Add Qwen3-TTS Custom Voice Models

qwen3_tts/code_decoder/12hz-0.6b-customvoice/W8A16-stateful/CodeDecoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbc1ea6ac3fa1e7a07a9e19a4407badc3cdf153bc2382b6a6d408a880f3964c3
+size 243

qwen3_tts/code_decoder/12hz-0.6b-customvoice/W8A16-stateful/CodeDecoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c9e439b240cd4bf60f1a142b4c3d34ea5ae64f57c9e2fd407606cbdae76a4d2
+size 671

qwen3_tts/code_decoder/12hz-0.6b-customvoice/W8A16-stateful/CodeDecoder.mlmodelc/metadata.json

@@ -0,0 +1,159 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Int32, Palettized (8 bits), UInt8)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 3072)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 3072]",
+        "name" : "logits",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 1]",
+        "name" : "hidden_states",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 28672 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 28672, 1, 1]",
+        "name" : "key_cache_updates",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 28672 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 28672, 1, 1]",
+        "name" : "value_cache_updates",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.expandDims" : 8,
+      "Ios18.softmax" : 28,
+      "Ios18.mul" : 675,
+      "Ios18.matmul" : 56,
+      "Ios18.rsqrt" : 113,
+      "Ios16.reduceMean" : 113,
+      "Split" : 2,
+      "Ios18.greaterEqual" : 2,
+      "Select" : 2,
+      "Ios18.readState" : 2,
+      "Ios18.gather" : 2,
+      "Ios18.add" : 311,
+      "Ios18.reshape" : 224,
+      "Ios18.constexprLutToDense" : 199,
+      "Ios18.conv" : 197,
+      "Ios18.concat" : 114,
+      "Ios18.cast" : 5,
+      "Ios18.sub" : 1,
+      "Ios18.silu" : 28,
+      "Ios18.transpose" : 1,
+      "Ios18.sliceByIndex" : 560,
+      "Ios18.squeeze" : 1
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int16, Int32, UInt16)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+      {
+        "dataType" : "Float16",
+        "isOptional" : "0",
+        "formattedType" : "State (Float16 1 × 28672 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 28672, 1, 256]",
+        "name" : "self_attn_key_cache",
+        "type" : "State"
+      },
+      {
+        "dataType" : "Float16",
+        "isOptional" : "0",
+        "formattedType" : "State (Float16 1 × 28672 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 28672, 1, 256]",
+        "name" : "self_attn_value_cache",
+        "type" : "State"
+      }
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-17",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 1]",
+        "name" : "input_embeds",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "cache_length",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 256]",
+        "name" : "kv_cache_update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 256]",
+        "name" : "key_padding_mask",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "CodeDecoderWithStatefulSelfAttention_8_bit",
+    "method" : "predict"
+  }
+]

qwen3_tts/code_decoder/12hz-0.6b-customvoice/W8A16-stateful/CodeDecoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9618fdb2a96db80646b257659ec4cf01240283c211ea7fc3ad0d986d35304fa
+size 443885568

qwen3_tts/code_decoder/12hz-1.7b-customvoice/W8A16-stateful/CodeDecoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ae0466e386343e0079070977b713c72abd29e9653af179fbac1a0c656a7408b
+size 243

qwen3_tts/code_decoder/12hz-1.7b-customvoice/W8A16-stateful/CodeDecoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bac65aa923aecda6c84e1439769c47c440ab8f8afcdfd47b8d100b992dc7be1a
+size 671

qwen3_tts/code_decoder/12hz-1.7b-customvoice/W8A16-stateful/CodeDecoder.mlmodelc/metadata.json

