Player not loading? Watch on YouTube
This tutorial uses Claude Code to install a community copy of VibeVoice, download a model, and start a local browser interface in Chrome. The presenter chooses the 1.5B model and identifies 7B as a larger option that takes more storage. Her download takes about 30 minutes; that is an example from her setup, not a general installation time. The video does not specify minimum hardware requirements.
The interface supports up to four speakers with separate voice choices. The presenter tests a preset voice, an English and Chinese sample, and a conversation between two speakers. She compares the results with ElevenLabs and describes the podcast mode as similar to NotebookLM, with customizable voices. These are her assessments rather than measured quality comparisons.
For voice cloning, she recommends at least 30 seconds of speech recorded in a quiet room. She records on an iPhone, transfers the sample to her computer, and asks Claude Code to convert it to WAV and place it in the appropriate folder. After restarting the server, she selects the custom voice and generates a short introduction.