AssetStore Price down information

beta

Sound Finder is a smart audio browser that listens to your files.
A search finds the sounds closest to what you describe, even when no file name contains your words.⚠️ Requires a GPU with compute shaders (SystemInfo.supportsComputeShaders), no CPU fallback. Tested on Windows with a dedicated GPU.⚠️ Memory: about 6 GB of VRAM and RAM combined, not 6 GB of each. What doesn't fit in VRAM moves to RAM: same results, slower text search. Editor-only, works with Built-in, URP and HDRP. Tested and working on a device with 8GB RAM and 512MB VRAM.Do you already own the sound you're about to download?With file names alone you can't tell. Sound Finder finds sounds by what they sound like, not what they're named.Type a plain-English description - "punchy kick", "glass shatter", "warm analog pad" - and get ranked matches from your own audio files in seconds. No tagging, no manual metadata, no naming discipline required.Can't put it into words? Search another way:Drag a reference clip onto the search bar to find sounds like itHold the mic and make the sound: (hum a melody, tap a rhythm, do beatbox) and search by what you recordedRecord your PC's audio: a Youtube clip, a game, a reference track playing through your speakers... find the closest matches in your library!It works on the audio already in your project and on external sample libraries. Point it at a folder anywhere on disk, search it in place, and the file is copied into your project only when you pick it. A 100 GB SFX collection never bloats your repo.✅ Everything runs 100% on your machine: no cloud service, no API keys, no subscriptions, no per-search cost. Your audio never leaves your computer, and it works offline.✅ Nothing ships in your builds, the entire pipeline is 100% editor only!Describe-it search: type what you're after in plain language and get ranked results.Search by example: drag any clip to find similar sounds.Search by recording: capture from a microphone or from your system's audio output and search by it.Project + external library search: index audio inside your project or any folder on disk; external picks are copied in on demand (WAV, AIFF, MP3, OGG).In-Inspector picker: replaces the default picker on AudioClip / AudioResource fields; AudioSource supported via right-click.Built-in clip cropping before import.Fully on-device & offline: all processing runs locally, GPU-accelerated. No server, no account, no data leaves your machine.Editor-only utility: nothing is added to your runtime build.⚠️ Requires a compute-shader-capable GPU (no CPU fallback). See compatibility notes.AI/ML is central to this tool's function. It bundles pre-trained third-party ML models - LAION CLAP, MeanAudio, and Google's T5-large text encoder - and can optionally download Stable Audio Open Small after explicit in-Editor user consent. All models run locally in the Unity Editor to power audio search; text search synthesizes audio internally for similarity matching, which is never saved, played back, or redistributed. The package also includes small pitch, onset, and melody-detection models trained from scratch by myself on public-domain and permissively-licensed data. No user data or audio leaves the machine.Some of the plugin's code was written with the assistance of AI coding tools and reviewed by myself.No AI-generated images, audio, or other media are distributed in this package. All bundled sample audio (Kenney, CC0) and UI icons (Lucide, ISC) are human-authored.