Upload 10 files

silero-vad-unified-256ms-v6.0.0.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:718705a52f6031e9a10814115fdfa694d60ddf349a3867915620e67e37f5a9ba
+size 243

silero-vad-unified-256ms-v6.0.0.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28efb7187c42b378ea90183a363742c5c0123ae17dd08de8db2e369f40280602
+size 625

silero-vad-unified-256ms-v6.0.0.mlmodelc/metadata.json

@@ -0,0 +1,121 @@
+[
+  {
+    "shortDescription" : "Silero VAD Unified Model 256ms (STFT + Encoder + Decoder) with noisy-OR aggregation",
+    "metadataOutputVersion" : "3.0",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1]",
+        "name" : "vad_output",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 128)",
+        "shortDescription" : "",
+        "shape" : "[1, 128]",
+        "name" : "new_hidden_state",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 128)",
+        "shortDescription" : "",
+        "shape" : "[1, 128]",
+        "name" : "new_cell_state",
+        "type" : "MultiArray"
+      }
+    ],
+    "version" : "6.0.0",
+    "modelParameters" : [
+
+    ],
+    "author" : "Fluid Infernece + Silero Team",
+    "specificationVersion" : 6,
+    "storagePrecision" : "Mixed (Float16, Float32)",
+    "mlProgramOperationTypeHistogram" : {
+      "Log" : 8,
+      "Concat" : 9,
+      "Lstm" : 8,
+      "SliceByIndex" : 41,
+      "Clip" : 32,
+      "Pow" : 16,
+      "Transpose" : 16,
+      "Sub" : 2,
+      "Relu" : 32,
+      "Squeeze" : 18,
+      "Cast" : 54,
+      "Sigmoid" : 8,
+      "Add" : 24,
+      "Sqrt" : 8,
+      "ExpandDims" : 26,
+      "ReduceMean" : 8,
+      "Conv" : 48,
+      "Mul" : 7
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int32)",
+    "stateSchema" : [
+
+    ],
+    "isUpdatable" : "0",
+    "availability" : {
+      "macOS" : "12.0",
+      "tvOS" : "15.0",
+      "visionOS" : "1.0",
+      "watchOS" : "8.0",
+      "iOS" : "15.0",
+      "macCatalyst" : "15.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 4160)",
+        "shortDescription" : "",
+        "shape" : "[1, 4160]",
+        "name" : "audio_input",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 128)",
+        "shortDescription" : "",
+        "shape" : "[1, 128]",
+        "name" : "hidden_state",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 128)",
+        "shortDescription" : "",
+        "shape" : "[1, 128]",
+        "name" : "cell_state",
+        "type" : "MultiArray"
+      }
+    ],
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2025-09-15",
+      "com.github.apple.coremltools.source" : "torch==2.7.0",
+      "com.github.apple.coremltools.version" : "9.0b1",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "generatedClassName" : "silero_vad_unified_256ms_v6_0_0",
+    "method" : "predict"
+  }
+]

silero-vad-unified-256ms-v6.0.0.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:853cf34740d3f5061f977ebe2976f7c921b064261c9c4753b3a1196f2dba42b4
+size 882304

silero-vad-unified-v6.0.0.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d622bd31042ab8fdd009a40a45c2cd8c9611927841bdd7bfc8ad40d16b6e3f7e
+size 243

silero-vad-unified-v6.0.0.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43933b65aab49a674b94a61e6af794f1fae327961a02160a0823dbfb174e91ce
+size 593

silero-vad-unified-v6.0.0.mlmodelc/metadata.json

@@ -0,0 +1,118 @@
+[
+  {
+    "shortDescription" : "Silero VAD Unified Model (STFT + Encoder + Decoder)",
+    "metadataOutputVersion" : "3.0",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1]",
+        "name" : "vad_output",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 128)",
+        "shortDescription" : "",
+        "shape" : "[1, 128]",
+        "name" : "new_hidden_state",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 128)",
