// docs / overview

LiveAudio documentation

LiveAudio is a free, open-source (MIT) desktop app that generates real-time Whisper speech captions 100% locally and streams them to OBS over a local WebSocket. v1.5.1 runs on Windows x64 as a native Rust + Tauri 2 build — no cloud, no API key, no per-minute cost.

What LiveAudio is

Plynte LiveAudio is a local-first speech-recognition engine for streamers and creators. It transcribes with Whisper (tiny, base, small, turbo), trims silence with Silero VAD, and broadcasts subtitle JSON over a local WebSocket so OBS — or any localhost client — can render live captions. v1.5.1 is a native Rust + Tauri 2 desktop app: faster startup, lower RAM, and no Python runtime.

Local processing. No audio or transcribed text leaves your machine. No telemetry.
Memory-safe core. The transcription engine is written in Rust, so entire classes of memory bugs are ruled out at compile time.
Hot-swap. Change the audio device or model size without restarting the app.

System requirements

Component	Requirement
Operating system	Windows 10/11 x64. No macOS. Linux: legacy Python v1.2.7 tarball.
GPU	Optional — CPU works. NVIDIA CUDA is recommended and auto-detected; needs driver `≥ 525` and VRAM `≥ 4 GiB`.
RAM	8 GB minimum, 16 GB recommended.
Disk	Installer ~26.2 MB (portable exe ~16.2 MB), plus transcription-model storage.
Runtime	None — v1.5.1 is a native binary. No Python runtime.
Internet	First run only (downloads transcription models). Fully offline afterward.

Capture and platforms

LiveAudio captures from a physical microphone. On Windows it can also capture system audio via WASAPI loopback.

How it installs (native build)

What you download is the whole app — a native Rust + Tauri 2 binary, not a bootstrapper. The Windows installer (LiveAudio_1.5.1_x64-setup.exe, ~26.2 MB) sets up the desktop app directly: no Python runtime, no dependency provisioning. On first transcription it downloads only the transcription models, then runs fully offline.

native install flow

LiveAudio_1.5.1_x64-setup.exe  /  liveaudio-desktop.exe
  │
  ├─ 1. Detect legacy Python builds (v1.0.0–v1.2.7) via the Windows registry
  ├─ 2. Run the legacy uninstaller silently + stop leftover processes
  ├─ 3. Remove obsolete Python folders (settings in %APPDATA%\LiveAudio kept)
  ├─ 4. Install the native Rust + Tauri 2 desktop app (~26.2 MB, no Python)
  └─ 5. Launch — transcription models download once on first use

Your settings and sessions live in %APPDATA%\LiveAudio and survive updates. Coming from a Python build (v1.0.0–v1.2.7)? The installer removes it silently first — your data is preserved.

Where your data lives

v1.5.1 keeps your data in one place — no virtual environments, no managed Python:

Location	Content
`%APPDATA%\LiveAudio`	Your settings (`config.json`) and `sessions/` — preserved across installs and the Python-to-Rust migration.
Legacy Python installs (v1.0.0–v1.2.7)	Detected via the Windows registry and removed silently by the v1.5.1 installer.

Hardware detection

The app chooses between cuda and cpu with a first-match-wins ladder:

Your explicit choice in the Advanced tab (compute-device selector) — picking CUDA dims the CPU-threads control because inference runs on the GPU.
Otherwise NVML + nvidia-smi: if the driver is ≥ 525 and VRAM is ≥ 4096 MiB → cuda.
Anything else (no NVML, smi failure, or a GPU below requirements) → cpu, with a note when a GPU was found but did not qualify.

Updates

Update in one click from inside the app. New releases — installer, portable exe, changelog and hashes — are published on GitHub Releases. Your settings and sessions in %APPDATA%\LiveAudio are preserved.