xuning / sherpaonnx

转到一个项目
Toggle navigation Toggle navigation pinning
yujinqiu
Committed by GitHub
9091917e 1 parent 552a267c

Generate SRT from audio (#341)

内嵌 并排对比
正在显示 19 个修改的文件 包含 984 行增加 和 0 行删除
... ... @@ -18,3 +18,37 @@ extension AVAudioPCMBuffer {
return self.audioBufferList.pointee.mBuffers.array()
}
}
extension TimeInterval {
var hourMinuteSecondMS: String {
String(format: "%d:%02d:%02d,%03d", hour, minute, second, millisecond)
}
var hour: Int {
Int((self / 3600).truncatingRemainder(dividingBy: 3600))
}
var minute: Int {
Int((self / 60).truncatingRemainder(dividingBy: 60))
}
var second: Int {
Int(truncatingRemainder(dividingBy: 60))
}
var millisecond: Int {
Int((self * 1000).truncatingRemainder(dividingBy: 1000))
}
}
extension String {
var fileURL: URL {
return URL(fileURLWithPath: self)
}
var pathExtension: String {
return fileURL.pathExtension
}
var lastPathComponent: String {
return fileURL.lastPathComponent
}
var stringByDeletingPathExtension: String {
return fileURL.deletingPathExtension().path
}
}
... ...
tiny.en-tokens.txt
*.onnx
*.ort
... ...
// !$*UTF8*$!
{
archiveVersion = 1;
classes = {
};
objectVersion = 56;
objects = {
/* Begin PBXBuildFile section */
DE081A8F2ABF287C00E8CD63 /* SherpaOnnx.swift in Sources */ = {isa = PBXBuildFile; fileRef = DE081A8E2ABF287C00E8CD63 /* SherpaOnnx.swift */; };
DE081A922ABF28D400E8CD63 /* SubtitleViewModel.swift in Sources */ = {isa = PBXBuildFile; fileRef = DE081A912ABF28D400E8CD63 /* SubtitleViewModel.swift */; };
DE081A952ABFC60E00E8CD63 /* Model.swift in Sources */ = {isa = PBXBuildFile; fileRef = DE081A942ABFC60E00E8CD63 /* Model.swift */; };
DE081AAF2ABFF35400E8CD63 /* UTType.swift in Sources */ = {isa = PBXBuildFile; fileRef = DE081AAE2ABFF35400E8CD63 /* UTType.swift */; };
DE081AB12ABFFEEE00E8CD63 /* Document.swift in Sources */ = {isa = PBXBuildFile; fileRef = DE081AB02ABFFEEE00E8CD63 /* Document.swift */; };
DE081AB32ABFFF2600E8CD63 /* Errors.swift in Sources */ = {isa = PBXBuildFile; fileRef = DE081AB22ABFFF2600E8CD63 /* Errors.swift */; };
DE8C85A62ABF23E100F667E3 /* onnxruntime.xcframework in Frameworks */ = {isa = PBXBuildFile; fileRef = DE8C85A52ABF23E100F667E3 /* onnxruntime.xcframework */; };
DE8C85AA2ABF23FA00F667E3 /* sherpa-onnx.xcframework in Frameworks */ = {isa = PBXBuildFile; fileRef = DE8C85A92ABF23FA00F667E3 /* sherpa-onnx.xcframework */; };
DE8C85B22ABF257200F667E3 /* SpeechSegment.swift in Sources */ = {isa = PBXBuildFile; fileRef = DE8C85B12ABF257200F667E3 /* SpeechSegment.swift */; };
DEA22DF12AC1796C00549373 /* tiny.en-encoder.int8.onnx in Resources */ = {isa = PBXBuildFile; fileRef = DEA22DEE2AC1796C00549373 /* tiny.en-encoder.int8.onnx */; };
DEA22DF22AC1796C00549373 /* tiny.en-decoder.int8.onnx in Resources */ = {isa = PBXBuildFile; fileRef = DEA22DEF2AC1796C00549373 /* tiny.en-decoder.int8.onnx */; };
DEA22DF32AC1796C00549373 /* tiny.en-tokens.txt in Resources */ = {isa = PBXBuildFile; fileRef = DEA22DF02AC1796C00549373 /* tiny.en-tokens.txt */; };
DEA22DF52AC179E500549373 /* silero_vad.onnx in Resources */ = {isa = PBXBuildFile; fileRef = DEA22DF42AC179CA00549373 /* silero_vad.onnx */; };
DEA657152ABF19730066A81D /* SherpaOnnxSubtitleApp.swift in Sources */ = {isa = PBXBuildFile; fileRef = DEA657142ABF19730066A81D /* SherpaOnnxSubtitleApp.swift */; };
DEA657172ABF19730066A81D /* ContentView.swift in Sources */ = {isa = PBXBuildFile; fileRef = DEA657162ABF19730066A81D /* ContentView.swift */; };
DEA657192ABF19740066A81D /* Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = DEA657182ABF19740066A81D /* Assets.xcassets */; };