+        "shortDescription" : "",
+        "shape" : "[1, 128]",
+        "name" : "new_cell_state",
+        "type" : "MultiArray"
+      }
+    ],
+    "version" : "6.0.0",
+    "modelParameters" : [
+
+    ],
+    "author" : "Fluid Infernece + Silero Team",
+    "specificationVersion" : 6,
+    "storagePrecision" : "Mixed (Float16, Float32)",
+    "mlProgramOperationTypeHistogram" : {
+      "Log" : 1,
+      "Lstm" : 1,
+      "SliceByIndex" : 2,
+      "Clip" : 4,
+      "Transpose" : 2,
+      "Pow" : 2,
+      "Relu" : 4,
+      "Squeeze" : 4,
+      "Cast" : 12,
+      "Sigmoid" : 1,
+      "Add" : 3,
+      "Sqrt" : 1,
+      "ExpandDims" : 5,
+      "ReduceMean" : 1,
+      "Conv" : 6
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int32)",
+    "stateSchema" : [
+
+    ],
+    "isUpdatable" : "0",
+    "availability" : {
+      "macOS" : "12.0",
+      "tvOS" : "15.0",
+      "visionOS" : "1.0",
+      "watchOS" : "8.0",
+      "iOS" : "15.0",
+      "macCatalyst" : "15.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 576)",
+        "shortDescription" : "",
+        "shape" : "[1, 576]",
+        "name" : "audio_input",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 128)",
+        "shortDescription" : "",
+        "shape" : "[1, 128]",
+        "name" : "hidden_state",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 128)",
+        "shortDescription" : "",
+        "shape" : "[1, 128]",
+        "name" : "cell_state",
+        "type" : "MultiArray"
+      }
+    ],
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2025-09-15",
+      "com.github.apple.coremltools.source" : "torch==2.7.0",
+      "com.github.apple.coremltools.version" : "9.0b1",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "generatedClassName" : "silero_vad_unified_v6_0_0",
+    "method" : "predict"
+  }
+]

silero-vad-unified-v6.0.0.mlmodelc/model.mil

@@ -0,0 +1,145 @@
+program(1.0)
+[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3500.14.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.7.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0b1"}})]
+{
+    func main<ios15>(tensor<fp32, [1, 576]> audio_input, tensor<fp32, [1, 128]> cell_state, tensor<fp32, [1, 128]> hidden_state) {
+            tensor<int32, [1]> x_1_axes_0 = const()[name = tensor<string, []>("x_1_axes_0"), val = tensor<int32, [1]>([1])];
+            tensor<string, []> audio_input_to_fp16_dtype_0 = const()[name = tensor<string, []>("audio_input_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
+            tensor<fp16, [1, 576]> audio_input_to_fp16 = cast(dtype = audio_input_to_fp16_dtype_0, x = audio_input)[name = tensor<string, []>("cast_11")];
+            tensor<fp16, [1, 1, 576]> x_1_cast_fp16 = expand_dims(axes = x_1_axes_0, x = audio_input_to_fp16)[name = tensor<string, []>("x_1_cast_fp16")];
+            tensor<string, []> stft_out_pad_type_0 = const()[name = tensor<string, []>("stft_out_pad_type_0"), val = tensor<string, []>("custom")];
+            tensor<int32, [2]> stft_out_pad_0 = const()[name = tensor<string, []>("stft_out_pad_0"), val = tensor<int32, [2]>([128, 128])];
+            tensor<int32, [1]> stft_out_strides_0 = const()[name = tensor<string, []>("stft_out_strides_0"), val = tensor<int32, [1]>([256])];
+            tensor<int32, [1]> stft_out_dilations_0 = const()[name = tensor<string, []>("stft_out_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> stft_out_groups_0 = const()[name = tensor<string, []>("stft_out_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [258, 1, 256]> stft_forward_basis_to_fp16 = const()[name = tensor<string, []>("stft_forward_basis_to_fp16"), val = tensor<fp16, [258, 1, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
