand.

Pricing and business model

  • Subscription: $10/month or $100/year for unlimited downloads.
  • Multiple users criticize hidden pricing that appears only after signup; founders say pricing will be added to the landing page.
  • Some feel cost is high vs one‑time-purchase plugins/DAWs; others think it’s reasonable relative to subscription sample libraries.
  • Questions about market size and VC-scale growth; team claims a large TAM by counting current musicians and potential new ones.

Technical approach and quality

  • Uses a custom autoencoder rather than off‑the‑shelf Encodec to better handle silence in samples and avoid noise artifacts.
  • Generation latency: text‑to‑sound ~10 seconds; “Infinite Sample Packs” feel instant because outputs are precomputed.
  • Some see it as “AI-powered Splice”; concerns that “infinite” samples may be generic and time‑wasting if quality varies.

Control, workflow, and target users

  • Strong interest in controllability: editable stems, one‑shots, MIDI, and parameter-level control vs black‑box full tracks.
  • Several musicians want tools that:
    • Generate drum kits + patterns, not just loops.
    • Coach novices and configure synths from reference tracks.
    • Convert samples into synth patches or SYSEX for hardware.
  • Debate over whether the product solves real problems for serious producers vs mainly helping beginners who may never level up.

Creativity, ethics, and “soul” of music

  • Enthusiasts see it as another creative tool, akin to drum machines or samplers, enabling faster idea capture and new sounds.
  • Skeptics worry it encourages generic, “artless” music and over-reliance on AI; some reject AI in art altogether.
  • Counterpoint: today’s music is already heavily algorithm-driven; artists will appropriate any tool for expressive work.

UX issues and misc. feedback

  • Requests for: API access, mastering features, humming/audio prompts, more generous free tier, and better clarity on loops vs samples.
  • Reports of minor bugs (e.g., prompt parsing) that were quickly fixed.