Summary
Highlights
The video introduces Voicebox, a free and open-source app that allows users to clone any voice on their PC. It promises quick setup, no subscriptions, and the ability to generate new speech from short voice samples in seconds.
Voicebox utilizes advanced AI voice models like Qwen 3 and Chatterbox. Users can download the appropriate version for their operating system from the website and follow the installation steps.
Once installed, navigate to the 'create voice' section. You can record your voice directly (up to 30 seconds for better quality), upload an audio file, or capture system audio. After recording/uploading, transcribe the audio and name your voice profile.
Select your newly created voice profile and type in the text you want it to speak. The video recommends using the Qwen 3 TTS model for best results. Click 'generate' to create the audio and play it back. You can export, regenerate, or delete the audio.
Voicebox offers a 'stories' feature to create multi-speaker narratives. Add a new story, type in lines, assign a speaker (from your created voices) to each line, and generate the speech. The timeline allows for adjusting timing and order. The final story can be exported as a single audio file.
The app includes sections for reviewing created voices, applying audio effects like 'robotic,' 'radio,' or 'echo chamber,' and customizing effects. Users can also configure audio devices (microphone/speakers) and view the AI models used. Advanced settings allow for GPU acceleration for faster processing.
For optimal results, use clean audio samples with minimal background noise, record in quiet environments, and speak clearly. Longer samples (20-30 seconds) tend to yield better quality. If dissatisfied, regenerate or adjust the text. The video concludes by emphasizing the free nature of Voicebox for PC voice cloning, urging users to only clone voices they have permission to use.