+            tensor<fp16, [1, 258, 3]> stft_out_cast_fp16 = conv(dilations = stft_out_dilations_0, groups = stft_out_groups_0, pad = stft_out_pad_0, pad_type = stft_out_pad_type_0, strides = stft_out_strides_0, weight = stft_forward_basis_to_fp16, x = x_1_cast_fp16)[name = tensor<string, []>("stft_out_cast_fp16")];
+            tensor<int32, [3]> var_25_begin_0 = const()[name = tensor<string, []>("op_25_begin_0"), val = tensor<int32, [3]>([0, 0, 0])];
+            tensor<int32, [3]> var_25_end_0 = const()[name = tensor<string, []>("op_25_end_0"), val = tensor<int32, [3]>([1, 129, 3])];
+            tensor<bool, [3]> var_25_end_mask_0 = const()[name = tensor<string, []>("op_25_end_mask_0"), val = tensor<bool, [3]>([true, false, true])];
+            tensor<fp16, [1, 129, 3]> var_25_cast_fp16 = slice_by_index(begin = var_25_begin_0, end = var_25_end_0, end_mask = var_25_end_mask_0, x = stft_out_cast_fp16)[name = tensor<string, []>("op_25_cast_fp16")];
+            tensor<int32, [3]> var_28_begin_0 = const()[name = tensor<string, []>("op_28_begin_0"), val = tensor<int32, [3]>([0, 129, 0])];
+            tensor<int32, [3]> var_28_end_0 = const()[name = tensor<string, []>("op_28_end_0"), val = tensor<int32, [3]>([1, 258, 3])];
+            tensor<bool, [3]> var_28_end_mask_0 = const()[name = tensor<string, []>("op_28_end_mask_0"), val = tensor<bool, [3]>([true, true, true])];
+            tensor<fp16, [1, 129, 3]> var_28_cast_fp16 = slice_by_index(begin = var_28_begin_0, end = var_28_end_0, end_mask = var_28_end_mask_0, x = stft_out_cast_fp16)[name = tensor<string, []>("op_28_cast_fp16")];
+            tensor<fp16, []> var_7_promoted_to_fp16 = const()[name = tensor<string, []>("op_7_promoted_to_fp16"), val = tensor<fp16, []>(0x1p+1)];
+            tensor<fp16, [1, 129, 3]> var_30_cast_fp16 = pow(x = var_25_cast_fp16, y = var_7_promoted_to_fp16)[name = tensor<string, []>("op_30_cast_fp16")];
+            tensor<fp16, []> var_7_promoted_1_to_fp16 = const()[name = tensor<string, []>("op_7_promoted_1_to_fp16"), val = tensor<fp16, []>(0x1p+1)];
+            tensor<fp16, [1, 129, 3]> var_31_cast_fp16 = pow(x = var_28_cast_fp16, y = var_7_promoted_1_to_fp16)[name = tensor<string, []>("op_31_cast_fp16")];
+            tensor<fp16, [1, 129, 3]> var_32_cast_fp16 = add(x = var_30_cast_fp16, y = var_31_cast_fp16)[name = tensor<string, []>("op_32_cast_fp16")];
+            tensor<fp16, []> var_33_to_fp16 = const()[name = tensor<string, []>("op_33_to_fp16"), val = tensor<fp16, []>(0x1p-24)];
+            tensor<fp16, [1, 129, 3]> var_34_cast_fp16 = add(x = var_32_cast_fp16, y = var_33_to_fp16)[name = tensor<string, []>("op_34_cast_fp16")];
+            tensor<fp16, [1, 129, 3]> magnitude_cast_fp16 = sqrt(x = var_34_cast_fp16)[name = tensor<string, []>("magnitude_cast_fp16")];
+            tensor<fp16, []> var_36_to_fp16 = const()[name = tensor<string, []>("op_36_to_fp16"), val = tensor<fp16, []>(0x1p-24)];
+            tensor<fp16, [1, 129, 3]> var_37_cast_fp16 = add(x = magnitude_cast_fp16, y = var_36_to_fp16)[name = tensor<string, []>("op_37_cast_fp16")];
+            tensor<fp16, []> input_1_epsilon_0_to_fp16 = const()[name = tensor<string, []>("input_1_epsilon_0_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
+            tensor<fp16, [1, 129, 3]> input_1_cast_fp16 = log(epsilon = input_1_epsilon_0_to_fp16, x = var_37_cast_fp16)[name = tensor<string, []>("input_1_cast_fp16")];
+            tensor<string, []> input_3_pad_type_0 = const()[name = tensor<string, []>("input_3_pad_type_0"), val = tensor<string, []>("custom")];
+            tensor<int32, [2]> input_3_pad_0 = const()[name = tensor<string, []>("input_3_pad_0"), val = tensor<int32, [2]>([1, 1])];
