Add audio model (pth + model.py + labels + meta)

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text

baseline_mean_female.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b83b047f116cefab3b0780856dd304dd69bfc461037f36e588c2c665394d60c
+size 180

baseline_mean_male.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0d71d187a2dda7888286a3c9b68cefb7bc7c41dcc998ded431e9833b3d18ff3
+size 180

baseline_std_female.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce87caf7c7a888500c0c8c612b8f49f1b401ec015512471bd8065805a428e18e
+size 180

baseline_std_male.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:30ee24a5ecd49255f381bdcc5de0bb5c1783871d2309596188336f652e8a041c
+size 180

labels.json

@@ -0,0 +1,8 @@
+[
+  "HAPPY",
+  "SAD",
+  "ANGRY",
+  "ANXIOUS",
+  "SURPRISE",
+  "FEAR"
+]

model.py

@@ -0,0 +1,26 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class PyTorchAudioModel(nn.Module):
+    def __init__(self, num_labels=6):
+        super().__init__()
+        self.conv1 = nn.Conv1d(13, 64, kernel_size=5, padding="same")
+        self.bn1 = nn.BatchNorm1d(64)
+        self.pool1 = nn.MaxPool1d(2, 2)
+        self.conv2 = nn.Conv1d(64, 128, kernel_size=3, padding="same")
+        self.bn2 = nn.BatchNorm1d(128)
+        self.pool2 = nn.MaxPool1d(2, 2)
+        self.bilstm = nn.LSTM(128, 64, bidirectional=True, batch_first=True)
+        self.dense1 = nn.Linear(128, 128)
+        self.dense2 = nn.Linear(128, num_labels)
+
+    def forward(self, x):
+        x = self.pool1(F.relu(self.bn1(self.conv1(x))))
+        x = self.pool2(F.relu(self.bn2(self.conv2(x))))
+        x = x.permute(0, 2, 1)  # (B,C,L)->(B,L,C)
+        _, (h_n, _) = self.bilstm(x)
+        x = torch.cat([h_n[-2], h_n[-1]], dim=1)
+        x = F.relu(self.dense1(x))
+        return self.dense2(x)

model_meta.json

@@ -0,0 +1,19 @@
+{
+  "model_version": "v1.0",
+  "framework": "keras-3.x / tf-2.15+",
+  "input_shape": [
+    100,
+    13
+  ],
+  "output_classes": 6,
+  "labels_order": [
+    "HAPPY",
+    "SAD",
+    "ANGRY",
+    "ANXIOUS",
+    "SURPRISE",
+    "FEAR"
+  ],
+  "delta_required": true,
+  "delta_formula": "(X - mean) / (std + 1e-8)"
+}

pytorch_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a560dc67d3063ae125e7be22753dcd8803f6b1b148b9e44c880a59f9a9bc942
+size 593404