Magpie TTS Studio

A modern PyQt6 desktop GUI for the Hugging Face model nvidia/magpie_tts_multilingual_357m.

Features

• modern dark/light desktop UI

• separate settings window for app language, TTS language, speaker, device and storage paths

• local cache folder for Hugging Face downloads

• first-run or manual model pre-download

• WAV export with timestamp-based filenames

• last-output playback inside the app

• virtual-environment friendly Windows setup scripts

  

Direct english output (example):

magpie_20260417_001837_en_Sofia.mp4

German Voice Output (example random generated sci-fi story):

magpie_20260423_210746_de_Aria.mp4

Project structure

• app.py — GUI entry point
• src/main_window.py — main application window
• src/options_dialog.py — settings dialog
• src/tts_backend.py — Magpie/NeMo loading and synthesis backend
• tools/preload_models.py — optional model prefetch tool
• install_windows.bat — creates venv, installs dependencies, optionally pre-downloads model
• run_windows.bat — launches the app inside the venv

Quick start

Windows

1. Extract the ZIP.
2. Run install_windows.bat.
3. The installer now tries Python 3.12, 3.11, 3.10, then any available py/python interpreter. If your version is older than 3.10, it warns but still continues.
4. After setup, run run_windows.bat.

Manual install

python -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip setuptools wheel
python -m pip install -r requirements.txt
python -m pip install torch torchvision torchaudio
python -m pip install "nemo_toolkit[asr,tts] @ git+https://github.com/NVIDIA/NeMo.git"
python tools/preload_models.py
python app.py

Notes

• This model is large, so keep enough free disk space for the .nemo checkpoint plus cache files.
• If the GitHub NeMo install fails, the Windows installer tries a published-package fallback.
• Apply text normalization is intentionally optional because text-normalization availability can vary by environment.
• The app stores settings in app_data/settings.json.

Supported model languages

• English (en)
• German (de)
• Spanish (es)
• French (fr)
• Italian (it)
• Vietnamese (vi)
• Chinese (zh)
• Hindi (hi)
• Japanese (ja)

Speakers

• Sofia
• Aria
• Jason
• Leo
• John

Windows note

This project uses a Windows-specific NeMo install path. nemo_text_processing / pynini is intentionally not installed by default because pip-based Windows installs are not officially supported for that dependency chain. The app can still synthesize speech normally, and text normalization remains optional.

GUI Source

github.com/zeittresor/Magpie_TTS_Studio