Voice Lock is enabled automatically in supported buffers. It maps faces and other structural text properties to speech personalities, allowing comments, strings, headings, links, and other elements to sound distinct without adding words to the document.
Try it in source code, Org, Info, or EWW. Move between two differently styled elements, such as a comment and a string or a heading and a link. Press C-e d v to turn Voice Lock off in that buffer, compare the same text, then press the key again to restore it.
Use C-e d C-v when you want to change Voice Lock globally. Turning it off removes this layer of voice changes; it does not silence ordinary speech, semantic feedback, tones, or auditory icons.
The exact personalities depend on the speech engine and active presentation settings. For an explanation of the separate Voice Lock, face-presentation, and semantic-presentation layers, see Voice Lock Compatibility. Use Aural Presentation: Speech, Voices, And Sound when you want to choose voices, sounds, profiles, or other reusable presentation behavior.