DEA6571C2ABF19740066A81D /* Preview Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = DEA6571B2ABF19740066A81D /* Preview Assets.xcassets */; };
DEA657232ABF20130066A81D /* Audio.swift in Sources */ = {isa = PBXBuildFile; fileRef = DEA657222ABF20130066A81D /* Audio.swift */; };
DED059702AC136FF00122A60 /* Extension.swift in Sources */ = {isa = PBXBuildFile; fileRef = DED0596F2AC136FF00122A60 /* Extension.swift */; };
/* End PBXBuildFile section */
/* Begin PBXFileReference section */
DE081A8E2ABF287C00E8CD63 /* SherpaOnnx.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; name = SherpaOnnx.swift; path = "../../../swift-api-examples/SherpaOnnx.swift"; sourceTree = "<group>"; };
DE081A912ABF28D400E8CD63 /* SubtitleViewModel.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = SubtitleViewModel.swift; sourceTree = "<group>"; };
DE081A942ABFC60E00E8CD63 /* Model.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; name = Model.swift; path = ../../SherpaOnnx2Pass/SherpaOnnx2Pass/Model.swift; sourceTree = "<group>"; };
DE081AAC2ABFF30A00E8CD63 /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist; path = Info.plist; sourceTree = "<group>"; };
DE081AAE2ABFF35400E8CD63 /* UTType.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = UTType.swift; sourceTree = "<group>"; };
DE081AB02ABFFEEE00E8CD63 /* Document.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = Document.swift; sourceTree = "<group>"; };
DE081AB22ABFFF2600E8CD63 /* Errors.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = Errors.swift; sourceTree = "<group>"; };
DE8C85A52ABF23E100F667E3 /* onnxruntime.xcframework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.xcframework; name = onnxruntime.xcframework; path = "../../build-ios/ios-onnxruntime/1.16.0/onnxruntime.xcframework"; sourceTree = "<group>"; };
DE8C85A92ABF23FA00F667E3 /* sherpa-onnx.xcframework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.xcframework; name = "sherpa-onnx.xcframework"; path = "../../build-ios/sherpa-onnx.xcframework"; sourceTree = "<group>"; };
DE8C85B12ABF257200F667E3 /* SpeechSegment.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = SpeechSegment.swift; sourceTree = "<group>"; };
DEA22DEE2AC1796C00549373 /* tiny.en-encoder.int8.onnx */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = "tiny.en-encoder.int8.onnx"; sourceTree = "<group>"; };
DEA22DEF2AC1796C00549373 /* tiny.en-decoder.int8.onnx */ = {isa = PBXFileReference; lastKnownFileType = file; path = "tiny.en-decoder.int8.onnx"; sourceTree = "<group>"; };
DEA22DF02AC1796C00549373 /* tiny.en-tokens.txt */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = "tiny.en-tokens.txt"; sourceTree = "<group>"; };
DEA22DF42AC179CA00549373 /* silero_vad.onnx */ = {isa = PBXFileReference; lastKnownFileType = file; path = silero_vad.onnx; sourceTree = "<group>"; };
DEA657112ABF19730066A81D /* SherpaOnnxSubtitle.app */ = {isa = PBXFileReference; explicitFileType = wrapper.application; includeInIndex = 0; path = SherpaOnnxSubtitle.app; sourceTree = BUILT_PRODUCTS_DIR; };
DEA657142ABF19730066A81D /* SherpaOnnxSubtitleApp.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = SherpaOnnxSubtitleApp.swift; sourceTree = "<group>"; };
DEA657162ABF19730066A81D /* ContentView.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = ContentView.swift; sourceTree = "<group>"; };
DEA657182ABF19740066A81D /* Assets.xcassets */ = {isa = PBXFileReference; lastKnownFileType = folder.assetcatalog; path = Assets.xcassets; sourceTree = "<group>"; };