+            tensor<int32, [1]> input_3_strides_0 = const()[name = tensor<string, []>("input_3_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [1]> input_3_dilations_0 = const()[name = tensor<string, []>("input_3_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> input_3_groups_0 = const()[name = tensor<string, []>("input_3_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [128, 129, 3]> encoder_layers_0_weight_to_fp16 = const()[name = tensor<string, []>("encoder_layers_0_weight_to_fp16"), val = tensor<fp16, [128, 129, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(132224)))];
+            tensor<fp16, [128]> encoder_layers_0_bias_to_fp16 = const()[name = tensor<string, []>("encoder_layers_0_bias_to_fp16"), val = tensor<fp16, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(231360)))];
+            tensor<fp16, [1, 128, 3]> input_3_cast_fp16 = conv(bias = encoder_layers_0_bias_to_fp16, dilations = input_3_dilations_0, groups = input_3_groups_0, pad = input_3_pad_0, pad_type = input_3_pad_type_0, strides = input_3_strides_0, weight = encoder_layers_0_weight_to_fp16, x = input_1_cast_fp16)[name = tensor<string, []>("input_3_cast_fp16")];
+            tensor<fp16, [1, 128, 3]> x_3_cast_fp16 = relu(x = input_3_cast_fp16)[name = tensor<string, []>("x_3_cast_fp16")];
+            tensor<fp16, []> const_0_to_fp16 = const()[name = tensor<string, []>("const_0_to_fp16"), val = tensor<fp16, []>(-inf)];
+            tensor<fp16, []> var_39_to_fp16 = const()[name = tensor<string, []>("op_39_to_fp16"), val = tensor<fp16, []>(0x1.388p+13)];
+            tensor<fp16, [1, 128, 3]> clip_0_cast_fp16 = clip(alpha = const_0_to_fp16, beta = var_39_to_fp16, x = x_3_cast_fp16)[name = tensor<string, []>("clip_0_cast_fp16")];
+            tensor<string, []> input_7_pad_type_0 = const()[name = tensor<string, []>("input_7_pad_type_0"), val = tensor<string, []>("custom")];
+            tensor<int32, [2]> input_7_pad_0 = const()[name = tensor<string, []>("input_7_pad_0"), val = tensor<int32, [2]>([1, 1])];
+            tensor<int32, [1]> input_7_strides_0 = const()[name = tensor<string, []>("input_7_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [1]> input_7_dilations_0 = const()[name = tensor<string, []>("input_7_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> input_7_groups_0 = const()[name = tensor<string, []>("input_7_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [64, 128, 3]> encoder_layers_2_weight_to_fp16 = const()[name = tensor<string, []>("encoder_layers_2_weight_to_fp16"), val = tensor<fp16, [64, 128, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(231680)))];
+            tensor<fp16, [64]> encoder_layers_2_bias_to_fp16 = const()[name = tensor<string, []>("encoder_layers_2_bias_to_fp16"), val = tensor<fp16, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(280896)))];
+            tensor<fp16, [1, 64, 3]> input_7_cast_fp16 = conv(bias = encoder_layers_2_bias_to_fp16, dilations = input_7_dilations_0, groups = input_7_groups_0, pad = input_7_pad_0, pad_type = input_7_pad_type_0, strides = input_7_strides_0, weight = encoder_layers_2_weight_to_fp16, x = clip_0_cast_fp16)[name = tensor<string, []>("input_7_cast_fp16")];
+            tensor<fp16, [1, 64, 3]> x_5_cast_fp16 = relu(x = input_7_cast_fp16)[name = tensor<string, []>("x_5_cast_fp16")];
+            tensor<fp16, []> const_1_to_fp16 = const()[name = tensor<string, []>("const_1_to_fp16"), val = tensor<fp16, []>(-inf)];
+            tensor<fp16, [1, 64, 3]> clip_1_cast_fp16 = clip(alpha = const_1_to_fp16, beta = var_39_to_fp16, x = x_5_cast_fp16)[name = tensor<string, []>("clip_1_cast_fp16")];
+            tensor<string, []> input_11_pad_type_0 = const()[name = tensor<string, []>("input_11_pad_type_0"), val = tensor<string, []>("custom")];
