Upload 10 files

.gitattributes

@@ -45,3 +45,6 @@ Transcribe_and_Translate_Subtitles/LLM/Gemma/Tokenizer/tokenizer.json filter=lfs
 Transcribe_and_Translate_Subtitles/LLM/Gemma/12B/5c2befb3-2b4e-11f0-846e-ec8e77077d52 filter=lfs diff=lfs merge=lfs -text
 Transcribe_and_Translate_Subtitles/VAD/HumAware/HumAwareVAD.jit filter=lfs diff=lfs merge=lfs -text
 Transcribe_and_Translate_Subtitles/VAD/NVIDIA_Frame_VAD_Multilingual_MarbleNet/frame_vad_multilingual_marblenet_v2.0.nemo filter=lfs diff=lfs merge=lfs -text
+Transcribe_and_Translate_Subtitles/VAD/TEN/lib/Linux/x64/libten_vad.so filter=lfs diff=lfs merge=lfs -text
+Transcribe_and_Translate_Subtitles/VAD/TEN/lib/macOS/ten_vad.framework/Versions/A/ten_vad filter=lfs diff=lfs merge=lfs -text
+Transcribe_and_Translate_Subtitles/VAD/TEN/lib/Windows/x64/ten_vad.dll filter=lfs diff=lfs merge=lfs -text

Transcribe_and_Translate_Subtitles/VAD/NVIDIA_Frame_VAD_Multilingual_MarbleNet/F16/NVIDIA_MarbleNet.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccbd7a6ad40d64299eb6502d3b672c332b061dddc7a8152716f9cb31410fa3f4
+size 766700

Transcribe_and_Translate_Subtitles/VAD/NVIDIA_Frame_VAD_Multilingual_MarbleNet/F32/NVIDIA_MarbleNet.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cc079144772762b81fed23ab538ab1948761ed3d58aed9f580b274641dffa18
+size 1515526

Transcribe_and_Translate_Subtitles/VAD/TEN/include/ten_vad.h

@@ -0,0 +1,81 @@
+#ifndef TEN_VAD_H
+#define TEN_VAD_H
+
+#if defined(__APPLE__) || defined(__ANDROID__) || defined(__linux__)
+#define TENVAD_API __attribute__((visibility("default")))
+#elif defined(_WIN32) || defined(__CYGWIN__)
+#ifdef TENVAD_EXPORTS
+#define TENVAD_API __declspec(dllexport)
+#else
+#define TENVAD_API __declspec(dllimport)
+#endif
+#else
+#define TENVAD_API
+#endif
+
+#include <stddef.h> /* size_t */
+#include <stdint.h> /* int16_t */
+
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+
+  /**
+   * @typedef ten_vad_handle
+   * @brief Opaque handle for ten_vad instance.
+   */
+  typedef void *ten_vad_handle_t;
+
+  /**
+   * @brief Create and initialize a ten_vad instance.
+   *
+   * @param[out] handle       Pointer to receive the vad handle.
+   * @param[in]  hop_size     The number of samples between the start points of
+   * two consecutive analysis frames. (e.g., 256).
+   * @param[in]  threshold    VAD detection threshold ranging from [0.0, 1.0]
+   * This threshold is used to determine voice activity by comparing with the output probability.
+   * When probability >= threshold, voice is detected.
+   * @return 0 on success, or -1 error occurs.
+   */
+  TENVAD_API int ten_vad_create(ten_vad_handle_t *handle, size_t hop_size,
+                                float threshold);
+
+  /**
+   * @brief Process one audio frame for voice activity detection.
+   * Must call ten_vad_init() before calling this, and ten_vad_destroy() when done.
+   *
+   * @param[in]  handle           Valid VAD handle returned by ten_vad_create().
+   * @param[in]  audio_data       Pointer to an array of int16_t samples,
+   * buffer length must equal the hop size specified at ten_vad_create.
+   * @param[in]  audio_data_length  size of audio_data buffer, here should be equal to hop_size.
+   * @param[out] out_probability  Pointer to a float (size 1) that receives the
+   * voice activity probability in the range [0.0, 1.0], where higher values indicate higher confidence in voice presence.
+   * @param[out] out_flag         Pointer to an int (size 1) that receives the
+   * binary voice activity decision: 0: no voice, 1: voice detected.
+   * This flag is set to 1 when out_probability >= threshold, and 0 otherwise.
+   * @return 0 on success, or -1 error occurs.
+   */
+  TENVAD_API int ten_vad_process(ten_vad_handle_t handle, const int16_t *audio_data, size_t audio_data_length,
+                                 float *out_probability, int *out_flag);
+
+  /**
+   * @brief Destroy a ten_vad instance and release its resources.
+   *
+   * @param[in,out] handle Pointer to the ten_vad handle; set to NULL on return.
+   * @return 0 on success, or -1 error occurs.
+   */
+  TENVAD_API int ten_vad_destroy(ten_vad_handle_t *handle);
+
+  /**
+   * @brief Get the ten_vad library version string.
+   *
+   * @return The version string (e.g., "1.0.0").
+   */
+  TENVAD_API const char *ten_vad_get_version(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* TEN_VAD_H */