DEA6571B2ABF19740066A81D /* Preview Assets.xcassets */ = {isa = PBXFileReference; lastKnownFileType = folder.assetcatalog; path = "Preview Assets.xcassets"; sourceTree = "<group>"; };
DEA657222ABF20130066A81D /* Audio.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Audio.swift; sourceTree = "<group>"; };
DED0596F2AC136FF00122A60 /* Extension.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; name = Extension.swift; path = ../../../SherpaOnnx2Pass/SherpaOnnx2Pass/Extension.swift; sourceTree = "<group>"; };
/* End PBXFileReference section */
/* Begin PBXFrameworksBuildPhase section */
DEA6570E2ABF19730066A81D /* Frameworks */ = {
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
DE8C85A62ABF23E100F667E3 /* onnxruntime.xcframework in Frameworks */,
DE8C85AA2ABF23FA00F667E3 /* sherpa-onnx.xcframework in Frameworks */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXFrameworksBuildPhase section */
/* Begin PBXGroup section */
DE081A902ABF28BE00E8CD63 /* Models */ = {
isa = PBXGroup;
children = (
DEA657222ABF20130066A81D /* Audio.swift */,
DE8C85B12ABF257200F667E3 /* SpeechSegment.swift */,
DE081AB02ABFFEEE00E8CD63 /* Document.swift */,
DE081AB22ABFFF2600E8CD63 /* Errors.swift */,
);
path = Models;
sourceTree = "<group>";
};
DE081AAD2ABFF34900E8CD63 /* Extensions */ = {
isa = PBXGroup;
children = (
DED0596F2AC136FF00122A60 /* Extension.swift */,
DE081AAE2ABFF35400E8CD63 /* UTType.swift */,
);
path = Extensions;
sourceTree = "<group>";
};
DE8C85A42ABF23E100F667E3 /* Frameworks */ = {
isa = PBXGroup;
children = (
DE8C85A92ABF23FA00F667E3 /* sherpa-onnx.xcframework */,
DE8C85A52ABF23E100F667E3 /* onnxruntime.xcframework */,
);
name = Frameworks;
sourceTree = "<group>";
};
DEA657082ABF19730066A81D = {
isa = PBXGroup;
children = (
DEA657132ABF19730066A81D /* SherpaOnnxSubtitle */,
DEA657122ABF19730066A81D /* Products */,
DE8C85A42ABF23E100F667E3 /* Frameworks */,
);
sourceTree = "<group>";
};
DEA657122ABF19730066A81D /* Products */ = {
isa = PBXGroup;
children = (
DEA657112ABF19730066A81D /* SherpaOnnxSubtitle.app */,
);
name = Products;
sourceTree = "<group>";
};
DEA657132ABF19730066A81D /* SherpaOnnxSubtitle */ = {
isa = PBXGroup;
children = (
DEA22DEF2AC1796C00549373 /* tiny.en-decoder.int8.onnx */,
DEA22DEE2AC1796C00549373 /* tiny.en-encoder.int8.onnx */,
DEA22DF02AC1796C00549373 /* tiny.en-tokens.txt */,
DEA22DF42AC179CA00549373 /* silero_vad.onnx */,
DE081AAC2ABFF30A00E8CD63 /* Info.plist */,
DE081A8E2ABF287C00E8CD63 /* SherpaOnnx.swift */,
DEA657142ABF19730066A81D /* SherpaOnnxSubtitleApp.swift */,
DEA657162ABF19730066A81D /* ContentView.swift */,
DE081A912ABF28D400E8CD63 /* SubtitleViewModel.swift */,
DE081AAD2ABFF34900E8CD63 /* Extensions */,
DE081A942ABFC60E00E8CD63 /* Model.swift */,
DE081A902ABF28BE00E8CD63 /* Models */,
DEA657182ABF19740066A81D /* Assets.xcassets */,
DEA6571A2ABF19740066A81D /* Preview Content */,
);
path = SherpaOnnxSubtitle;
sourceTree = "<group>";
};
DEA6571A2ABF19740066A81D /* Preview Content */ = {
isa = PBXGroup;
children = (
DEA6571B2ABF19740066A81D /* Preview Assets.xcassets */,
);
path = "Preview Content";
sourceTree = "<group>";
};
/* End PBXGroup section */
/* Begin PBXNativeTarget section */
DEA657102ABF19730066A81D /* SherpaOnnxSubtitle */ = {
isa = PBXNativeTarget;
buildConfigurationList = DEA6571F2ABF19740066A81D /* Build configuration list for PBXNativeTarget "SherpaOnnxSubtitle" */;
buildPhases = (
DEA6570D2ABF19730066A81D /* Sources */,
DEA6570E2ABF19730066A81D /* Frameworks */,
DEA6570F2ABF19730066A81D /* Resources */,
);
buildRules = (
);
dependencies = (
);
name = SherpaOnnxSubtitle;
productName = SherpaOnnxSubtitle;