@@ -0,0 +1,159 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Int32, Palettized (8 bits), UInt8)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 3072)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 3072]",
+        "name" : "logits",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 2048 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 2048, 1, 1]",
+        "name" : "hidden_states",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 28672 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 28672, 1, 1]",
+        "name" : "key_cache_updates",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 28672 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 28672, 1, 1]",
+        "name" : "value_cache_updates",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.expandDims" : 8,
+      "Ios18.softmax" : 28,
+      "Ios18.mul" : 675,
+      "Ios18.matmul" : 56,
+      "Ios18.rsqrt" : 113,
+      "Ios16.reduceMean" : 113,
+      "Split" : 2,
+      "Ios18.greaterEqual" : 2,
+      "Select" : 2,
+      "Ios18.readState" : 2,
+      "Ios18.gather" : 2,
+      "Ios18.add" : 311,
+      "Ios18.reshape" : 224,
+      "Ios18.constexprLutToDense" : 199,
+      "Ios18.conv" : 197,
+      "Ios18.concat" : 114,
+      "Ios18.cast" : 5,
+      "Ios18.sub" : 1,
+      "Ios18.silu" : 28,
+      "Ios18.transpose" : 1,
+      "Ios18.sliceByIndex" : 560,
+      "Ios18.squeeze" : 1
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int16, Int32, UInt16)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+      {
+        "dataType" : "Float16",
+        "isOptional" : "0",
+        "formattedType" : "State (Float16 1 × 28672 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 28672, 1, 256]",
+        "name" : "self_attn_key_cache",
+        "type" : "State"
+      },
+      {
+        "dataType" : "Float16",
+        "isOptional" : "0",
+        "formattedType" : "State (Float16 1 × 28672 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 28672, 1, 256]",
+        "name" : "self_attn_value_cache",
+        "type" : "State"
+      }
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-17",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 2048 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 2048, 1, 1]",
+        "name" : "input_embeds",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "cache_length",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 256]",
+        "name" : "kv_cache_update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 256]",
+        "name" : "key_padding_mask",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "CodeDecoderWithStatefulSelfAttention_8_bit",
+    "method" : "predict"
+  }
+]

qwen3_tts/code_decoder/12hz-1.7b-customvoice/W8A16-stateful/CodeDecoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dc0cb784a9df3f11cb4c12710e962ab108d84081f25d7e0f068a0eb0aa3ace7
+size 1416032256

qwen3_tts/code_embedder/12hz-0.6b-customvoice/W16A16/CodeEmbedder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5880de6256184e5a3bdbbdd631e8fccf030f2cf82d8930c36bc9542cd5d31645
+size 243

qwen3_tts/code_embedder/12hz-0.6b-customvoice/W16A16/CodeEmbedder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d19846f4edbd27cef1e45f4c155aad182318620c53e94897a21622774c604b8
+size 380

qwen3_tts/code_embedder/12hz-0.6b-customvoice/W16A16/CodeEmbedder.mlmodelc/metadata.json

@@ -0,0 +1,66 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Float16",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 1]",
+        "name" : "input_embeds",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.greaterEqual" : 1,
+      "Ios18.add" : 1,
+      "Ios18.cast" : 3,
+      "Select" : 1,
+      "Ios18.gather" : 1,
+      "Ios18.expandDims" : 2
+    },
+    "computePrecision" : "Mixed (Float16, Int16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-06",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "input_ids",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "CodeEmbedder",
+    "method" : "predict"
+  }
+]

qwen3_tts/code_embedder/12hz-0.6b-customvoice/W16A16/CodeEmbedder.mlmodelc/model.mil

@@ -0,0 +1,26 @@
+program(1.3)
+[buildInfo = dict<string, string>({{"coremlc-component-MIL", "3510.2.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.8.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
+{
+    func main<ios18>(tensor<int32, [1]> input_ids) {
+            int32 embeddings_batch_dims_0 = const()[name = string("embeddings_batch_dims_0"), val = int32(0)];
+            bool embeddings_validate_indices_0 = const()[name = string("embeddings_validate_indices_0"), val = bool(false)];
+            tensor<fp16, [3072, 1024]> codec_embedding_weight_to_fp16 = const()[name = string("codec_embedding_weight_to_fp16"), val = tensor<fp16, [3072, 1024]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(64)))];
+            string input_ids_to_int16_dtype_0 = const()[name = string("input_ids_to_int16_dtype_0"), val = string("int16")];
+            string cast_2_dtype_0 = const()[name = string("cast_2_dtype_0"), val = string("int32")];
+            int32 greater_equal_0_y_0 = const()[name = string("greater_equal_0_y_0"), val = int32(0)];
+            tensor<int16, [1]> input_ids_to_int16 = cast(dtype = input_ids_to_int16_dtype_0, x = input_ids)[name = string("cast_5")];
+            tensor<int32, [1]> cast_2 = cast(dtype = cast_2_dtype_0, x = input_ids_to_int16)[name = string("cast_4")];
+            tensor<bool, [1]> greater_equal_0 = greater_equal(x = cast_2, y = greater_equal_0_y_0)[name = string("greater_equal_0")];
+            int32 slice_by_index_0 = const()[name = string("slice_by_index_0"), val = int32(3072)];
+            tensor<int32, [1]> add_0 = add(x = cast_2, y = slice_by_index_0)[name = string("add_0")];
+            tensor<int32, [1]> select_0 = select(a = cast_2, b = add_0, cond = greater_equal_0)[name = string("select_0")];
+            int32 embeddings_cast_fp16_cast_uint16_axis_0 = const()[name = string("embeddings_cast_fp16_cast_uint16_axis_0"), val = int32(0)];
+            string select_0_to_int16_dtype_0 = const()[name = string("select_0_to_int16_dtype_0"), val = string("int16")];
+            tensor<int16, [1]> select_0_to_int16 = cast(dtype = select_0_to_int16_dtype_0, x = select_0)[name = string("cast_3")];
+            tensor<fp16, [1, 1024]> embeddings_cast_fp16_cast_uint16_cast_uint16 = gather(axis = embeddings_cast_fp16_cast_uint16_axis_0, batch_dims = embeddings_batch_dims_0, indices = select_0_to_int16, validate_indices = embeddings_validate_indices_0, x = codec_embedding_weight_to_fp16)[name = string("embeddings_cast_fp16_cast_uint16_cast_uint16")];
+            tensor<int32, [1]> var_18_axes_0 = const()[name = string("op_18_axes_0"), val = tensor<int32, [1]>([2])];
+            tensor<fp16, [1, 1024, 1]> var_18_cast_fp16 = expand_dims(axes = var_18_axes_0, x = embeddings_cast_fp16_cast_uint16_cast_uint16)[name = string("op_18_cast_fp16")];
+            tensor<int32, [1]> var_20_axes_0 = const()[name = string("op_20_axes_0"), val = tensor<int32, [1]>([3])];
+            tensor<fp16, [1, 1024, 1, 1]> input_embeds = expand_dims(axes = var_20_axes_0, x = var_18_cast_fp16)[name = string("op_20_cast_fp16")];
+        } -> (input_embeds);
+}

