Show HN: LatentScore – Type a mood, get procedural/ambient music (open source)
Source: Hacker News
Introduction
I’ve used Generative.fm for years and love it, but I always wanted to describe what I was in the mood for instead of scrolling through presets. So I built LatentScore.
How It Works
- You type a text description of anything — from “mountain sunrise” to “neon city”.
- The system generates a procedural/ambient stream that matches that mood.
- It runs locally, with no account, no tracking, and no ads.
Under the hood is a custom synthesizer driven by sentence embeddings, not a generative AI model (although you can choose to use one). This means:
- No GPU required.
- No API calls.
- Playback starts almost instantly.
The whole project is open source.
Programmatic Use
If you’re a developer, you can use LatentScore as a Python library:
pip install latentscore
One line of code renders audio, but you can also use the web player for everyday listening.
Limitations
Fair warning: LatentScore is still in alpha. The synth has limits, so don’t expect full songs or vocals. It’s designed for ambient/procedural music—ideal for focus music or background atmosphere.
Call for Feedback
I’d love to know what vibes you try and whether they land!
— Prabal