productReference = DEA657112ABF19730066A81D /* SherpaOnnxSubtitle.app */;
productType = "com.apple.product-type.application";
};
/* End PBXNativeTarget section */
/* Begin PBXProject section */
DEA657092ABF19730066A81D /* Project object */ = {
isa = PBXProject;
attributes = {
BuildIndependentTargetsInParallel = 1;
LastSwiftUpdateCheck = 1500;
LastUpgradeCheck = 1500;
TargetAttributes = {
DEA657102ABF19730066A81D = {
CreatedOnToolsVersion = 15.0;
};
};
};
buildConfigurationList = DEA6570C2ABF19730066A81D /* Build configuration list for PBXProject "SherpaOnnxSubtitle" */;
compatibilityVersion = "Xcode 14.0";
developmentRegion = en;
hasScannedForEncodings = 0;
knownRegions = (
en,
Base,
);
mainGroup = DEA657082ABF19730066A81D;
productRefGroup = DEA657122ABF19730066A81D /* Products */;
projectDirPath = "";
projectRoot = "";
targets = (
DEA657102ABF19730066A81D /* SherpaOnnxSubtitle */,
);
};
/* End PBXProject section */
/* Begin PBXResourcesBuildPhase section */
DEA6570F2ABF19730066A81D /* Resources */ = {
isa = PBXResourcesBuildPhase;
buildActionMask = 2147483647;
files = (
DEA22DF52AC179E500549373 /* silero_vad.onnx in Resources */,
DEA6571C2ABF19740066A81D /* Preview Assets.xcassets in Resources */,
DEA22DF12AC1796C00549373 /* tiny.en-encoder.int8.onnx in Resources */,
DEA657192ABF19740066A81D /* Assets.xcassets in Resources */,
DEA22DF22AC1796C00549373 /* tiny.en-decoder.int8.onnx in Resources */,
DEA22DF32AC1796C00549373 /* tiny.en-tokens.txt in Resources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXResourcesBuildPhase section */
/* Begin PBXSourcesBuildPhase section */
DEA6570D2ABF19730066A81D /* Sources */ = {
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
DE081AAF2ABFF35400E8CD63 /* UTType.swift in Sources */,
DE8C85B22ABF257200F667E3 /* SpeechSegment.swift in Sources */,
DE081A922ABF28D400E8CD63 /* SubtitleViewModel.swift in Sources */,
DE081AB12ABFFEEE00E8CD63 /* Document.swift in Sources */,
DED059702AC136FF00122A60 /* Extension.swift in Sources */,
DEA657172ABF19730066A81D /* ContentView.swift in Sources */,
DEA657152ABF19730066A81D /* SherpaOnnxSubtitleApp.swift in Sources */,
DE081AB32ABFFF2600E8CD63 /* Errors.swift in Sources */,
DEA657232ABF20130066A81D /* Audio.swift in Sources */,
DE081A8F2ABF287C00E8CD63 /* SherpaOnnx.swift in Sources */,
DE081A952ABFC60E00E8CD63 /* Model.swift in Sources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXSourcesBuildPhase section */
/* Begin XCBuildConfiguration section */
DEA6571D2ABF19740066A81D /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_SEARCH_USER_PATHS = NO;
ASSETCATALOG_COMPILER_GENERATE_SWIFT_ASSET_SYMBOL_EXTENSIONS = YES;
CLANG_ANALYZER_NONNULL = YES;
CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
CLANG_CXX_LANGUAGE_STANDARD = "gnu++20";
CLANG_ENABLE_MODULES = YES;
CLANG_ENABLE_OBJC_ARC = YES;
CLANG_ENABLE_OBJC_WEAK = YES;
CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
CLANG_WARN_BOOL_CONVERSION = YES;
CLANG_WARN_COMMA = YES;
CLANG_WARN_CONSTANT_CONVERSION = YES;
CLANG_WARN_DEPRECATED_OBJC_IMPLEMENTATIONS = YES;
CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
CLANG_WARN_EMPTY_BODY = YES;
CLANG_WARN_ENUM_CONVERSION = YES;
CLANG_WARN_INFINITE_RECURSION = YES;
CLANG_WARN_INT_CONVERSION = YES;
CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
CLANG_WARN_OBJC_IMPLICIT_RETAIN_SELF = YES;
CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
CLANG_WARN_QUOTED_INCLUDE_IN_FRAMEWORK_HEADER = YES;
CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
CLANG_WARN_STRICT_PROTOTYPES = YES;
CLANG_WARN_SUSPICIOUS_MOVE = YES;
CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