qwen3_tts/code_embedder/12hz-0.6b-customvoice/W16A16/CodeEmbedder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5bf4059e34abbb08bea47b47e1cc305a08c690c19e42526aa97c14e9dd66b4b4
+size 6291584

qwen3_tts/code_embedder/12hz-1.7b-customvoice/W16A16/CodeEmbedder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a67276394d403612b21f2042840c89e8b5feeb0d76162b0b10dc9f5be3273242
+size 243

qwen3_tts/code_embedder/12hz-1.7b-customvoice/W16A16/CodeEmbedder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52bea537a5853fa5f98b869ca6c68e052febb5257d5b5a5a14be9148e4599e6b
+size 380

qwen3_tts/code_embedder/12hz-1.7b-customvoice/W16A16/CodeEmbedder.mlmodelc/metadata.json

@@ -0,0 +1,66 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Float16",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 2048 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 2048, 1, 1]",
+        "name" : "input_embeds",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.greaterEqual" : 1,
+      "Ios18.add" : 1,
+      "Ios18.cast" : 3,
+      "Select" : 1,
+      "Ios18.gather" : 1,
+      "Ios18.expandDims" : 2
+    },
+    "computePrecision" : "Mixed (Float16, Int16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-12",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "input_ids",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "CodeEmbedder",
+    "method" : "predict"
+  }
+]

qwen3_tts/code_embedder/12hz-1.7b-customvoice/W16A16/CodeEmbedder.mlmodelc/model.mil

