The higher the quality of your voice samples, the better the resulting voice model will be. Your recordings should feature a single speaker, be clear, and be free from background noise, music, and effects such as echo or reverb. Avoid long silences, multiple speakers, and ambient noise like air conditioners or street sounds.