Transcribe_and_Translate_Subtitles/VAD/TEN/include/ten_vad.py

@@ -0,0 +1,85 @@
+from ctypes import c_int, c_int32, c_float, c_size_t, CDLL, c_void_p, POINTER
+import numpy as np
+import os
+
+
+class TenVad:
+    def __init__(self, hop_size: int = 256, threshold: float = 0.5, lib_path: str = r".lib/Linux/x64/libten_vad.so"):
+        self.hop_size = hop_size
+        self.threshold = threshold
+        if os.path.exists(
+                lib_path
+        ):
+            self.vad_library = CDLL(
+                lib_path
+            )
+        else:
+            self.vad_library = CDLL(
+                lib_path
+            )
+        self.vad_handler = c_void_p(0)
+        self.out_probability = c_float()
+        self.out_flags = c_int32()
+
+        self.vad_library.ten_vad_create.argtypes = [
+            POINTER(c_void_p),
+            c_size_t,
+            c_float,
+        ]
+        self.vad_library.ten_vad_create.restype = c_int
+
+        self.vad_library.ten_vad_destroy.argtypes = [POINTER(c_void_p)]
+        self.vad_library.ten_vad_destroy.restype = c_int
+
+        self.vad_library.ten_vad_process.argtypes = [
+            c_void_p,
+            c_void_p,
+            c_size_t,
+            POINTER(c_float),
+            POINTER(c_int32),
+        ]
+        self.vad_library.ten_vad_process.restype = c_int
+        self.create_and_init_handler()
+
+    def create_and_init_handler(self):
+        assert (
+                self.vad_library.ten_vad_create(
+                    POINTER(c_void_p)(self.vad_handler),
+                    c_size_t(self.hop_size),
+                    c_float(self.threshold),
+                )
+                == 0
+        ), "[TEN VAD]: create handler failure!"
+
+    def __del__(self):
+        assert (
+                self.vad_library.ten_vad_destroy(
+                    POINTER(c_void_p)(self.vad_handler)
+                )
+                == 0
+        ), "[TEN VAD]: destroy handler failure!"
+
+    def get_input_data(self, audio_data: np.ndarray):
+        audio_data = np.squeeze(audio_data)
+        assert (
+                len(audio_data.shape) == 1
+                and audio_data.shape[0] == self.hop_size
+        ), "[TEN VAD]: audio data shape should be [%d]" % (
+            self.hop_size
+        )
+        assert (
+                type(audio_data[0]) == np.int16
+        ), "[TEN VAD]: audio data type error, must be int16"
+        data_pointer = audio_data.__array_interface__["data"][0]
+        return c_void_p(data_pointer)
+
+    def process(self, audio_data: np.ndarray):
+        input_pointer = self.get_input_data(audio_data)
+        self.vad_library.ten_vad_process(
+            self.vad_handler,
+            input_pointer,
+            c_size_t(self.hop_size),
+            POINTER(c_float)(self.out_probability),
+            POINTER(c_int32)(self.out_flags),
+        )
+        return self.out_probability.value, self.out_flags.value

Transcribe_and_Translate_Subtitles/VAD/TEN/lib/Linux/x64/libten_vad.so

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5abfe6bf6e9a4fcea6b440240f0a9a0f431ab5006e48a4e16465ebe681ffd90f
+size 313240

Transcribe_and_Translate_Subtitles/VAD/TEN/lib/Windows/x64/ten_vad.dll

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38937f5604fa93a7941db7b9326992b792fa3731ebf9353973b3234457c6064b
+size 510464

Transcribe_and_Translate_Subtitles/VAD/TEN/lib/macOS/ten_vad.framework/Versions/A/Headers/ten_vad.h

