https://arxiv.org/abs/2011.12985
FBWave: Efficient and Scalable Neural Vocoders for Streaming Text-To-Speech on the Edge (Bichen Wu, Qing He, Peizhao Zhang, Thilo Koehler, Kurt Keutzer, Peter Vajda)
nonautoregressive flow와 autoregressive flow를 붙인 형태의 vocoder. quantization하고 갤럭시 S8에 올렸더니 real-time factor 0.7 정도가 나왔다고.
#non-autoregressive #vocoder #lightweight