Open shenbuguanni opened 8 months ago
It seems that the input of multi-channel audio cannot be supported, for example, shape is [B,samples], and B refers to the number of audio. In this case, the model needs to be processed one by one when training online, which affects the efficiency.
It seems that the input of multi-channel audio cannot be supported, for example, shape is [B,samples], and B refers to the number of audio. In this case, the model needs to be processed one by one when training online, which affects the efficiency.