Add cloud translation support with multiple providers and configurati… (

#183) * Add cloud translation support with multiple providers and configuration options * Refactor CMakeLists.txt for cloud translation sources formatting * Add support for translating only full sentences in cloud translation * Update ICU build configuration and fix header include case sensitivity * Fix CURL helper function signatures and improve URL encoding * Fix character type casting in DeepLTranslator for language conversion * Refactor file saving logic in transcription filter to streamline sentence handling and add support for saving translated sentences * Add support for Deepl Free API endpoint and enhance cloud translation configuration * Add ccache detection to ICU build configuration for improved compilation speed * Enhance ICU build configuration to use ccache as a compiler wrapper for improved performance
locaal-ai · Nov 25, 2024 · 04a6f6a · 04a6f6a
1 parent b7ab6a9
commit 04a6f6a
Show file tree

Hide file tree

Showing 28 changed files with 1,798 additions and 73 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -127,6 +127,8 @@ target_sources(
           src/translation/translation-language-utils.cpp
           src/ui/filter-replace-dialog.cpp)
 
+add_subdirectory(src/translation/cloud-translation)
+
 set_target_properties_plugin(${CMAKE_PROJECT_NAME} PROPERTIES OUTPUT_NAME ${_name})
 
 if(ENABLE_TESTS)

diff --git a/cmake/BuildICU.cmake b/cmake/BuildICU.cmake
@@ -48,26 +48,38 @@ if(WIN32)
                                                                                        "${ICU_LIB_${lib}}")
   endforeach()
 else()
+  # Add ccache detection at the start
+  find_program(CCACHE_PROGRAM ccache)
+  if(CCACHE_PROGRAM)
+    message(STATUS "Found ccache: ${CCACHE_PROGRAM}")
+    # Create compiler wrapper commands
+    set(C_LAUNCHER "${CCACHE_PROGRAM} ${CMAKE_C_COMPILER}")
+    set(CXX_LAUNCHER "${CCACHE_PROGRAM} ${CMAKE_CXX_COMPILER}")
+  endif()
+
   set(ICU_URL
       "https://github.com/unicode-org/icu/releases/download/release-${ICU_VERSION_DASH}/icu4c-${ICU_VERSION_UNDERSCORE}-src.tgz"
   )
   set(ICU_HASH "SHA256=cb968df3e4d2e87e8b11c49a5d01c787bd13b9545280fc6642f826527618caef")
   if(APPLE)
     set(ICU_PLATFORM "MacOSX")
     set(TARGET_ARCH -arch\ $ENV{MACOS_ARCH})
-    set(ICU_BUILD_ENV_VARS CFLAGS=${TARGET_ARCH} CXXFLAGS=${TARGET_ARCH} LDFLAGS=${TARGET_ARCH})
+    set(ICU_BUILD_ENV_VARS CFLAGS=${TARGET_ARCH} CXXFLAGS=${TARGET_ARCH} LDFLAGS=${TARGET_ARCH} CC=${C_LAUNCHER}
+                           CXX=${CXX_LAUNCHER})
   else()
     set(ICU_PLATFORM "Linux")
-    set(ICU_BUILD_ENV_VARS CFLAGS=-fPIC CXXFLAGS=-fPIC LDFLAGS=-fPIC)
+    set(ICU_BUILD_ENV_VARS CFLAGS=-fPIC CXXFLAGS=-fPIC LDFLAGS=-fPIC CC=${C_LAUNCHER} CXX=${CXX_LAUNCHER})
   endif()
 