@@ -0,0 +1,26 @@
+program(1.3)
+[buildInfo = dict<string, string>({{"coremlc-component-MIL", "3510.2.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.8.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
+{
+    func main<ios18>(tensor<int32, [1]> input_ids) {
+            int32 embeddings_batch_dims_0 = const()[name = string("embeddings_batch_dims_0"), val = int32(0)];
+            bool embeddings_validate_indices_0 = const()[name = string("embeddings_validate_indices_0"), val = bool(false)];
+            tensor<fp16, [3072, 2048]> codec_embedding_weight_to_fp16 = const()[name = string("codec_embedding_weight_to_fp16"), val = tensor<fp16, [3072, 2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(64)))];
+            string input_ids_to_int16_dtype_0 = const()[name = string("input_ids_to_int16_dtype_0"), val = string("int16")];
+            string cast_2_dtype_0 = const()[name = string("cast_2_dtype_0"), val = string("int32")];
+            int32 greater_equal_0_y_0 = const()[name = string("greater_equal_0_y_0"), val = int32(0)];
+            tensor<int16, [1]> input_ids_to_int16 = cast(dtype = input_ids_to_int16_dtype_0, x = input_ids)[name = string("cast_5")];
+            tensor<int32, [1]> cast_2 = cast(dtype = cast_2_dtype_0, x = input_ids_to_int16)[name = string("cast_4")];
+            tensor<bool, [1]> greater_equal_0 = greater_equal(x = cast_2, y = greater_equal_0_y_0)[name = string("greater_equal_0")];
+            int32 slice_by_index_0 = const()[name = string("slice_by_index_0"), val = int32(3072)];
+            tensor<int32, [1]> add_0 = add(x = cast_2, y = slice_by_index_0)[name = string("add_0")];
+            tensor<int32, [1]> select_0 = select(a = cast_2, b = add_0, cond = greater_equal_0)[name = string("select_0")];
+            int32 embeddings_cast_fp16_cast_uint16_axis_0 = const()[name = string("embeddings_cast_fp16_cast_uint16_axis_0"), val = int32(0)];
+            string select_0_to_int16_dtype_0 = const()[name = string("select_0_to_int16_dtype_0"), val = string("int16")];
+            tensor<int16, [1]> select_0_to_int16 = cast(dtype = select_0_to_int16_dtype_0, x = select_0)[name = string("cast_3")];
+            tensor<fp16, [1, 2048]> embeddings_cast_fp16_cast_uint16_cast_uint16 = gather(axis = embeddings_cast_fp16_cast_uint16_axis_0, batch_dims = embeddings_batch_dims_0, indices = select_0_to_int16, validate_indices = embeddings_validate_indices_0, x = codec_embedding_weight_to_fp16)[name = string("embeddings_cast_fp16_cast_uint16_cast_uint16")];
+            tensor<int32, [1]> var_18_axes_0 = const()[name = string("op_18_axes_0"), val = tensor<int32, [1]>([2])];
+            tensor<fp16, [1, 2048, 1]> var_18_cast_fp16 = expand_dims(axes = var_18_axes_0, x = embeddings_cast_fp16_cast_uint16_cast_uint16)[name = string("op_18_cast_fp16")];
+            tensor<int32, [1]> var_20_axes_0 = const()[name = string("op_20_axes_0"), val = tensor<int32, [1]>([3])];
+            tensor<fp16, [1, 2048, 1, 1]> input_embeds = expand_dims(axes = var_20_axes_0, x = var_18_cast_fp16)[name = string("op_20_cast_fp16")];
+        } -> (input_embeds);
+}

qwen3_tts/code_embedder/12hz-1.7b-customvoice/W16A16/CodeEmbedder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb500c72e2fb86bbba8b89e6810c64485e177bb4bcfc3725373d2c756f28b624
+size 12583040

qwen3_tts/multi_code_decoder/12hz-0.6b-customvoice/W8A16/MultiCodeDecoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7096ffa7daeda1488227df4f181dd3a628d53238752488eb4f64c08e547992a0
+size 243

qwen3_tts/multi_code_decoder/12hz-0.6b-customvoice/W8A16/MultiCodeDecoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff1f98f3c58bd419916a56b58d94bf2a203efbbab376e7b12353a6eb7ec394a9
+size 611

qwen3_tts/multi_code_decoder/12hz-0.6b-customvoice/W8A16/MultiCodeDecoder.mlmodelc/metadata.json

@@ -0,0 +1,151 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (8 bits), UInt8)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 15 × 2048)",
+        "shortDescription" : "",
+        "shape" : "[1, 15, 2048]",
+        "name" : "all_logits",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 5120, 1, 1]",
+        "name" : "key_cache_updates",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 5120, 1, 1]",
+        "name" : "value_cache_updates",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.expandDims" : 8,
+      "Ios18.softmax" : 5,
+      "Ios18.mul" : 123,
+      "Ios18.matmul" : 10,
+      "Ios18.rsqrt" : 21,
+      "Ios16.reduceMean" : 21,
+      "Split" : 2,
+      "Ios18.greaterEqual" : 2,
+      "Select" : 2,
+      "Ios18.gather" : 2,
+      "Ios18.add" : 58,
+      "Ios18.reshape" : 40,
+      "Ios18.constexprLutToDense" : 50,
+      "Ios18.conv" : 50,
+      "Ios18.concat" : 23,
+      "Ios18.cast" : 5,
+      "Ios18.sub" : 1,
+      "Ios18.silu" : 5,
+      "Ios18.transpose" : 1,
+      "Ios18.sliceByIndex" : 100,
+      "Ios18.squeeze" : 15
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int16, Int32, UInt16)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-12",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 1]",
+        "name" : "input_embeds",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "cache_length",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 16)",
+        "shortDescription" : "",
+        "shape" : "[1, 5120, 1, 16]",
+        "name" : "key_cache",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 16)",
+        "shortDescription" : "",
+        "shape" : "[1, 5120, 1, 16]",
+        "name" : "value_cache",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 16)",
+        "shortDescription" : "",
+        "shape" : "[1, 16]",
+        "name" : "kv_cache_update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 16)",
+        "shortDescription" : "",
+        "shape" : "[1, 16]",
+        "name" : "key_padding_mask",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "MultiCodeDecoder_8_bit",
+    "method" : "predict"
+  }
+]

