VoXtream2: Full-stream TTS with dynamic speaking rate control

TL;DR

VoXtream2 is a real-time, controllable, zero-shot full-stream TTS model that enables dynamic speaking rate adjustments mid-utterance with high efficiency and quality.

Contribution

It introduces VoXtream2, a novel TTS model capable of dynamic speaking rate control and prompt-text masking, achieving high-quality synthesis with minimal latency in a zero-shot, full-stream setting.

Findings

Runs 4 times faster than real time on consumer GPU

Achieves competitive results on zero-shot benchmarks

Supports mid-utterance speaking rate adjustments

Abstract

Full-stream text-to-speech (TTS) for interactive systems must start speaking with minimal delay while remaining controllable as text arrives incrementally. We present VoXtream2, a zero-shot full-stream TTS model with dynamic speaking-rate control that can be updated mid-utterance on the fly. VoXtream2 combines a distribution matching mechanism over duration states with classifier-free guidance across conditioning signals to improve controllability and synthesis quality. Prompt-text masking enables textless audio prompting, removing the need for prompt transcription. Across standard zero-shot benchmarks and a dedicated speaking-rate test set, VoXtream2 achieves competitive objective and subjective results against public baselines despite a smaller model and less training data. In full-stream mode, it runs 4 times faster than real time with 74 ms first-packet latency on a consumer GPU.