   ExternalProject_Add(
     ICU_build
     DOWNLOAD_EXTRACT_TIMESTAMP true
     GIT_REPOSITORY "https://github.com/unicode-org/icu.git"
     GIT_TAG "release-${ICU_VERSION_DASH}"
-    CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${ICU_BUILD_ENV_VARS} <SOURCE_DIR>/icu4c/source/runConfigureICU
-                      ${ICU_PLATFORM} --prefix=<INSTALL_DIR> --enable-static --disable-shared
+    CONFIGURE_COMMAND
+      ${CMAKE_COMMAND} -E env ${ICU_BUILD_ENV_VARS} <SOURCE_DIR>/icu4c/source/runConfigureICU ${ICU_PLATFORM}
+      --prefix=<INSTALL_DIR> --enable-static --disable-shared --disable-tools --disable-samples --disable-layout
+      --disable-layoutex --disable-tests --disable-draft --disable-extras --disable-icuio
     BUILD_COMMAND make -j4
     BUILD_BYPRODUCTS
       <INSTALL_DIR>/lib/${CMAKE_STATIC_LIBRARY_PREFIX}icudata${CMAKE_STATIC_LIBRARY_SUFFIX}

diff --git a/data/locale/en-US.ini b/data/locale/en-US.ini
@@ -16,6 +16,8 @@ whisper_sampling_method="Whisper Sampling Method"
 n_threads="Number of threads"
 n_max_text_ctx="Max text context"
 translate="Translate"
+translate_local="Local Translation"
+translate_cloud="Cloud Translation"
 no_context="No context"
 single_segment="Single segment"
 print_special="Print special"
@@ -75,6 +77,11 @@ general_group="General"
 transcription_group="Transcription"
 file_output_group="File Output Configuration"
 translate_explaination="Enabling translation will increase the processing load on your machine, This feature uses additional resources to translate content in real-time, which may impact performance. <a href='#'>Learn More</a>"
+translate_cloud_explaination="Cloud translation requires an active internet connection and API keys to the translation provider."
+translate_cloud_provider="Translation Provider"
+translate_cloud_only_full_sentences="Translate only full sentences"
+translate_cloud_api_key="Access Key"
+translate_cloud_secret_key="Secret Key"
 log_group="Logging"
 advanced_group="Advanced Configuration"
 buffered_output_parameters="Buffered Output Configuration"
@@ -89,4 +96,20 @@ translate_only_full_sentences="Translate only full sentences"
 duration_filter_threshold="Duration filter"
 segment_duration="Segment duration"
 n_context_sentences="# Context sentences"
-max_sub_duration="Max. sub duration (ms)"
+max_sub_duration="Max. sub duration (ms)"
+Google-Cloud-Translation="Google Cloud Translation"
+Microsoft-Translator="Microsoft Azure Translator"
+Amazon-Translate="AWS Translate"
+IBM-Watson-Translate="IBM Watson Translate"
+Yandex-Translate="Yandex Translate"
+Baidu-Translate="Baidu Translate"
+Tencent-Translate="Tencent Translate"
+Alibaba-Translate="Alibaba Translate"
+Naver-Translate="Naver Translate"
+Kakao-Translate="Kakao Translate"
+Papago-Translate="Papago"
+Deepl-Translate="Deepl"
+Bing-Translate="Bing Translate"
+OpenAI-Translate="OpenAI"
+Claude-Translate="Claude"
+translate_cloud_deepl_free="Use Deepl Free API Endpoint"
diff --git a/src/transcription-filter-callbacks.cpp b/src/transcription-filter-callbacks.cpp
@@ -21,6 +21,7 @@
 #include "whisper-utils/whisper-utils.h"
 #include "whisper-utils/whisper-model-utils.h"
 #include "translation/language_codes.h"
+#include "translation/cloud-translation/translation-cloud.h"
 
 void send_caption_to_source(const std::string &target_source_name, const std::string &caption,
 			    struct transcription_filter_data *gf)
@@ -80,30 +81,60 @@ std::string send_sentence_to_translation(const std::string &sentence,
 	return "";
 }
 