qwen3_tts/multi_code_decoder/12hz-0.6b-customvoice/W8A16/MultiCodeDecoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:469c3db16d54f85f771d5fbb6817f86c26a92c19e1545e0a452edaeda09f7a59
+size 110173568

qwen3_tts/multi_code_decoder/12hz-1.7b-customvoice/W8A16/MultiCodeDecoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90d4090f7d667dc9e55ca0698f543928678e69a645d71fc343c52e7398724f65
+size 243

qwen3_tts/multi_code_decoder/12hz-1.7b-customvoice/W8A16/MultiCodeDecoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91e4d1bd298329d7300432d70c3054ccc58f5f250cb500a44f0b7d4075720186
+size 611

qwen3_tts/multi_code_decoder/12hz-1.7b-customvoice/W8A16/MultiCodeDecoder.mlmodelc/metadata.json

@@ -0,0 +1,151 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (8 bits), UInt8)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 15 × 2048)",
+        "shortDescription" : "",
+        "shape" : "[1, 15, 2048]",
+        "name" : "all_logits",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 5120, 1, 1]",
+        "name" : "key_cache_updates",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 5120, 1, 1]",
+        "name" : "value_cache_updates",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.expandDims" : 8,
+      "Ios18.softmax" : 5,
+      "Ios18.mul" : 123,
+      "Ios18.matmul" : 10,
+      "Ios18.rsqrt" : 21,
+      "Ios16.reduceMean" : 21,
+      "Split" : 2,
+      "Ios18.greaterEqual" : 2,
+      "Select" : 2,
+      "Ios18.gather" : 2,
+      "Ios18.add" : 58,
+      "Ios18.reshape" : 40,
+      "Ios18.constexprLutToDense" : 51,
+      "Ios18.conv" : 51,
+      "Ios18.concat" : 23,
+      "Ios18.cast" : 5,
+      "Ios18.sub" : 1,
+      "Ios18.silu" : 5,
+      "Ios18.transpose" : 1,
+      "Ios18.sliceByIndex" : 100,
+      "Ios18.squeeze" : 15
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int16, Int32, UInt16)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-11",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 2048 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 2048, 1, 1]",
+        "name" : "input_embeds",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "cache_length",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 16)",
+        "shortDescription" : "",
+        "shape" : "[1, 5120, 1, 16]",
+        "name" : "key_cache",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 16)",
+        "shortDescription" : "",
+        "shape" : "[1, 5120, 1, 16]",
+        "name" : "value_cache",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 16)",
+        "shortDescription" : "",
+        "shape" : "[1, 16]",
+        "name" : "kv_cache_update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 16)",
+        "shortDescription" : "",
+        "shape" : "[1, 16]",
+        "name" : "key_padding_mask",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "MultiCodeDecoder_8_bit",
+    "method" : "predict"
+  }
+]

qwen3_tts/multi_code_decoder/12hz-1.7b-customvoice/W8A16/MultiCodeDecoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c3d559c2413b8bffbd5f1d4e1a3cab6b084e66c369c8d02f8d1e5cee509cc85
+size 112273472

qwen3_tts/multi_code_embedder/12hz-0.6b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18324dfd306ef038fcb0e0a279809ca8f8ea24e92d6bac7c15fbd81df6102220
+size 243

qwen3_tts/multi_code_embedder/12hz-0.6b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08f3c2e364b7d50fd16600725143bdfe80127688d9dd05a9391a90fdcf30d29c
+size 380

qwen3_tts/multi_code_embedder/12hz-0.6b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/metadata.json

@@ -0,0 +1,66 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Float16",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 1]",
+        "name" : "input_embeds",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.greaterEqual" : 1,
+      "Ios18.add" : 1,
+      "Ios18.cast" : 3,
+      "Select" : 1,
+      "Ios18.gather" : 1,
+      "Ios18.expandDims" : 2
+    },
+    "computePrecision" : "Mixed (Float16, Int16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-06",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "input_ids",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "MultiCodeEmbedder",
+    "method" : "predict"
+  }
+]

qwen3_tts/multi_code_embedder/12hz-0.6b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/model.mil

