FlashSpeech: Efficient Zero-Shot Speech Synthesis

arXiv:2404.14700v1 Announce Type: cross
Abstract: Recent progress in large-scale zero-shot speech synthesis has been significantly advanced by language models and diffusion models. However, the generation process of both methods is slow and computationally intensive. Efficient speech synthesis using a lower computing budget to achieve quality on par with previous work remains a significant challenge. In this paper, we present FlashSpeech, a large-scale zero-shot speech synthesis system with approximately 5\% of the inference time compared with previous work. FlashSpeech is built …

arxiv cs.ai cs.cl cs.lg cs.sd eess.as speech synthesis type zero-shot