modelscope / 3D-Speaker

A Repository for Single- and Multi-modal Speaker Verification, Speaker Recognition and Speaker Diarization
Apache License 2.0
1.07k stars 93 forks source link

damo/speech_eres2net_sv_zh-cn_16k-common 模型不支持输入numpy数组 #46

Closed kwaihua closed 8 months ago

kwaihua commented 8 months ago

隔壁cam++推理时,可以直接输入提取好的音频numpy数组进行推理。 希望eres2net也可以统一加上,谢谢

yfchenlucky commented 8 months ago

您好,我check了modelscope上damo/speech_eres2net_sv_zh-cn_16k-common模型推理代码,是支持numpy数组输入的,所以不知道具体问题出现在哪里?