update

funasr/bin/eend_ola_inference.py

@@ -209,7 +209,7 @@ def inference_modelscope(
         if data_path_and_name_and_type is None and raw_inputs is not None:
             if isinstance(raw_inputs, torch.Tensor):
                 raw_inputs = raw_inputs.numpy()
-            data_path_and_name_and_type = [raw_inputs[0], "speech", "bytes"]
+            data_path_and_name_and_type = [raw_inputs[0], "speech", "sound"]
         loader = EENDOLADiarTask.build_streaming_iterator(
             data_path_and_name_and_type,
             dtype=dtype,

funasr/modules/eend_ola/encoder.py

@@ -87,7 +87,7 @@ class EENDOLATransformerEncoder(nn.Module):
                  n_layers: int,
                  n_units: int,
                  e_units: int = 2048,
-                 h: int = 8,
+                 h: int = 4,
                  dropout_rate: float = 0.1,
                  use_pos_emb: bool = False):
         super(EENDOLATransformerEncoder, self).__init__()

setup.py

@@ -17,7 +17,7 @@ requirements = {
         "humanfriendly",
         "scipy>=1.4.1",
         # "filelock",
-        "librosa>=0.8.0",
+        "librosa==0.8.1",
         "jamo==0.4.1",  # For kss
         "PyYAML>=5.1.2",
         "soundfile>=0.10.2",