CLANG_WARN_UNREACHABLE_CODE = YES;
CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
COPY_PHASE_STRIP = NO;
DEBUG_INFORMATION_FORMAT = dwarf;
ENABLE_STRICT_OBJC_MSGSEND = YES;
ENABLE_TESTABILITY = YES;
ENABLE_USER_SCRIPT_SANDBOXING = YES;
GCC_C_LANGUAGE_STANDARD = gnu17;
GCC_DYNAMIC_NO_PIC = NO;
GCC_NO_COMMON_BLOCKS = YES;
GCC_OPTIMIZATION_LEVEL = 0;
GCC_PREPROCESSOR_DEFINITIONS = (
"DEBUG=1",
"$(inherited)",
);
GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
GCC_WARN_UNDECLARED_SELECTOR = YES;
GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
GCC_WARN_UNUSED_FUNCTION = YES;
GCC_WARN_UNUSED_VARIABLE = YES;
IPHONEOS_DEPLOYMENT_TARGET = 16.0;
LOCALIZATION_PREFERS_STRING_CATALOGS = YES;
MTL_ENABLE_DEBUG_INFO = INCLUDE_SOURCE;
MTL_FAST_MATH = YES;
ONLY_ACTIVE_ARCH = YES;
SDKROOT = iphoneos;
SWIFT_ACTIVE_COMPILATION_CONDITIONS = "DEBUG $(inherited)";
SWIFT_OPTIMIZATION_LEVEL = "-Onone";
};
name = Debug;
};
DEA6571E2ABF19740066A81D /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_SEARCH_USER_PATHS = NO;
ASSETCATALOG_COMPILER_GENERATE_SWIFT_ASSET_SYMBOL_EXTENSIONS = YES;
CLANG_ANALYZER_NONNULL = YES;
CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
CLANG_CXX_LANGUAGE_STANDARD = "gnu++20";
CLANG_ENABLE_MODULES = YES;
CLANG_ENABLE_OBJC_ARC = YES;
CLANG_ENABLE_OBJC_WEAK = YES;
CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
CLANG_WARN_BOOL_CONVERSION = YES;
CLANG_WARN_COMMA = YES;
CLANG_WARN_CONSTANT_CONVERSION = YES;
CLANG_WARN_DEPRECATED_OBJC_IMPLEMENTATIONS = YES;
CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
CLANG_WARN_EMPTY_BODY = YES;
CLANG_WARN_ENUM_CONVERSION = YES;
CLANG_WARN_INFINITE_RECURSION = YES;
CLANG_WARN_INT_CONVERSION = YES;
CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
CLANG_WARN_OBJC_IMPLICIT_RETAIN_SELF = YES;
CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
CLANG_WARN_QUOTED_INCLUDE_IN_FRAMEWORK_HEADER = YES;
CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
CLANG_WARN_STRICT_PROTOTYPES = YES;
CLANG_WARN_SUSPICIOUS_MOVE = YES;
CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
CLANG_WARN_UNREACHABLE_CODE = YES;
CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
COPY_PHASE_STRIP = NO;
DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
ENABLE_NS_ASSERTIONS = NO;
ENABLE_STRICT_OBJC_MSGSEND = YES;
ENABLE_USER_SCRIPT_SANDBOXING = YES;
GCC_C_LANGUAGE_STANDARD = gnu17;
GCC_NO_COMMON_BLOCKS = YES;
GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
GCC_WARN_UNDECLARED_SELECTOR = YES;
GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
GCC_WARN_UNUSED_FUNCTION = YES;
GCC_WARN_UNUSED_VARIABLE = YES;
IPHONEOS_DEPLOYMENT_TARGET = 16.0;
LOCALIZATION_PREFERS_STRING_CATALOGS = YES;
MTL_ENABLE_DEBUG_INFO = NO;
MTL_FAST_MATH = YES;
SDKROOT = iphoneos;
SWIFT_COMPILATION_MODE = wholemodule;
VALIDATE_PRODUCT = YES;
};
name = Release;
};
DEA657202ABF19740066A81D /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
ASSETCATALOG_COMPILER_GLOBAL_ACCENT_COLOR_NAME = AccentColor;
CODE_SIGN_STYLE = Automatic;
CURRENT_PROJECT_VERSION = 1;
DEVELOPMENT_ASSET_PATHS = "\"SherpaOnnxSubtitle/Preview Content\"";
DEVELOPMENT_TEAM = 896WS4KUPV;
ENABLE_PREVIEWS = YES;
GENERATE_INFOPLIST_FILE = YES;
HEADER_SEARCH_PATHS = "${PROJECT_DIR}/../../build-ios/sherpa-onnx.xcframework/Headers/";
INFOPLIST_FILE = SherpaOnnxSubtitle/Info.plist;
INFOPLIST_KEY_UIApplicationSceneManifest_Generation = YES;
INFOPLIST_KEY_UIApplicationSupportsIndirectInputEvents = YES;
INFOPLIST_KEY_UILaunchScreen_Generation = YES;
