Voicebox: Windows setup and Italian voice cloning

This tutorial follows a Windows installation of Voicebox and an Italian voice cloning test using the presenter's own recording. The supplied metadata describes the app as open source and available for Windows, macOS and Linux. The presenter installs it through an MSI file and opens the interface without creating an account.

The workflow covers selecting Italian and Qwen 3, uploading a recording, creating a voice profile and generating speech from pasted text. Microphone recording and system audio capture are also shown as input options. The presenter understands the reference audio limit to be 30 seconds. A profile text field has a 1,000-character limit; the example uses 977 characters. In the closing remarks, he separately states a 10,000-character generation limit and suggests splitting longer text into sections.

The model download is 3.6 GB. The presenter reports a 24-minute wait before the first result and says the generated speech sounds similar to his voice. That is his assessment of this example, rather than a measured comparison with ElevenLabs, which he introduces as a paid option.

The final steps cover WAV export, regeneration and voice effects such as robotic and radio treatments. The model panel also includes Whisper transcription models. The tutorial gives a practical local AI setup example, but does not specify hardware requirements or benchmark subsequent generation speed.