Skip to content

Home

Jump to bottom
aivrar edited this page Aug 30, 2026 · 6 revisions

LocalSoundsAPI Wiki

LocalSoundsAPI is a fully portable, local AI audio studio for Windows. It includes text-to-speech, voice cloning, transcription, music and sound-effect generation, video production, and a local API.

Current Release

LocalSoundsAPI v1.1 - One-Click Installer

The v1.1 source release adds automatic first-time setup. The portable Python and audio-tool dependency bundle remains on v1.0; the installer finds that compatible bundle automatically.

One-Click Installation

Requirements

  • 64-bit Windows 10 or Windows 11
  • About 2.1 GB for the initial dependency download
  • Additional storage for AI models as you install them
  • No administrator rights, system Python, or preinstalled 7-Zip required

Setup

  1. Clone the repository or download and extract the v1.1 source archive.
  2. Double-click install.bat.
  3. Wait while setup downloads, verifies, and extracts the portable environment.
  4. The LocalSoundsAPI Launcher opens automatically.
  5. Download the AI models you want from the Launcher's Models & Tools tab, or let each engine download its model on first use.

You can also double-click launcher.bat on a fresh clone. It detects missing dependencies and starts the one-click installer automatically.

What the Installer Does

  • Finds the newest GitHub release containing both required dependency assets
  • Downloads portable-python-env-v1.7z and bin.zip
  • Verifies the SHA-256 digest GitHub publishes for each asset
  • Resumes interrupted downloads when possible
  • Extracts the portable Python environment into python/
  • Installs FFmpeg, RubberBand, and eSpeak-ng under bin/
  • Leaves the operating system and system Python untouched

The optional lib.zip and triton.zip files on v1.0 are troubleshooting packages, not standard installation requirements.

Manual Installation Fallback

The original manual setup remains supported:

  1. Open the v1.0 dependency bundle.

  2. Download portable-python-env-v1.7z and extract it into the LocalSoundsAPI project root. It creates python/.

  3. Download bin.zip and extract it into the same project root. It creates or populates bin/.

  4. Confirm these required files exist:

    python/python.exe
bin/ffmpeg/bin/ffmpeg.exe
bin/rubberband/rubberband.exe
bin/espeak-ng/libespeak-ng.dll

  5. Double-click launcher.bat.

Do not substitute a raw pip install -r requirements.txt environment for the provided portable Python bundle; the complete ML and CUDA stack is preassembled in the release asset.

Starting LocalSoundsAPI

  • Launcher GUI: launcher.bat
  • Simple single instance: (portable) LocalSoundsAPI-Single.bat
  • Multiple command-line instances: (portable) LocalSoundsAPI-Multi.bat

The default single-instance address is http://127.0.0.1:5006.

Updating an Existing Checkout

git pull origin main
install.bat

If the portable runtime and tools are already present, install.bat verifies the installation and opens the Launcher without downloading them again. Run install.bat -Force from a terminal only when you intentionally want to reinstall the dependency bundle.

Documentation and Help

After the dependency and model downloads are complete, LocalSoundsAPI can run fully offline.

Clone this wiki locally