+void send_sentence_to_cloud_translation_async(const std::string &sentence,
+					      struct transcription_filter_data *gf,
+					      const std::string &source_language,
+					      std::function<void(const std::string &)> callback)
+{
+	std::thread([sentence, gf, source_language, callback]() {
+		const std::string last_text = gf->last_text_for_cloud_translation;
+		gf->last_text_for_cloud_translation = sentence;
+		if (gf->translate_cloud && !sentence.empty()) {
+			obs_log(gf->log_level, "Translating text with cloud provider %s. %s -> %s",
+				gf->translate_cloud_provider.c_str(), source_language.c_str(),
+				gf->translate_cloud_target_language.c_str());
+			std::string translated_text;
+			if (sentence == last_text) {
+				// do not translate the same sentence twice
+				callback(gf->last_text_cloud_translation);
+				return;
+			}
+			CloudTranslatorConfig config;
+			config.provider = gf->translate_cloud_provider;
+			config.access_key = gf->translate_cloud_api_key;
+			config.secret_key = gf->translate_cloud_secret_key;
+			config.free = gf->translate_cloud_deepl_free;
+			config.region = gf->translate_cloud_region;
+
+			translated_text = translate_cloud(config, sentence,
+							  gf->translate_cloud_target_language,
+							  source_language);
+			if (!translated_text.empty()) {
+				if (gf->log_words) {
+					obs_log(LOG_INFO, "Cloud Translation: '%s' -> '%s'",
+						sentence.c_str(), translated_text.c_str());
+				}
+				gf->last_text_translation = translated_text;
+				callback(translated_text);
+				return;
+			} else {
+				obs_log(gf->log_level, "Failed to translate text");
+			}
+		}
+		callback("");
+	}).detach();
+}
+
 void send_sentence_to_file(struct transcription_filter_data *gf,
-			   const DetectionResultWithText &result, const std::string &str_copy,
-			   const std::string &translated_sentence)
+			   const DetectionResultWithText &result, const std::string &sentence,
+			   const std::string &file_path, bool bump_sentence_number)
 {
 	// Check if we should save the sentence
 	if (gf->save_only_while_recording && !obs_frontend_recording_active()) {
 		// We are not recording, do not save the sentence to file
 		return;
 	}
 
-	std::string translated_file_path = "";
-	bool write_translations = gf->translate && !translated_sentence.empty();
-
-	// if translation is enabled, save the translated sentence to another file
-	if (write_translations) {
-		// add a postfix to the file name (without extension) with the translation target language
-		std::string output_file_path = gf->output_file_path;
-		std::string file_extension =
-			output_file_path.substr(output_file_path.find_last_of(".") + 1);
-		std::string file_name =
-			output_file_path.substr(0, output_file_path.find_last_of("."));
-		translated_file_path = file_name + "_" + gf->target_lang + "." + file_extension;
-	}
-
 	// should the file be truncated?
 	std::ios_base::openmode openmode = std::ios::out;
 	if (gf->truncate_output_file) {
@@ -114,15 +145,9 @@ void send_sentence_to_file(struct transcription_filter_data *gf,
 	if (!gf->save_srt) {
 		// Write raw sentence to file
 		try {
-			std::ofstream output_file(gf->output_file_path, openmode);
-			output_file << str_copy << std::endl;
+			std::ofstream output_file(file_path, openmode);
+			output_file << sentence << std::endl;
 			output_file.close();
-			if (write_translations) {
-				std::ofstream translated_output_file(translated_file_path,
-								     openmode);
-				translated_output_file << translated_sentence << std::endl;
-				translated_output_file.close();
-			}
 		} catch (const std::ofstream::failure &e) {
 			obs_log(LOG_ERROR, "Exception opening/writing/closing file: %s", e.what());
 		}
@@ -133,9 +158,9 @@ void send_sentence_to_file(struct transcription_filter_data *gf,
 		}
 
 		obs_log(gf->log_level, "Saving sentence to file %s, sentence #%d",
-			gf->output_file_path.c_str(), gf->sentence_number);
+			file_path.c_str(), gf->sentence_number);
 		// Append sentence to file in .srt format
-		std::ofstream output_file(gf->output_file_path, openmode);
+		std::ofstream output_file(file_path, openmode);
 		output_file << gf->sentence_number << std::endl;
 		// use the start and end timestamps to calculate the start and end time in srt format
 		auto format_ts_for_srt = [](std::ofstream &output_stream, uint64_t ts) {
@@ -156,28 +181,34 @@ void send_sentence_to_file(struct transcription_filter_data *gf,
 		format_ts_for_srt(output_file, result.end_timestamp_ms);
 		output_file << std::endl;
 
