Would you be interested in converting the ncnn model of these two？

[Baiyuetribe]

1. audio denoise src: https://github.com/haoheliu/voicefixer online demo: https://haoheliu.github.io/demopage-voicefixer
2. audio enhance src: https://huggingface.co/spaces/brentspell/hifi-gan-bwe online demo: https://huggingface.co/spaces/brentspell/hifi-gan-bwe

ncnn has many cases in the image field, but you are the first taster in the audio field, hope to add more cases

[lhwcv]

Thank you for your good advice.

I 'll try this and deploy use NCNN: https://github.com/brentspell/hifi-gan-bwe

Maybe next month I'll have time to do this.

[zuowanbushiwo]

@lhwcv 能不能把这个工程也转一下 DPCRN_DNS3，这个和DTLN 一样 提供了实时脚本，预训练模型。 但看不懂tensorflow 写的Conv2D_stateful 和 DeConv2D_stateful，搜索了一下也没找到pytorch 实现。

Models	Para. (M)	PESQ	STOI	SI-SDR
Noisy	-	1.97	92.1	8.41
RNNoise19 [2018]	0.06	2.34	-	-
PerceptNet3 [2020]	8.0	2.73	-	-
DeepFilterNet4 [2022]	1.80	2.81	-	16.63
S-DCCRN5 [2022]	2.34	2.84	94.0	-
DPARN [2022]	0.89	2.92	94.2	18.28