microsoft / AEC-Challenge

AEC Challenge
MIT License
378 stars 128 forks source link

Does the local version of AECMOS support inference in batch size bigger than 1? (the local version) #22

Open ahikaml opened 2 years ago

ahikaml commented 2 years ago

The shape of the input to the model is defined in the onnx file as[batch_size, 3, frames, 160], but When I run it with a batch size of 8 samples I get the following error: onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Non-zero status code returned while running GRU node. Name:'GRU_17' Status Message: Input initial_h must have shape {2,8,64}. Actual:{2,1,64}