+            tensor<int32, [2]> input_11_pad_0 = const()[name = tensor<string, []>("input_11_pad_0"), val = tensor<int32, [2]>([1, 1])];
+            tensor<int32, [1]> input_11_strides_0 = const()[name = tensor<string, []>("input_11_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [1]> input_11_dilations_0 = const()[name = tensor<string, []>("input_11_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> input_11_groups_0 = const()[name = tensor<string, []>("input_11_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [64, 64, 3]> encoder_layers_4_weight_to_fp16 = const()[name = tensor<string, []>("encoder_layers_4_weight_to_fp16"), val = tensor<fp16, [64, 64, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(281088)))];
+            tensor<fp16, [64]> encoder_layers_4_bias_to_fp16 = const()[name = tensor<string, []>("encoder_layers_4_bias_to_fp16"), val = tensor<fp16, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(305728)))];
+            tensor<fp16, [1, 64, 3]> input_11_cast_fp16 = conv(bias = encoder_layers_4_bias_to_fp16, dilations = input_11_dilations_0, groups = input_11_groups_0, pad = input_11_pad_0, pad_type = input_11_pad_type_0, strides = input_11_strides_0, weight = encoder_layers_4_weight_to_fp16, x = clip_1_cast_fp16)[name = tensor<string, []>("input_11_cast_fp16")];
+            tensor<fp16, [1, 64, 3]> x_7_cast_fp16 = relu(x = input_11_cast_fp16)[name = tensor<string, []>("x_7_cast_fp16")];
+            tensor<fp16, []> const_2_to_fp16 = const()[name = tensor<string, []>("const_2_to_fp16"), val = tensor<fp16, []>(-inf)];
+            tensor<fp16, [1, 64, 3]> clip_2_cast_fp16 = clip(alpha = const_2_to_fp16, beta = var_39_to_fp16, x = x_7_cast_fp16)[name = tensor<string, []>("clip_2_cast_fp16")];
+            tensor<string, []> input_15_pad_type_0 = const()[name = tensor<string, []>("input_15_pad_type_0"), val = tensor<string, []>("custom")];
+            tensor<int32, [2]> input_15_pad_0 = const()[name = tensor<string, []>("input_15_pad_0"), val = tensor<int32, [2]>([1, 1])];
+            tensor<int32, [1]> input_15_strides_0 = const()[name = tensor<string, []>("input_15_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [1]> input_15_dilations_0 = const()[name = tensor<string, []>("input_15_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> input_15_groups_0 = const()[name = tensor<string, []>("input_15_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [128, 64, 3]> encoder_layers_6_weight_to_fp16 = const()[name = tensor<string, []>("encoder_layers_6_weight_to_fp16"), val = tensor<fp16, [128, 64, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(305920)))];
+            tensor<fp16, [128]> encoder_layers_6_bias_to_fp16 = const()[name = tensor<string, []>("encoder_layers_6_bias_to_fp16"), val = tensor<fp16, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(355136)))];
+            tensor<fp16, [1, 128, 3]> input_15_cast_fp16 = conv(bias = encoder_layers_6_bias_to_fp16, dilations = input_15_dilations_0, groups = input_15_groups_0, pad = input_15_pad_0, pad_type = input_15_pad_type_0, strides = input_15_strides_0, weight = encoder_layers_6_weight_to_fp16, x = clip_2_cast_fp16)[name = tensor<string, []>("input_15_cast_fp16")];
+            tensor<fp16, [1, 128, 3]> x_9_cast_fp16 = relu(x = input_15_cast_fp16)[name = tensor<string, []>("x_9_cast_fp16")];
+            tensor<fp16, []> const_3_to_fp16 = const()[name = tensor<string, []>("const_3_to_fp16"), val = tensor<fp16, []>(-inf)];
+            tensor<fp16, [1, 128, 3]> clip_3_cast_fp16 = clip(alpha = const_3_to_fp16, beta = var_39_to_fp16, x = x_9_cast_fp16)[name = tensor<string, []>("clip_3_cast_fp16")];
+            tensor<int32, [3]> transpose_0_perm_0 = const()[name = tensor<string, []>("transpose_0_perm_0"), val = tensor<int32, [3]>([2, 0, 1])];