@@ -0,0 +1,26 @@
+program(1.3)
+[buildInfo = dict<string, string>({{"coremlc-component-MIL", "3510.2.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.8.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
+{
+    func main<ios18>(tensor<int32, [1]> input_ids) {
+            int32 embeddings_batch_dims_0 = const()[name = string("embeddings_batch_dims_0"), val = int32(0)];
+            bool embeddings_validate_indices_0 = const()[name = string("embeddings_validate_indices_0"), val = bool(false)];
+            tensor<fp16, [30720, 1024]> codec_embedding_weight_to_fp16 = const()[name = string("codec_embedding_weight_to_fp16"), val = tensor<fp16, [30720, 1024]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(64)))];
+            string input_ids_to_int16_dtype_0 = const()[name = string("input_ids_to_int16_dtype_0"), val = string("int16")];
+            string cast_2_dtype_0 = const()[name = string("cast_2_dtype_0"), val = string("int32")];
+            int32 greater_equal_0_y_0 = const()[name = string("greater_equal_0_y_0"), val = int32(0)];
+            tensor<int16, [1]> input_ids_to_int16 = cast(dtype = input_ids_to_int16_dtype_0, x = input_ids)[name = string("cast_5")];
+            tensor<int32, [1]> cast_2 = cast(dtype = cast_2_dtype_0, x = input_ids_to_int16)[name = string("cast_4")];
+            tensor<bool, [1]> greater_equal_0 = greater_equal(x = cast_2, y = greater_equal_0_y_0)[name = string("greater_equal_0")];
+            int32 slice_by_index_0 = const()[name = string("slice_by_index_0"), val = int32(30720)];
+            tensor<int32, [1]> add_0 = add(x = cast_2, y = slice_by_index_0)[name = string("add_0")];
+            tensor<int32, [1]> select_0 = select(a = cast_2, b = add_0, cond = greater_equal_0)[name = string("select_0")];
+            int32 embeddings_cast_fp16_cast_uint16_axis_0 = const()[name = string("embeddings_cast_fp16_cast_uint16_axis_0"), val = int32(0)];
+            string select_0_to_int16_dtype_0 = const()[name = string("select_0_to_int16_dtype_0"), val = string("int16")];
+            tensor<int16, [1]> select_0_to_int16 = cast(dtype = select_0_to_int16_dtype_0, x = select_0)[name = string("cast_3")];
+            tensor<fp16, [1, 1024]> embeddings_cast_fp16_cast_uint16_cast_uint16 = gather(axis = embeddings_cast_fp16_cast_uint16_axis_0, batch_dims = embeddings_batch_dims_0, indices = select_0_to_int16, validate_indices = embeddings_validate_indices_0, x = codec_embedding_weight_to_fp16)[name = string("embeddings_cast_fp16_cast_uint16_cast_uint16")];
+            tensor<int32, [1]> var_18_axes_0 = const()[name = string("op_18_axes_0"), val = tensor<int32, [1]>([2])];
+            tensor<fp16, [1, 1024, 1]> var_18_cast_fp16 = expand_dims(axes = var_18_axes_0, x = embeddings_cast_fp16_cast_uint16_cast_uint16)[name = string("op_18_cast_fp16")];
+            tensor<int32, [1]> var_20_axes_0 = const()[name = string("op_20_axes_0"), val = tensor<int32, [1]>([3])];
+            tensor<fp16, [1, 1024, 1, 1]> input_embeds = expand_dims(axes = var_20_axes_0, x = var_18_cast_fp16)[name = string("op_20_cast_fp16")];
+        } -> (input_embeds);
+}

qwen3_tts/multi_code_embedder/12hz-0.6b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f318d23d7db53b299b3aa472844a721d2b6245550fb57558b0d2c5ebbcd3fd82
+size 62914688

qwen3_tts/multi_code_embedder/12hz-1.7b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59e7994cafd7a43d8c1af61044f330357ba05ee9817d11a3e51bcc3000d598f0
+size 243

qwen3_tts/multi_code_embedder/12hz-1.7b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d20121b9a714012ce61f8257ebd047be23c2994a7ef0a1c40225fdbdc72c2dd8
+size 380

qwen3_tts/multi_code_embedder/12hz-1.7b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/metadata.json

@@ -0,0 +1,66 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Float16",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 2048 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 2048, 1, 1]",
+        "name" : "input_embeds",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.greaterEqual" : 1,
+      "Ios18.add" : 1,
+      "Ios18.cast" : 3,
+      "Select" : 1,
+      "Ios18.gather" : 1,
+      "Ios18.expandDims" : 2
+    },
+    "computePrecision" : "Mixed (Float16, Int16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-12",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "input_ids",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "MultiCodeEmbedder",
+    "method" : "predict"
+  }
+]

qwen3_tts/multi_code_embedder/12hz-1.7b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/model.mil