@@ -0,0 +1,90 @@
+/*
+ * @file ten_vad.h
+ * @brief Ten Voice Activity Detection (ten_vad) C API
+ * Version: 1.0.0
+ *
+ * Provides functions to create, process, and destroy a VAD instance.
+ */
+#ifndef TEN_VAD_H
+#define TEN_VAD_H
+
+#if defined(__APPLE__) || defined(__ANDROID__) || defined(__linux__)
+#define TENVAD_API __attribute__((visibility("default")))
+#elif defined(_WIN32) || defined(__CYGWIN__)
+/**
+ * @def TENVAD_API
+ * @brief Export/import macro for ten_vad shared library symbols.
+ */
+#ifdef TENVAD_EXPORTS
+#define TENVAD_API __declspec(dllexport)
+#else
+#define TENVAD_API __declspec(dllimport)
+#endif
+#else
+#define TENVAD_API
+#endif
+
+#include <stddef.h> /* size_t */
+#include <stdint.h> /* int16_t */
+
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+
+  /**
+   * @typedef ten_vad_handle
+   * @brief Opaque handle for ten_vad instance.
+   */
+  typedef void *ten_vad_handle_t;
+
+  /**
+   * @brief Create and initialize a ten_vad instance.
+   *
+   * @param[out] handle       Pointer to receive the vad handle.
+   * @param[in]  hop_size     The number of samples between the start points of
+   * two consecutive analysis frames. (e.g., 256).
+   * @param[in]  threshold    VAD detection threshold ranging from [0.0, 1.0]
+   * (default: 0.5).
+   * @return 0 on success, or -1 error occurs.
+   */
+  TENVAD_API int ten_vad_create(ten_vad_handle_t *handle, size_t hop_size,
+                                float threshold);
+
+  /**
+   * @brief Process one audio frame for voice activity detection.
+   * Must call ten_vad_init() before calling this, and ten_vad_destroy() when done.
+   *
+   * @param[in]  handle           Valid VAD handle returned by ten_vad_create().
+   * @param[in]  audio_data       Pointer to an array of int16_t samples,
+   * buffer length must equal the hop size specified at ten_vad_create.
+   * @param[in]  audio_data_length  size of audio_data buffer, here should be equal to hop_size.
+   * @param[out] out_probability  Pointer to a float (size 1) that receives the
+   * voice activity probability in the range [0.0, 1.0].
+   * @param[out] out_flag         Pointer to an int (size 1) that receives the
+   * detection result: 0 = no voice, 1 = voice detected.
+   * @return 0 on success, or -1 error occurs.
+   */
+  TENVAD_API int ten_vad_process(ten_vad_handle_t handle, const int16_t *audio_data, size_t audio_data_length,
+                                 float *out_probability, int *out_flag);
+
+  /**
+   * @brief Destroy a ten_vad instance and release its resources.
+   *
+   * @param[in,out] handle Pointer to the ten_vad handle; set to NULL on return.
+   * @return 0 on success, or -1 error occurs.
+   */
+  TENVAD_API int ten_vad_destroy(ten_vad_handle_t *handle);
+
+  /**
+   * @brief Get the ten_vad library version string.
+   *
+   * @return The version string (e.g., "1.0.0").
+   */
+  TENVAD_API const char *ten_vad_get_version(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* TEN_VAD_H */

Transcribe_and_Translate_Subtitles/VAD/TEN/lib/macOS/ten_vad.framework/Versions/A/Resources/Info.plist

@@ -0,0 +1,44 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+	<key>BuildMachineOSBuild</key>
+	<string>23D60</string>
+	<key>CFBundleDevelopmentRegion</key>
+	<string>English</string>
+	<key>CFBundleExecutable</key>
+	<string>ten_vad</string>
+	<key>CFBundleIdentifier</key>
+	<string>com.yourcompany.ten_vad</string>
+	<key>CFBundleInfoDictionaryVersion</key>
+	<string>6.0</string>
+	<key>CFBundlePackageType</key>
+	<string>FMWK</string>
+	<key>CFBundleSignature</key>
+	<string>????</string>
+	<key>CFBundleSupportedPlatforms</key>
+	<array>
+		<string>MacOSX</string>
+	</array>
+	<key>CSResourcesFileMapped</key>
+	<true/>
+	<key>DTCompiler</key>
+	<string>com.apple.compilers.llvm.clang.1_0</string>
+	<key>DTPlatformBuild</key>
+	<string></string>
+	<key>DTPlatformName</key>
+	<string>macosx</string>
+	<key>DTPlatformVersion</key>
+	<string>14.2</string>
+	<key>DTSDKBuild</key>
+	<string>23C53</string>
+	<key>DTSDKName</key>
+	<string>macosx14.2</string>
+	<key>DTXcode</key>
+	<string>1520</string>
+	<key>DTXcodeBuild</key>
+	<string>15C500b</string>
+	<key>LSMinimumSystemVersion</key>
+	<string>10.10</string>
+</dict>
+</plist>

Transcribe_and_Translate_Subtitles/VAD/TEN/lib/macOS/ten_vad.framework/Versions/A/ten_vad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81b2de13710670bb94fef315ab50fedc903a21c04c4290c6c2ac28d8b42e715a
+size 744600