+            tensor<string, []> transpose_0_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("transpose_0_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
+            tensor<int32, [1]> hx_1_axes_0 = const()[name = tensor<string, []>("hx_1_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<string, []> hidden_state_to_fp16_dtype_0 = const()[name = tensor<string, []>("hidden_state_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
+            tensor<fp16, [1, 128]> hidden_state_to_fp16 = cast(dtype = hidden_state_to_fp16_dtype_0, x = hidden_state)[name = tensor<string, []>("cast_9")];
+            tensor<fp16, [1, 1, 128]> hx_1_cast_fp16 = expand_dims(axes = hx_1_axes_0, x = hidden_state_to_fp16)[name = tensor<string, []>("hx_1_cast_fp16")];
+            tensor<int32, [1]> hx_axes_0 = const()[name = tensor<string, []>("hx_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<string, []> cell_state_to_fp16_dtype_0 = const()[name = tensor<string, []>("cell_state_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
+            tensor<fp16, [1, 128]> cell_state_to_fp16 = cast(dtype = cell_state_to_fp16_dtype_0, x = cell_state)[name = tensor<string, []>("cast_8")];
+            tensor<fp16, [1, 1, 128]> hx_cast_fp16 = expand_dims(axes = hx_axes_0, x = cell_state_to_fp16)[name = tensor<string, []>("hx_cast_fp16")];
+            tensor<fp32, [512]> concat_0 = const()[name = tensor<string, []>("concat_0"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(355456)))];
+            tensor<fp32, [512, 128]> concat_1 = const()[name = tensor<string, []>("concat_1"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(357568)))];
+            tensor<fp32, [512, 128]> concat_2 = const()[name = tensor<string, []>("concat_2"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(619776)))];
+            tensor<int32, [1]> lstm_out_batch_first_lstm_h0_squeeze_axes_0 = const()[name = tensor<string, []>("lstm_out_batch_first_lstm_h0_squeeze_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 128]> lstm_out_batch_first_lstm_h0_squeeze_cast_fp16 = squeeze(axes = lstm_out_batch_first_lstm_h0_squeeze_axes_0, x = hx_1_cast_fp16)[name = tensor<string, []>("lstm_out_batch_first_lstm_h0_squeeze_cast_fp16")];
+            tensor<string, []> lstm_out_batch_first_lstm_h0_squeeze_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("lstm_out_batch_first_lstm_h0_squeeze_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
+            tensor<int32, [1]> lstm_out_batch_first_lstm_c0_squeeze_axes_0 = const()[name = tensor<string, []>("lstm_out_batch_first_lstm_c0_squeeze_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 128]> lstm_out_batch_first_lstm_c0_squeeze_cast_fp16 = squeeze(axes = lstm_out_batch_first_lstm_c0_squeeze_axes_0, x = hx_cast_fp16)[name = tensor<string, []>("lstm_out_batch_first_lstm_c0_squeeze_cast_fp16")];
+            tensor<string, []> lstm_out_batch_first_lstm_c0_squeeze_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("lstm_out_batch_first_lstm_c0_squeeze_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
+            tensor<string, []> lstm_out_batch_first_direction_0 = const()[name = tensor<string, []>("lstm_out_batch_first_direction_0"), val = tensor<string, []>("forward")];
+            tensor<bool, []> lstm_out_batch_first_output_sequence_0 = const()[name = tensor<string, []>("lstm_out_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
+            tensor<string, []> lstm_out_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("lstm_out_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
+            tensor<string, []> lstm_out_batch_first_cell_activation_0 = const()[name = tensor<string, []>("lstm_out_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