INFOPLIST_KEY_UISupportedInterfaceOrientations_iPad = "UIInterfaceOrientationPortrait UIInterfaceOrientationPortraitUpsideDown UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight";
INFOPLIST_KEY_UISupportedInterfaceOrientations_iPhone = "UIInterfaceOrientationPortrait UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight";
IPHONEOS_DEPLOYMENT_TARGET = 16.0;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
);
MARKETING_VERSION = 1.0;
OTHER_LDFLAGS = "-lc++";
PRODUCT_BUNDLE_IDENTIFIER = net.duoziwei.SherpaOnnxSubtitle;
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_EMIT_LOC_STRINGS = YES;
SWIFT_OBJC_BRIDGING_HEADER = "${PROJECT_DIR}/../../swift-api-examples/SherpaOnnx-Bridging-Header.h";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
};
name = Debug;
};
DEA657212ABF19740066A81D /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
ASSETCATALOG_COMPILER_GLOBAL_ACCENT_COLOR_NAME = AccentColor;
CODE_SIGN_STYLE = Automatic;
CURRENT_PROJECT_VERSION = 1;
DEVELOPMENT_ASSET_PATHS = "\"SherpaOnnxSubtitle/Preview Content\"";
DEVELOPMENT_TEAM = 896WS4KUPV;
ENABLE_PREVIEWS = YES;
GENERATE_INFOPLIST_FILE = YES;
HEADER_SEARCH_PATHS = "${PROJECT_DIR}/../../build-ios/sherpa-onnx.xcframework/Headers/";
INFOPLIST_FILE = SherpaOnnxSubtitle/Info.plist;
INFOPLIST_KEY_UIApplicationSceneManifest_Generation = YES;
INFOPLIST_KEY_UIApplicationSupportsIndirectInputEvents = YES;
INFOPLIST_KEY_UILaunchScreen_Generation = YES;
INFOPLIST_KEY_UISupportedInterfaceOrientations_iPad = "UIInterfaceOrientationPortrait UIInterfaceOrientationPortraitUpsideDown UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight";
INFOPLIST_KEY_UISupportedInterfaceOrientations_iPhone = "UIInterfaceOrientationPortrait UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight";
IPHONEOS_DEPLOYMENT_TARGET = 16.0;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
);
MARKETING_VERSION = 1.0;
OTHER_LDFLAGS = "-lc++";
PRODUCT_BUNDLE_IDENTIFIER = net.duoziwei.SherpaOnnxSubtitle;
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_EMIT_LOC_STRINGS = YES;
SWIFT_OBJC_BRIDGING_HEADER = "${PROJECT_DIR}/../../swift-api-examples/SherpaOnnx-Bridging-Header.h";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
};
name = Release;
};
/* End XCBuildConfiguration section */
/* Begin XCConfigurationList section */
DEA6570C2ABF19730066A81D /* Build configuration list for PBXProject "SherpaOnnxSubtitle" */ = {
isa = XCConfigurationList;
buildConfigurations = (
DEA6571D2ABF19740066A81D /* Debug */,
DEA6571E2ABF19740066A81D /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
DEA6571F2ABF19740066A81D /* Build configuration list for PBXNativeTarget "SherpaOnnxSubtitle" */ = {
isa = XCConfigurationList;
buildConfigurations = (
DEA657202ABF19740066A81D /* Debug */,
DEA657212ABF19740066A81D /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
/* End XCConfigurationList section */
};
rootObject = DEA657092ABF19730066A81D /* Project object */;
}
... ...
<?xml version="1.0" encoding="UTF-8"?>
<Workspace
version = "1.0">
<FileRef
location = "self:">
</FileRef>
</Workspace>
... ...
<?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>IDEDidComputeMac32BitWarning</key>
<true/>
</dict>
</plist>
... ...
{
"colors" : [
{
"idiom" : "universal"
}
],
"info" : {
"author" : "xcode",
"version" : 1
}
}
... ...
{
"images" : [
{
"filename" : "k2-1024x1024.png",
"idiom" : "universal",
"platform" : "ios",
"size" : "1024x1024"
}
],
"info" : {
"author" : "xcode",
"version" : 1
}
}
... ...
{
"info" : {
"author" : "xcode",
"version" : 1
}
}
... ...
//
// ContentView.swift
// SherpaOnnxSubtitle
//
// Created by knight on 2023/9/23.
