Sovereign Voice 0.5.2 — on-device voice synthesis with Qwen3-TTS + MLX on Apple Silicon

Описание и применение

Sovereign Voice 0.5.2 – автономное приложение для macOS, использующее модель Qwen3‑TTS и MLX для генерации речи без подключения к облаку, обеспечивая конфиденциальность и экономию средств.

Задача

Users of cloud voice services face high per-character costs and lack of offline control.

Результат

Sovereign Voice delivers offline, private, and cost‑effective voice synthesis on macOS, with a free tier that supports up to 25 generations and a one‑time $19 upgrade for unlimited use.

Рабочий процесс

• Download and install the signed DMG from the official landing page.
• Launch the app; the first run downloads ~1.9 GB model weights to disk.
• Use the menu bar daemon or studio window to input text and generate speech.
• Optionally clone a voice from a short audio sample.
• Generate up to 25 free voice outputs; beyond that, continue using the free tier without a paywall.
• Upgrade to the Pro tier ($19) for unlimited generations, custom clones, studio mode, and long‑form synthesis.

Доказательства

No audio, text, or personal data sent anywhere after the one-time weight download

Источник

First audio in ~1s; a full short sentence in ~5-6s on an M4

Источник