@@ -0,0 +1,26 @@
+program(1.3)
+[buildInfo = dict<string, string>({{"coremlc-component-MIL", "3510.2.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.8.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
+{
+    func main<ios18>(tensor<int32, [1]> input_ids) {
+            int32 embeddings_batch_dims_0 = const()[name = string("embeddings_batch_dims_0"), val = int32(0)];
+            bool embeddings_validate_indices_0 = const()[name = string("embeddings_validate_indices_0"), val = bool(false)];
+            tensor<fp16, [30720, 2048]> codec_embedding_weight_to_fp16 = const()[name = string("codec_embedding_weight_to_fp16"), val = tensor<fp16, [30720, 2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(64)))];
+            string input_ids_to_int16_dtype_0 = const()[name = string("input_ids_to_int16_dtype_0"), val = string("int16")];
+            string cast_2_dtype_0 = const()[name = string("cast_2_dtype_0"), val = string("int32")];
+            int32 greater_equal_0_y_0 = const()[name = string("greater_equal_0_y_0"), val = int32(0)];
+            tensor<int16, [1]> input_ids_to_int16 = cast(dtype = input_ids_to_int16_dtype_0, x = input_ids)[name = string("cast_5")];
+            tensor<int32, [1]> cast_2 = cast(dtype = cast_2_dtype_0, x = input_ids_to_int16)[name = string("cast_4")];
+            tensor<bool, [1]> greater_equal_0 = greater_equal(x = cast_2, y = greater_equal_0_y_0)[name = string("greater_equal_0")];
+            int32 slice_by_index_0 = const()[name = string("slice_by_index_0"), val = int32(30720)];
+            tensor<int32, [1]> add_0 = add(x = cast_2, y = slice_by_index_0)[name = string("add_0")];
+            tensor<int32, [1]> select_0 = select(a = cast_2, b = add_0, cond = greater_equal_0)[name = string("select_0")];
+            int32 embeddings_cast_fp16_cast_uint16_axis_0 = const()[name = string("embeddings_cast_fp16_cast_uint16_axis_0"), val = int32(0)];
+            string select_0_to_int16_dtype_0 = const()[name = string("select_0_to_int16_dtype_0"), val = string("int16")];
+            tensor<int16, [1]> select_0_to_int16 = cast(dtype = select_0_to_int16_dtype_0, x = select_0)[name = string("cast_3")];
+            tensor<fp16, [1, 2048]> embeddings_cast_fp16_cast_uint16_cast_uint16 = gather(axis = embeddings_cast_fp16_cast_uint16_axis_0, batch_dims = embeddings_batch_dims_0, indices = select_0_to_int16, validate_indices = embeddings_validate_indices_0, x = codec_embedding_weight_to_fp16)[name = string("embeddings_cast_fp16_cast_uint16_cast_uint16")];
+            tensor<int32, [1]> var_18_axes_0 = const()[name = string("op_18_axes_0"), val = tensor<int32, [1]>([2])];
+            tensor<fp16, [1, 2048, 1]> var_18_cast_fp16 = expand_dims(axes = var_18_axes_0, x = embeddings_cast_fp16_cast_uint16_cast_uint16)[name = string("op_18_cast_fp16")];
+            tensor<int32, [1]> var_20_axes_0 = const()[name = string("op_20_axes_0"), val = tensor<int32, [1]>([3])];
+            tensor<fp16, [1, 2048, 1, 1]> input_embeds = expand_dims(axes = var_20_axes_0, x = var_18_cast_fp16)[name = string("op_20_cast_fp16")];
+        } -> (input_embeds);
+}

qwen3_tts/multi_code_embedder/12hz-1.7b-customvoice/W16A16/MultiCodeEmbedder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46f8ac7940915bf46cb15cd16ac06c1cf306044aebf6ade6253831539e61cb88
+size 125829248

qwen3_tts/speech_decoder/12hz-0.6b-customvoice/W8A16/SpeechDecoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e5662e1e7fb3aa28cbbf01e35c3dd4cc11c795ce151f05068ec220092157eb8
+size 243

qwen3_tts/speech_decoder/12hz-0.6b-customvoice/W8A16/SpeechDecoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:608c8c933a512f362e79193430195402b64781ee57731099225c2e83ab9dbb8d
+size 681

qwen3_tts/speech_decoder/12hz-0.6b-customvoice/W8A16/SpeechDecoder.mlmodelc/metadata.json