//
import AVKit
import MediaPlayer
import PhotosUI
import SwiftUI
struct ContentView: View {
@StateObject var subtitleViewModel = SubtitleViewModel()
var body: some View {
VStack {
VStack {
Text("SherpaOnnxSubtitle")
.font(.title)
VStack(alignment: .leading) {
Text("Audio format should be **mono** channel and **16khz** sample rate")
Text("You can convert file with the help of ffmpeg")
Text("```ffmpeg -i ./foo.mov -acodec pcm_s16le -ac 1 -ar 16000 foo.wav```")
}
}
.padding(.vertical)
PhotosPicker(
selection: $subtitleViewModel.selectedItem,
matching: .videos
) {
Label("Open Audio from Photo Library", systemImage: "photo")
.frame(minWidth: 0, maxWidth: .infinity)
.padding()
.background(.blue, in: .rect(cornerRadius: 8.0))
.foregroundColor(.white)
}
Button(action: {
subtitleViewModel.importNow = true
}, label: {
Text("Open Audio from Files")
.frame(minWidth: 0, maxWidth: .infinity)
.padding()
.background(.blue, in: .rect(cornerRadius: 8.0))
})
.foregroundColor(.white)
switch subtitleViewModel.loadState {
case .initial, .loaded(_), .done:
EmptyView()
case .loading:
ProgressView()
case .failed:
Text("Gen SRT failed")
}
}
.fileImporter(isPresented: $subtitleViewModel.importNow, allowedContentTypes: [.movie, .audio], onCompletion: handleImportCompletion)
.onChange(of: subtitleViewModel.importNow) { importNow in
if !importNow {
subtitleViewModel.restoreState()
}
}
.fileExporter(isPresented: $subtitleViewModel.exportNow,
document: subtitleViewModel.srtDocument, contentType: .srt,
defaultFilename: subtitleViewModel.srtName,
onCompletion: handleExportCompletion)
.task(id: subtitleViewModel.selectedItem) {
do {
if !subtitleViewModel.hasAudio {
return
}
subtitleViewModel.loadState = .loading
if let movie = try await subtitleViewModel.selectedItem?.loadTransferable(type: Audio.self) {
subtitleViewModel.loadState = .loaded(movie)
subtitleViewModel.generateSRT(from: movie.url)
} else {
subtitleViewModel.loadState = .failed
}
} catch {
subtitleViewModel.loadState = .failed
}
}
.padding()
}
private func handleImportCompletion(result: Result<URL, Error>) {
print("file import...")
switch result {
case let .success(file):
let accessing = file.startAccessingSecurityScopedResource()
defer {
if accessing {
file.stopAccessingSecurityScopedResource()
}
}
subtitleViewModel.generateSRT(from: file)
case let .failure(error):
print(error.localizedDescription)
subtitleViewModel.loadState = .failed
}
}
private func handleExportCompletion(result: Result<URL, any Error>) {
switch result {
case let .success(url):
print("audio export to: \(url)")
subtitleViewModel.loadState = .done
case let .failure(error):
print("export audio error: \(error.localizedDescription)")
subtitleViewModel.loadState = .failed
}
}
}
struct ContentView_Previews: PreviewProvider {
static var previews: some View {
ContentView()
}
}
... ...
//
// UTType.swift
// YPlayer
//
// Created by knight on 2023/7/7.
//
import UniformTypeIdentifiers
extension UTType {
static var srt: UTType {
UTType(exportedAs: "com.k2.srt")
}
}
... ...
<?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>UTExportedTypeDeclarations</key>
<array>
<dict>
<key>UTTypeConformsTo</key>
<array>
<string>public.plain-text</string>
</array>
<key>UTTypeDescription</key>
<string>SubRip Subtitle File</string>
<key>UTTypeIconFiles</key>
<array/>
<key>UTTypeIdentifier</key>
<string>com.k2.srt</string>
<key>UTTypeTagSpecification</key>
<dict>
<key>public.filename-extension</key>
<array>
<string>srt</string>
</array>
</dict>
</dict>
</array>
</dict>
</plist>
... ...
//
// Audio.swift
// SherpaOnnxSubtitle
//
// Created by knight on 2023/9/23.
//
import SwiftUI
struct Audio: Transferable {
let url: URL
static var transferRepresentation: some TransferRepresentation {
FileRepresentation(contentType: .movie) { movie in
SentTransferredFile(movie.url)
} importing: { received in
let copy = URL.documentsDirectory.appending(path: "audio.wav")
if FileManager.default.fileExists(atPath: copy.path()) {
try FileManager.default.removeItem(at: copy)
}
try FileManager.default.copyItem(at: received.file, to: copy)
return Self(url: copy)
}
}
}
... ...
//
// Document.swift
// YPlayer
//
// Created by knight on 2023/6/5.
//
import SwiftUI
import UniformTypeIdentifiers
struct Document: FileDocument {
static var readableContentTypes = [UTType.srt]
static var writableContentTypes = [UTType.srt]
var data: Data?
init(data: Data?) {
self.data = data
}
init(configuration: ReadConfiguration) throws {
if let data = configuration.file.regularFileContents {
self.data = data
}
}
func fileWrapper(configuration _: WriteConfiguration) throws -> FileWrapper {
guard let data = data else {
throw ExportError.fileNotFound
}
return FileWrapper(regularFileWithContents: data)
}
}
... ...
//
// Errors.swift
// YPlayer
//
// Created by knight on 2023/8/26.
//
import Foundation
enum ExportError: String, Error {
case fileNotFound = "export file not found"
}
... ...
//
// SpeechSegment.swift
// SherpaOnnxSubtitle
//
// Created by knight on 2023/9/23.