-		output_file << str_copy << std::endl;
+		output_file << sentence << std::endl;
 		output_file << std::endl;
 		output_file.close();
 
-		if (write_translations) {
-			obs_log(gf->log_level, "Saving translation to file %s, sentence #%d",
-				translated_file_path.c_str(), gf->sentence_number);
-
-			// Append translated sentence to file in .srt format
-			std::ofstream translated_output_file(translated_file_path, openmode);
-			translated_output_file << gf->sentence_number << std::endl;
-			format_ts_for_srt(translated_output_file, result.start_timestamp_ms);
-			translated_output_file << " --> ";
-			format_ts_for_srt(translated_output_file, result.end_timestamp_ms);
-			translated_output_file << std::endl;
-
-			translated_output_file << translated_sentence << std::endl;
-			translated_output_file << std::endl;
-			translated_output_file.close();
+		if (bump_sentence_number) {
+			gf->sentence_number++;
 		}
+	}
+}
 
-		gf->sentence_number++;
+void send_translated_sentence_to_file(struct transcription_filter_data *gf,
+				      const DetectionResultWithText &result,
+				      const std::string &translated_sentence,
+				      const std::string &target_lang)
+{
+	// if translation is enabled, save the translated sentence to another file
+	if (translated_sentence.empty()) {
+		obs_log(gf->log_level, "Translation is empty, not saving to file");
+	} else {
+		// add a postfix to the file name (without extension) with the translation target language
+		std::string translated_file_path = "";
+		std::string output_file_path = gf->output_file_path;
+		std::string file_extension =
+			output_file_path.substr(output_file_path.find_last_of(".") + 1);
+		std::string file_name =
+			output_file_path.substr(0, output_file_path.find_last_of("."));
+		translated_file_path = file_name + "_" + target_lang + "." + file_extension;
+		send_sentence_to_file(gf, result, translated_sentence, translated_file_path, false);
 	}
 }
 
@@ -235,41 +266,76 @@ void set_text_callback(struct transcription_filter_data *gf,
 		}
 	}
 
-	bool should_translate =
+	bool should_translate_local =
 		gf->translate_only_full_sentences ? result.result == DETECTION_RESULT_SPEECH : true;
 
 	// send the sentence to translation (if enabled)
-	std::string translated_sentence =
-		should_translate ? send_sentence_to_translation(str_copy, gf, result.language) : "";
+	std::string translated_sentence_local =
+		should_translate_local ? send_sentence_to_translation(str_copy, gf, result.language)
+				       : "";
 
 	if (gf->translate) {
 		if (gf->translation_output == "none") {
 			// overwrite the original text with the translated text
-			str_copy = translated_sentence;
+			str_copy = translated_sentence_local;
 		} else {
 			if (gf->buffered_output) {
 				// buffered output - add the sentence to the monitor
 				gf->translation_monitor.addSentenceFromStdString(
-					translated_sentence,
+					translated_sentence_local,
 					get_time_point_from_ms(result.start_timestamp_ms),
 					get_time_point_from_ms(result.end_timestamp_ms),
 					result.result == DETECTION_RESULT_PARTIAL);
 			} else {
 				// non-buffered output - send the sentence to the selected source
-				send_caption_to_source(gf->translation_output, translated_sentence,
-						       gf);
+				send_caption_to_source(gf->translation_output,
+						       translated_sentence_local, gf);
 			}
 		}
+		if (gf->save_to_file && gf->output_file_path != "") {
+			send_translated_sentence_to_file(gf, result, translated_sentence_local,
+							 gf->target_lang);
+		}
 	}
 
