Context—
VibeVoice Realtime 0.5B is a lightweight real-time text-to-speech model from Microsoft with streaming text input and long-form speech generation, producing the first audible speech in about 300 ms. It encodes incoming text chunks while continuing diffusion-based acoustic generation, using only an acoustic tokenizer at 7.5 Hz. It is built primarily for English, with nine more languages available to try: German, French, Italian, Japanese, Korean, Dutch, Polish, Portuguese and Spanish.