LatentScore Live

Spec sheet

What it does

LatentScore is an open-source Python library that turns text prompts into procedural music in real time.Read 12 Oct 2026 · prabal.ca

Product

LatentScore is an open-source Python library that generates ambient music from text descriptions.Read 5 Oct 2026 · latentscore.com

Intended users

The maker presents it as a Python library for generating ambient music from text descriptions and demonstrates it for research use.Read 12 Oct 2026 · github.com

Integrations

The optional external install connects to hosted LLMs through LiteLLM, including Anthropic, Gemini, and OpenAI.Read 12 Oct 2026 · github.com

License

The project is released under the Apache License 2.0.Read 12 Oct 2026 · github.com

Access

The maker links to a browser demo, GitHub source and documentation, Google Colab, and a PyPI installation.Read 5 Oct 2026 · latentscore.com

Access options

The maker links to a browser demo, a Colab notebook, Docker-based local demo, and a pip-installable Python library.Read 12 Oct 2026 · github.com

Audio export formats

wavRead 21 Sep 2026 · latentscore.com

CPU synthesis

The synthesizer is CPU-based, and the maker says the default retrieval backend runs in under a second without network access or inference.Read 12 Oct 2026 · prabal.ca

Live control

The SDK can continuously emit audio and crossfade to new configurations while playback continues.Read 12 Oct 2026 · prabal.ca

Live controls

The demo offers real-time adjustment of 18 or more musical parameters, including tempo, mode, and texture.Read 5 Oct 2026 · latentscore.com

Manual parameters

Users can manually change synthesis parameters and re-render with updates.Read 12 Oct 2026 · prabal.ca

Music focus

The maker says it handles ambient and electronic music, but cannot generate vocals.Read 12 Oct 2026 · prabal.ca

Optional models

The heavy extra adds CLAP audio-based retrieval, while the expressive extra for local LLM inference is described as extremely experimental and under testing.Read 5 Oct 2026 · github.com

Output

The README shows saving generated audio to a WAV file.Read 12 Oct 2026 · github.com

Output controls

The Python library supports configuring musical parameters directly and saving rendered audio as WAV.Read 5 Oct 2026 · github.com

Playlists

The maker says users can build playlists of different vibes and cross-fade between tracks for live performance sets.Read 5 Oct 2026 · latentscore.com

Supported SDK platforms

Local pip installation is supported on macOS and Linux with Python 3.11 or 3.12; native Windows pip is unsupported.Read 5 Oct 2026 · github.com

Supported setup

The local SDK install supports macOS and Linux with Python 3.11 or 3.12; the demo can run through Docker Desktop on Windows using WSL2.Read 12 Oct 2026 · github.com

Text to music

Users can describe a vibe in natural language to generate ambient music without musical training.Read 5 Oct 2026 · latentscore.com

Visual interface

The web interface displays generated titles, color palettes, and a parameter editor, with particle effects that respond to playback.Read 12 Oct 2026 · prabal.ca

Visuals

The maker says particle effects are driven by the generated color palette.Read 5 Oct 2026 · latentscore.com

Windows demo

The maker says the local Docker demo has been tested on Windows with Docker Desktop using the WSL2 backend.Read 5 Oct 2026 · github.com

Where Specifiction places it

• AI music generators#106 of 1237.3