加速自回归语音合成，实现实时流式生成与高并发部署。
Faster IndexTTS-2: Accelerating and Streaming Autoregressive Zero-Shot Text-to-Speech Synthesis on GPUs
- 用TensorRT优化各神经网络组件，提升推理速度
- 支持流式生成与批量处理，端到端提速3.6倍
- 适合需要低延迟的语音交互系统部署
自回归文本到语音模型虽具备高自然度，但因逐词生成导致推理缓慢，难以在生产环境中应用。IndexTTS-2 是当前领先的自回归语音合成模型，包含 GPT、基于流匹配的扩散变换器和声码器。尽管音质优秀，其推理速度仍无法达到实时，且缺乏流式与批处理支持。本文提出 Faster IndexTTS-2，利用 NVIDIA TensorRT 与 TensorRT-LLM 对所有神经网络组件进行加速，实现生产级 GPU 部署。该方案支持流式合成，适用于低延迟交互场景，并通过全组件批处理最大化 GPU 利用率。在英文与中文的 Seed-TTS 基准测试中，自回归 GPT 模块加速达 5.0×，端到端速度提升 3.6×，同时保持词错误率、说话人相似度与自然度的轻微下降。该方法为同类自回归语音模型的高效部署提供了实用参考。
原文摘要 · Abstract (English)
Autoregressive text-to-speech models achieve strong naturalness but suffer from slow inference due to sequential token generation, limiting their deployment in production applications that require low latency. IndexTTS-2 is a state-of-the-art autoregressive TTS model consisting of a GPT, a flow-matching Diffusion Transformer, and a vocoder. Despite its high synthesis quality, its inference speed barely reaches real-time without streaming or batching support. We present Faster IndexTTS-2, which accelerates all neural network components of IndexTTS-2 for production deployment on GPUs using NVIDIA TensorRT and TensorRT-LLM. Faster IndexTTS-2 also enables streaming synthesis for latency-sensitive interactive applications, and batched inference across all components to maximize GPU utilization. Experiments on the Seed-TTS benchmark for both English and Chinese demonstrate up to 5.0$\times$ speedup on the autoregressive GPT and 3.6$\times$ end-to-end, with minimal degradation in word error rate, speaker similarity, and naturalness. Our methodology provides a practical reference for efficiently accelerating similar autoregressive speech models on GPUs.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。