Skip to content
aivrar edited this page Nov 18, 2026 · 10 revisions

🎧 LocalSoundsAPI Wiki

A portable AI audio studio for Windows that runs entirely on your own machine. Clone voices, narrate scripts, generate music with vocals and sound effects, transcribe, and turn it all into subtitled videos, from a web UI or a local REST API.

Kokoro narrating while Stable Audio renders a thunderstorm, side by side

Latest: v1.2.1. Different engines run at the same time, every panel has a live progress bar with ETA, and Fish Speech keeps its models loaded for faster multi-chunk jobs. See Concurrency & Progress.

Start here

  1. Installation: double-click install.bat; no admin rights or system Python needed.
  2. Getting Started: your first generation in about five minutes.
  3. Launcher: start servers on each GPU and download models.

The engines

Engine What it does Listen
XTTS-v2 Voice Cloning Clone any voice from ~10 s of audio, 17 languages, 58 built-in speakers ▶ sample
Fish Speech Highest-quality voice cloning for final takes ▶ sample
Kokoro TTS Tiny and very fast: 54 voices, 9 languages, works on CPU ▶ sample
Stable Audio Sound effects, ambiences and instrumentals up to 47 s, ranked by CLAP ▶ sample
ACE-Step Music Full songs with sung lyrics from a style tag + [verse]/[chorus] ▶ sample
Upload & Transcribe Add cloning voices; Whisper transcribes and checks every generated chunk
Video Production Audio + images → subtitled video, 720p to 4K, vertical or transparent
Chatbot Local LLM, LM Studio or OpenRouter writes scripts and lyrics, then sends them to any engine

Every sample was generated with this app, using the settings shown on that engine's page. The cloned samples use a synthetic reference voice, and every speech sample passed the app's Whisper verification.

For developers

  • API Reference: every UI action is a JSON call. The API Payload button on each panel shows the exact request.
  • Concurrency & Progress: /job_status, 503 Busy with Retry-After, per-client cancel.

Help

Clone this wiki locally