Description

The G-Man AI voice model was trained on 10 minutes and 58 seconds of audio data, full of many samples including Half-Life 1 (enhanced and original), Half-Life 2 Beta, Half-Life 2 Retail, and Half-Life: Alyx. The model generates high-definition audio, but to replicate the Half-Life 1 G-Man, simply lower the sample rate to 11,025 Hz in Audacity or a similar program.