+            tensor<string, []> lstm_out_batch_first_activation_0 = const()[name = tensor<string, []>("lstm_out_batch_first_activation_0"), val = tensor<string, []>("tanh")];
+            tensor<fp32, [1, 128]> lstm_out_batch_first_lstm_c0_squeeze_cast_fp16_to_fp32 = cast(dtype = lstm_out_batch_first_lstm_c0_squeeze_cast_fp16_to_fp32_dtype_0, x = lstm_out_batch_first_lstm_c0_squeeze_cast_fp16)[name = tensor<string, []>("cast_6")];
+            tensor<fp32, [1, 128]> lstm_out_batch_first_lstm_h0_squeeze_cast_fp16_to_fp32 = cast(dtype = lstm_out_batch_first_lstm_h0_squeeze_cast_fp16_to_fp32_dtype_0, x = lstm_out_batch_first_lstm_h0_squeeze_cast_fp16)[name = tensor<string, []>("cast_7")];
+            tensor<fp16, [3, 1, 128]> transpose_0_cast_fp16 = transpose(perm = transpose_0_perm_0, x = clip_3_cast_fp16)[name = tensor<string, []>("transpose_3")];
+            tensor<fp32, [3, 1, 128]> transpose_0_cast_fp16_to_fp32 = cast(dtype = transpose_0_cast_fp16_to_fp32_dtype_0, x = transpose_0_cast_fp16)[name = tensor<string, []>("cast_10")];
+            tensor<fp32, [3, 1, 128]> lstm_out_batch_first_0, tensor<fp32, [1, 128]> lstm_out_batch_first_1, tensor<fp32, [1, 128]> lstm_out_batch_first_2 = lstm(activation = lstm_out_batch_first_activation_0, bias = concat_0, cell_activation = lstm_out_batch_first_cell_activation_0, direction = lstm_out_batch_first_direction_0, initial_c = lstm_out_batch_first_lstm_c0_squeeze_cast_fp16_to_fp32, initial_h = lstm_out_batch_first_lstm_h0_squeeze_cast_fp16_to_fp32, output_sequence = lstm_out_batch_first_output_sequence_0, recurrent_activation = lstm_out_batch_first_recurrent_activation_0, weight_hh = concat_2, weight_ih = concat_1, x = transpose_0_cast_fp16_to_fp32)[name = tensor<string, []>("lstm_out_batch_first")];
+            tensor<int32, [3]> transpose_1_perm_0 = const()[name = tensor<string, []>("transpose_1_perm_0"), val = tensor<int32, [3]>([1, 2, 0])];
+            tensor<string, []> lstm_out_batch_first_0_to_fp16_dtype_0 = const()[name = tensor<string, []>("lstm_out_batch_first_0_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
+            tensor<int32, [1]> hn_axes_0 = const()[name = tensor<string, []>("hn_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<string, []> lstm_out_batch_first_1_to_fp16_dtype_0 = const()[name = tensor<string, []>("lstm_out_batch_first_1_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
+            tensor<fp16, [1, 128]> lstm_out_batch_first_1_to_fp16 = cast(dtype = lstm_out_batch_first_1_to_fp16_dtype_0, x = lstm_out_batch_first_1)[name = tensor<string, []>("cast_4")];
+            tensor<fp16, [1, 1, 128]> hn_cast_fp16 = expand_dims(axes = hn_axes_0, x = lstm_out_batch_first_1_to_fp16)[name = tensor<string, []>("hn_cast_fp16")];
+            tensor<int32, [1]> cn_axes_0 = const()[name = tensor<string, []>("cn_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<string, []> lstm_out_batch_first_2_to_fp16_dtype_0 = const()[name = tensor<string, []>("lstm_out_batch_first_2_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
+            tensor<fp16, [1, 128]> lstm_out_batch_first_2_to_fp16 = cast(dtype = lstm_out_batch_first_2_to_fp16_dtype_0, x = lstm_out_batch_first_2)[name = tensor<string, []>("cast_3")];
+            tensor<fp16, [1, 1, 128]> cn_cast_fp16 = expand_dims(axes = cn_axes_0, x = lstm_out_batch_first_2_to_fp16)[name = tensor<string, []>("cn_cast_fp16")];
+            tensor<string, []> input_pad_type_0 = const()[name = tensor<string, []>("input_pad_type_0"), val = tensor<string, []>("valid")];