//
import Foundation
class SpeechSegment: CustomStringConvertible {
let start: Float
let end: Float
let text: String
init(start: Float, duration: Float, text: String) {
self.start = start
end = start + duration
self.text = text
}
public var description: String {
var s: String
s = TimeInterval(start).hourMinuteSecondMS
s += " --> "
s += TimeInterval(end).hourMinuteSecondMS
s += "\n"
s += text
return s
}
}
... ...
//
// SherpaOnnxSubtitleApp.swift
// SherpaOnnxSubtitle
//
// Created by knight on 2023/9/23.
//
import SwiftUI
@main
struct SherpaOnnxSubtitleApp: App {
var body: some Scene {
WindowGroup {
ContentView()
}
}
}
... ...
//
// SubtitleViewModel.swift
// SherpaOnnxSubtitle
//
// Created by knight on 2023/9/23.
//
import AVFoundation
import PhotosUI
import SwiftUI
enum LoadState {
case initial
case loading
case loaded(Audio)
case done
case failed
}
class SubtitleViewModel: ObservableObject {
var modelType = "whisper"
let sampleRate = 16000
var modelConfig: SherpaOnnxOfflineModelConfig?
// modelType = "paraformer"
var recognizer: SherpaOnnxOfflineRecognizer?
var vadModelConfig: SherpaOnnxVadModelConfig?
var vad: SherpaOnnxVoiceActivityDetectorWrapper?
@Published var loadState: LoadState = .initial
@Published var selectedItem: PhotosPickerItem? = nil
@Published var importNow: Bool = false {
didSet {
loadState = .loading
}
}
@Published var exportNow: Bool = false
var srtName: String = "unknown.srt"
var content: String = ""
var srtDocument: Document {
let content = content.data(using: .utf8)
return Document(data: content)
}
var hasAudio: Bool {
return selectedItem != nil
}
init() {
if modelType == "whisper" {
// for English
self.modelConfig = getNonStreamingWhisperTinyEn()
} else if modelType == "paraformer" {
// for Chinese
self.modelConfig = getNonStreamingZhParaformer20230328()
} else {
print("Please specify a supported modelType \(modelType)")
return
}
let featConfig = sherpaOnnxFeatureConfig(
sampleRate: sampleRate,
featureDim: 80
)
guard let modelConfig else {
return
}
var config = sherpaOnnxOfflineRecognizerConfig(
featConfig: featConfig,
modelConfig: modelConfig
)
recognizer = SherpaOnnxOfflineRecognizer(config: &config)
let sileroVadConfig = sherpaOnnxSileroVadModelConfig(
model: getResource("silero_vad", "onnx")
)
self.vadModelConfig = sherpaOnnxVadModelConfig(sileroVad: sileroVadConfig)
guard var vadModelConfig else {
return
}
vad = SherpaOnnxVoiceActivityDetectorWrapper(
config: &vadModelConfig, buffer_size_in_seconds: 120
)
}
func restoreState() {
loadState = .initial
}
func generateSRT(from file: URL) {
print("gen srt from: \(file)")
content = ""
// restore state
defer {
loadState = .done
}
guard let recognizer else {
return
}
guard let vadModelConfig else {
return
}
guard let vad else {
return
}
do {
let audioFile = try AVAudioFile(forReading: file)
let audioFormat = audioFile.processingFormat
assert(audioFormat.sampleRate == Double(sampleRate))
assert(audioFormat.channelCount == 1)
assert(audioFormat.commonFormat == AVAudioCommonFormat.pcmFormatFloat32)
let audioFrameCount = UInt32(audioFile.length)
let audioFileBuffer = AVAudioPCMBuffer(pcmFormat: audioFormat, frameCapacity: audioFrameCount)
try audioFile.read(into: audioFileBuffer!)
var array: [Float]! = audioFileBuffer?.array()
let windowSize = Int(vadModelConfig.silero_vad.window_size)
var segments: [SpeechSegment] = []
while array.count > windowSize {
// todo(fangjun): avoid extra copies here
vad.acceptWaveform(samples: [Float](array[0 ..< windowSize]))
array = [Float](array[windowSize ..< array.count])
while !vad.isEmpty() {
let s = vad.front()
vad.pop()
let result = recognizer.decode(samples: s.samples)
segments.append(
SpeechSegment(
start: Float(s.start) / Float(sampleRate),
duration: Float(s.samples.count) / Float(sampleRate),
text: result.text
))
print(segments.last!)
}
}
content = zip(segments.indices, segments).map { index, element in
"\(index + 1)\n\(element)"
}.joined(separator: "\n\n")
} catch {
print("error: \(error.localizedDescription)")
}
exportNow = true
let last = file.lastPathComponent
srtName = "\(last).srt"
}
}
... ...