Text-to-Speech (TTS) models are significantly more numerically fragile than Large Language Models (LLMs) due to their continuous waveform generation and perceptual sensitivity to small numerical perturbations. While aggressive precision reduction techniques such as BlockFloat8 (BFP8) and low-fidelity (LoFi) compute have been widely adopted in language models, applying similar strategies to TTS systems often results in audible artifacts, phase instability, and spectral distortion. In this work, we present Lightning V2, a production-grade TTS model co-optimized for Tenstorrent hardware. Through precision-aware architectural design and hardware-software co-optimization, we achieve over 95% LoFi computational fidelity and more than 80% BlockFloat8 deployment without measurable degradation in audio quality. Leveraging Tenstorrent's Network-on-Chip (NoC), distributed SRAM, and deterministic execution model, we reduce memory movement and redundant weight fetches, enabling efficient low-precision inference. Compared to an NVIDIA L40S baseline, Lightning V2 achieves approximately 4x lower on-prem accelerator cost at equivalent throughput, while maintaining production audio fidelity. Our results demonstrate that precision co-design, combined with hardware-aware optimization, can fundamentally reshape the economics of real-time speech inference.

翻译：文本转语音（TTS）模型在数值上比大语言模型（LLMs）更为脆弱，原因在于其连续的波形生成以及对微小数值扰动在感知上的敏感性。尽管诸如BlockFloat8（BFP8）和低保真度（LoFi）计算等激进的精度降低技术已在语言模型中得到广泛采用，但将类似策略应用于TTS系统时，常会导致可听伪影、相位不稳定和频谱失真。在本工作中，我们提出了Lightning V2，一个针对Tenstorrent硬件进行协同优化的生产级TTS模型。通过精度感知的架构设计以及硬件-软件协同优化，我们实现了超过95%的LoFi计算保真度和超过80%的BlockFloat8部署，同时音频质量无显著可测退化。利用Tenstorrent的片上网络（NoC）、分布式SRAM以及确定性执行模型，我们减少了内存移动和冗余权重获取，从而实现了高效的低精度推理。与NVIDIA L40S基线相比，Lightning V2在等吞吐量下实现了大约4倍的低部署加速器成本，同时保持了生产级音频保真度。我们的结果表明，精度协同设计与硬件感知优化相结合，可以从根本上重塑实时语音推理的经济性。