Grok Voice 3.0
Turn any script into a natural voiceover, or any recording into an editable transcript.
Week 17 · Nov 2 – 8, 2026Product
About
Grok Voice 3.0 puts text to speech and speech to text in one browser workspace, so creators, marketers and educators do not have to juggle a voice generator, a separate transcription service and a folder of exports. Text to speech: paste a script, listen to four voices (Eve, Ara, Atlas and Leo) on the same passage, then generate narration. Scripts can run up to 60,000 characters per request, with word level timestamps for lining narration up with captions or B-roll. Downloads include MP3 and WAV, and audio settings also offer PCM, mu-law and A-law with sample rates from 8 kHz to 48 kHz. Write the script in English, Spanish, French, German, Japanese, Chinese and more, or let auto detect pick the language. Speech to text: upload an audio file up to 500 MiB or paste a public audio link, then review, copy and download the transcript as a .txt file. Speaker labels help with interviews, panels and two host podcasts, up to 100 custom keywords keep names, brands and technical terms spelled correctly, and a clean or verbatim mode decides whether filler words stay in. A "Use as script" action moves any transcript into the voice editor, so a recorded interview can become a fresh voiceover. Every result is saved to My Creations, and estimated credits are shown before a task runs when an estimate is available.
