16k model

script.py

@@ -58,7 +58,12 @@ model = LCNN(return_emb=False).to(device)
 # model_path = './checkpoints/LCNN_ALL_DATA_AUG.pth'
 # model_path = './checkpoints/LCNN_ALL_DATA_TTS_AUG.pth'
 # model_path = './checkpoints/LCNN_ALL_DATA_TTS_MOD.pth'
-model_path = './checkpoints/LCNN_ALL_DATA_HI_FREQ_22050.pth'
+# model_path = './checkpoints/LCNN_ALL_DATA_HI_FREQ_22050.pth'
+
+model_path = './checkpoints/LCNN_ALL_DATA_FS_16000.pth'
+# model_path = './checkpoints/LCNN_ALL_DATA_FS_22050.pth'
+# model_path = './checkpoints/LCNN_ALL_DATA_FS_24000.pth'
+
 model.load_state_dict(torch.load(model_path, map_location=device))
 
 # # MOE MODEL
@@ -128,7 +133,7 @@ for el in tqdm.tqdm(dataset_remote):
 
     # RUNNING ON HUGGINGFACE
     file_like = io.BytesIO(el["audio"]["bytes"])
-    tensor, sr = preprocess(file_like, target_sr=22050)
+    tensor, sr = preprocess(file_like, target_sr=16000)
     # # RUNNING LOCALLY
     # tensor = preprocess(el)
 

src/lcnn_model.py

@@ -24,7 +24,7 @@ class LCNN(nn.Module):
     def __init__(self, return_emb=False, num_class=2):
         super(LCNN, self).__init__()
 
-        self.threshold = 0.5
+        self.threshold = 0.35
         self.return_emb = return_emb
 
         # Feature Extraction Part (First Part)