+            tensor<int32, [1]> input_strides_0 = const()[name = tensor<string, []>("input_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> input_pad_0 = const()[name = tensor<string, []>("input_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> input_dilations_0 = const()[name = tensor<string, []>("input_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> input_groups_0 = const()[name = tensor<string, []>("input_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [1, 128, 1]> decoder_final_conv_weight_to_fp16 = const()[name = tensor<string, []>("decoder_final_conv_weight_to_fp16"), val = tensor<fp16, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(881984)))];
+            tensor<fp16, [1]> decoder_final_conv_bias_to_fp16 = const()[name = tensor<string, []>("decoder_final_conv_bias_to_fp16"), val = tensor<fp16, [1]>([0x1.dfp-5])];
+            tensor<fp16, [3, 1, 128]> lstm_out_batch_first_0_to_fp16 = cast(dtype = lstm_out_batch_first_0_to_fp16_dtype_0, x = lstm_out_batch_first_0)[name = tensor<string, []>("cast_5")];
+            tensor<fp16, [1, 128, 3]> transpose_1_cast_fp16 = transpose(perm = transpose_1_perm_0, x = lstm_out_batch_first_0_to_fp16)[name = tensor<string, []>("transpose_2")];
+            tensor<fp16, [1, 1, 3]> input_cast_fp16 = conv(bias = decoder_final_conv_bias_to_fp16, dilations = input_dilations_0, groups = input_groups_0, pad = input_pad_0, pad_type = input_pad_type_0, strides = input_strides_0, weight = decoder_final_conv_weight_to_fp16, x = transpose_1_cast_fp16)[name = tensor<string, []>("input_cast_fp16")];
+            tensor<fp16, [1, 1, 3]> out_cast_fp16 = sigmoid(x = input_cast_fp16)[name = tensor<string, []>("out_cast_fp16")];
+            tensor<int32, [1]> var_124_axes_0 = const()[name = tensor<string, []>("op_124_axes_0"), val = tensor<int32, [1]>([2])];
+            tensor<bool, []> var_124_keep_dims_0 = const()[name = tensor<string, []>("op_124_keep_dims_0"), val = tensor<bool, []>(true)];
+            tensor<fp16, [1, 1, 1]> var_124_cast_fp16 = reduce_mean(axes = var_124_axes_0, keep_dims = var_124_keep_dims_0, x = out_cast_fp16)[name = tensor<string, []>("op_124_cast_fp16")];
+            tensor<string, []> var_124_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("op_124_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
+            tensor<int32, [1]> var_125_axes_0 = const()[name = tensor<string, []>("op_125_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 128]> var_125_cast_fp16 = squeeze(axes = var_125_axes_0, x = hn_cast_fp16)[name = tensor<string, []>("op_125_cast_fp16")];
+            tensor<string, []> var_125_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("op_125_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
+            tensor<int32, [1]> var_126_axes_0 = const()[name = tensor<string, []>("op_126_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 128]> var_126_cast_fp16 = squeeze(axes = var_126_axes_0, x = cn_cast_fp16)[name = tensor<string, []>("op_126_cast_fp16")];
+            tensor<string, []> var_126_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("op_126_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
+            tensor<fp32, [1, 128]> new_cell_state = cast(dtype = var_126_cast_fp16_to_fp32_dtype_0, x = var_126_cast_fp16)[name = tensor<string, []>("cast_0")];
+            tensor<fp32, [1, 128]> new_hidden_state = cast(dtype = var_125_cast_fp16_to_fp32_dtype_0, x = var_125_cast_fp16)[name = tensor<string, []>("cast_1")];
+            tensor<fp32, [1, 1, 1]> vad_output = cast(dtype = var_124_cast_fp16_to_fp32_dtype_0, x = var_124_cast_fp16)[name = tensor<string, []>("cast_2")];
+        } -> (vad_output, new_hidden_state, new_cell_state);
+}

silero-vad-unified-v6.0.0.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:853cf34740d3f5061f977ebe2976f7c921b064261c9c4753b3a1196f2dba42b4
+size 882304