-	if (gf->buffered_output) {
-		gf->captions_monitor.addSentenceFromStdString(
-			str_copy, get_time_point_from_ms(result.start_timestamp_ms),
-			get_time_point_from_ms(result.end_timestamp_ms),
-			result.result == DETECTION_RESULT_PARTIAL);
-	} else {
-		// non-buffered output - send the sentence to the selected source
-		send_caption_to_source(gf->text_source_name, str_copy, gf);
+	bool should_translate_cloud = (gf->translate_cloud_only_full_sentences
+					       ? result.result == DETECTION_RESULT_SPEECH
+					       : true) &&
+				      gf->translate_cloud;
+
+	if (should_translate_cloud) {
+		send_sentence_to_cloud_translation_async(
+			str_copy, gf, result.language,
+			[gf, result](const std::string &translated_sentence_cloud) {
+				if (gf->translate_cloud_output != "none") {
+					send_caption_to_source(gf->translate_cloud_output,
+							       translated_sentence_cloud, gf);
+				} else {
+					// overwrite the original text with the translated text
+					send_caption_to_source(gf->text_source_name,
+							       translated_sentence_cloud, gf);
+				}
+				if (gf->save_to_file && gf->output_file_path != "") {
+					send_translated_sentence_to_file(
+						gf, result, translated_sentence_cloud,
+						gf->translate_cloud_target_language);
+				}
+			});
+	}
+
+	// send the original text to the output
+	// unless the translation is enabled and set to overwrite the original text
+	if (!((should_translate_cloud && gf->translate_cloud_output == "none") ||
+	      (should_translate_local && gf->translation_output == "none"))) {
+		if (gf->buffered_output) {
+			gf->captions_monitor.addSentenceFromStdString(
+				str_copy, get_time_point_from_ms(result.start_timestamp_ms),
+				get_time_point_from_ms(result.end_timestamp_ms),
+				result.result == DETECTION_RESULT_PARTIAL);
+		} else {
+			// non-buffered output - send the sentence to the selected source
+			send_caption_to_source(gf->text_source_name, str_copy, gf);
+		}
 	}
 
 	if (gf->caption_to_stream && result.result == DETECTION_RESULT_SPEECH) {
@@ -279,7 +345,7 @@ void set_text_callback(struct transcription_filter_data *gf,
 
 	if (gf->save_to_file && gf->output_file_path != "" &&
 	    result.result == DETECTION_RESULT_SPEECH) {
-		send_sentence_to_file(gf, result, str_copy, translated_sentence);
+		send_sentence_to_file(gf, result, str_copy, gf->output_file_path, true);
 	}
 
 	if (!result.text.empty() && (result.result == DETECTION_RESULT_SPEECH ||

diff --git a/src/transcription-filter-data.h b/src/transcription-filter-data.h
@@ -89,9 +89,18 @@ struct transcription_filter_data {
 	float duration_filter_threshold = 2.25f;
 	int segment_duration = 7000;
 
-	// Last transcription result
-	std::string last_text_for_translation;
-	std::string last_text_translation;
+	// Cloud translation options
+	bool translate_cloud = false;
+	std::string translate_cloud_provider;
+	std::string translate_cloud_target_language;
+	std::string translate_cloud_output;
+	std::string translate_cloud_api_key;
+	std::string translate_cloud_secret_key;
+	bool translate_cloud_only_full_sentences = true;
+	std::string last_text_for_cloud_translation;
+	std::string last_text_cloud_translation;
+	bool translate_cloud_deepl_free;
+	std::string translate_cloud_region;
 
 	// Transcription context sentences
 	int n_context_sentences;
@@ -119,6 +128,9 @@ struct transcription_filter_data {
 	std::string translation_model_index;
 	std::string translation_model_path_external;
 	bool translate_only_full_sentences;
+	// Last transcription result
+	std::string last_text_for_translation;
+	std::string last_text_translation;
 
 	bool buffered_output = false;
 	TokenBufferThread captions_monitor;