@@ -0,0 +1,178 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Int32, Palettized (8 bits), UInt8)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 1920)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 1920]",
+        "name" : "audio",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 8192 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 8192, 1, 1]",
+        "name" : "key_cache_updates",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 8192 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 8192, 1, 1]",
+        "name" : "value_cache_updates",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 1]",
+        "name" : "hidden_context_update",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.expandDims" : 26,
+      "Ios18.mul" : 234,
+      "Ios18.softmax" : 8,
+      "Ios18.rsqrt" : 17,
+      "Ios18.matmul" : 16,
+      "Ios16.reduceMean" : 17,
+      "Split" : 3,
+      "Ios18.greaterEqual" : 1,
+      "Select" : 1,
+      "Ios18.sin" : 29,
+      "Tile" : 2,
+      "Ios18.gather" : 18,
+      "Ios18.add" : 132,
+      "Ios18.layerNorm" : 2,
+      "Ios18.reshape" : 56,
+      "Pad" : 17,
+      "Ios18.constexprLutToDense" : 116,
+      "Ios18.conv" : 93,
+      "Ios18.concat" : 19,
+      "Ios18.transpose" : 30,
+      "Ios18.sub" : 1,
+      "Ios18.cast" : 19,
+      "Ios18.silu" : 8,
+      "Ios18.gelu" : 2,
+      "Ios18.clip" : 1,
+      "Ios18.convTranspose" : 6,
+      "Ios18.sliceByIndex" : 39,
+      "Ios18.squeeze" : 18
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int16, Int32, UInt16)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-09",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1 × 16 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 16, 1]",
+        "name" : "audio_codes",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "cache_length",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 8192 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 8192, 1, 256]",
+        "name" : "key_cache",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 8192 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 8192, 1, 256]",
+        "name" : "value_cache",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 256]",
+        "name" : "kv_cache_update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 256]",
+        "name" : "key_padding_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 4)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 4]",
+        "name" : "hidden_context",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "SpeechDecoder_8_bit",
+    "method" : "predict"
+  }
+]

qwen3_tts/speech_decoder/12hz-0.6b-customvoice/W8A16/SpeechDecoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a9a352536383e01d53d1d0ab227145e6e02880b2281715bc0ff12a44f7be89f
+size 114215488

qwen3_tts/speech_decoder/12hz-1.7b-customvoice/W8A16/SpeechDecoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a7c001fcef8399080af242e87d12ad44d7fc37fded8007e7fce65461346a52c
+size 243

qwen3_tts/speech_decoder/12hz-1.7b-customvoice/W8A16/SpeechDecoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:554158c2df351252d84d8295d438b004a9bae5e65ae9dcd802f13de13bacd41a
+size 681

qwen3_tts/speech_decoder/12hz-1.7b-customvoice/W8A16/SpeechDecoder.mlmodelc/metadata.json

@@ -0,0 +1,177 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Int32)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 1920)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 1920]",
+        "name" : "audio",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 8192 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 8192, 1, 1]",
+        "name" : "key_cache_updates",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 8192 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 8192, 1, 1]",
+        "name" : "value_cache_updates",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 1]",
+        "name" : "hidden_context_update",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.expandDims" : 26,
+      "Ios18.mul" : 234,
+      "Ios18.softmax" : 8,
+      "Ios18.matmul" : 16,
+      "Ios18.rsqrt" : 17,
+      "Ios16.reduceMean" : 17,
+      "Split" : 3,
+      "Ios18.greaterEqual" : 1,
+      "Select" : 1,
+      "Ios18.sin" : 29,
+      "Tile" : 2,
+      "Ios18.gather" : 18,
+      "Ios18.add" : 132,
+      "Ios18.layerNorm" : 2,
+      "Ios18.reshape" : 56,
+      "Pad" : 17,
+      "Ios18.conv" : 93,
+      "Ios18.concat" : 19,
+      "Ios18.transpose" : 30,
+      "Ios18.sub" : 1,
+      "Ios18.cast" : 19,
+      "Ios18.silu" : 8,
+      "Ios18.gelu" : 2,
+      "Ios18.clip" : 1,
+      "Ios18.convTranspose" : 6,
+      "Ios18.sliceByIndex" : 39,
+      "Ios18.squeeze" : 18
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int16, Int32, UInt16)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-02-12",
+      "com.github.apple.coremltools.source" : "torch==2.8.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1 × 16 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 16, 1]",
+        "name" : "audio_codes",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "cache_length",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 8192 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 8192, 1, 256]",
+        "name" : "key_cache",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 8192 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 8192, 1, 256]",
+        "name" : "value_cache",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 256]",
+        "name" : "kv_cache_update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 256]",
+        "name" : "key_padding_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1024 × 1 × 4)",
+        "shortDescription" : "",
+        "shape" : "[1, 1024, 1, 4]",
+        "name" : "hidden_context",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "SpeechDecoder",
+    "method" : "predict"
+  }
+]

qwen3_tts/speech_decoder/12hz-1.7b-customvoice/W8A16/SpeechDecoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c90dc149272173de6cdae46826e1ef4265c53b8477b1de19c02843d4e38729e7
+size 228134208