CN115910069A - Automatic English bilingual subtitle generating system in traditional Chinese medicine video - Google Patents

Abstract

The invention discloses a system for automatically generating Chinese-English bilingual subtitles for videos of traditional Chinese medicine, which includes a front-end information acquisition module, an analysis processing module, and a back-end writing module based on the system architecture. Call the data of Chinese medicine terminology database. Compared with the existing technology, the present invention has the advantages of: reasonable system architecture design and process steps, importing the self-built corpus into the speech recognition engine for model training, and dynamically updating through the knowledge map technology, so as to continuously build and improve The model improves the accuracy of speech recognition in the field of Chinese medicine, effectively improves the efficiency and accuracy of bilingual translation of Chinese medicine videos, reduces the workload of translators, has good applicability, and is easy to promote.

Description

An automatic generation system of Chinese and English bilingual subtitles for Chinese medicine videos

technical field

The invention relates to the field of traditional Chinese medicine, in particular to a system for automatically generating Chinese-English bilingual subtitles for videos of traditional Chinese medicine.

Background technique

TCM theory comes from the summary of medical experience, and its content includes the theory of essence, qi, blood, body fluid, viscera, meridian, constitution, etiology, disease, pathogenesis, treatment principles, health preservation, etc. As early as more than 2,000 years ago, the Chinese medicine monograph "Huangdi Neijing" came out, which laid the foundation of Chinese medicine. Today, the theories, diagnostic methods, and treatment methods related to traditional Chinese medicine can be found in this book.

Contents of the invention

The technical problem to be solved by the present invention is that the current Chinese medicine videos all use Chinese as the source of commentary, and the translation process mostly relies on manual translation, resulting in low translation efficiency and low translation accuracy, and also increases the workload of translators.

In order to solve the above technical problems, the technical solution provided by the present invention is: a Chinese-English bilingual subtitle automatic generation system for Chinese medicine video, including a front-end information acquisition module, an analysis processing module and a back-end writing module based on the system architecture, and the front-end information acquisition module acquires Video data, and call the data of the traditional Chinese medicine terminology database according to the acquired video data type; the analysis processing module recognizes the voice content through the HKUST Xunfei interface, obtains the text after analyzing the result and generates corresponding subtitles, and the analysis processing module has a built-in judgment program to judge whether the subtitle results generated by the program detection need to be translated, specifically including: 1) translation is required, the system calls the Baidu machine translation machine port and the Chinese-English bilingual Chinese medicine term translation database to translate the subtitle results and generate bilingual subtitles; 2) not If translation is required, the subtitle results are directly exported; the judgment program extracts keywords from the subtitles, and the extracted data with keywords is transmitted to the back-end writing module through the channel. The back-end writing module is made through the web page and obtained through the channel The front-end information acquisition module and the instruction data sent by the analysis processing module are used for system production.

Compared with the prior art, the present invention has the advantages of reasonable system architecture design and process steps, importing the self-built corpus into the speech recognition engine for model training, and dynamically updating through the knowledge map technology, so as to continuously build and improve the The model improves the accuracy of speech recognition in the field of TCM, effectively improves the efficiency and accuracy of bilingual translation of TCM videos, reduces the workload of translators, has good applicability, and is easy to promote.

Further, the audio acquisition method of the front-end information acquisition module includes voice signal extraction and sound feature parameter extraction.

Furthermore, the analysis function of the analysis processing module for the recognized speech data is based on Ali's open source library FastJson.

Further, the parsing and processing module generates subtitle results based on the TextView.

Further, the functions of the back-end writing module include: 1) search using knowledge graph technology; 2) modify subtitles; 3) vocal extraction; 4) audio transcription; 5) voice broadcast.

Description of drawings

Figure 1 is a schematic diagram of the workflow of a system for automatically generating Chinese-English bilingual subtitles for Chinese medicine videos.

Fig. 2 is a schematic diagram of Embodiment 1.

Fig. 3 is a schematic diagram of the second embodiment.

Fig. 4 is a schematic diagram of the third embodiment.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings.

When the present invention is implemented, in the embodiment shown in Figure 1, the front-end information acquisition module acquires video data, and transfers the data of the traditional Chinese medicine terminology database according to the video data type acquired; Identify the content, get the text after parsing the result and generate corresponding subtitles. The parsing processing module has a built-in judgment program to judge whether the subtitle results generated by the program detection need to be translated, specifically including: 1) Translation is required, and the system calls the Baidu machine translation machine port Translate the subtitle results with the Chinese-English bilingual TCM terminology translation database and generate bilingual subtitles; 2) Directly export the subtitle results without translation; the judgment program extracts keywords from the subtitles, and the extracted data with keywords passes through The channel is transmitted to the back-end writing module, and the back-end writing module is made through the web page, and the command data sent by the front-end information acquisition module and the analysis processing module are obtained through the channel to make the system. Each imported video can generate a small knowledge graph corresponding to the video.

Further, the audio acquisition method of the front-end information acquisition module includes voice signal extraction and sound feature parameter extraction. The parsing and processing module analyzes the recognized voice data based on Ali's open source library FastJson. The parsing and processing module generates subtitle results based on TextView. The functions of the back-end writing module include: 1) search by knowledge map technology; 2) modify subtitles; 3) vocal extraction; 4) audio transcription; 5) voice broadcast.

In one embodiment of the present invention, as shown in Figure 2, what this embodiment provides is the various technologies carried by the system, specifically including: 1) speech recognition technology, speech recognition technology includes acoustic feature parameter extraction (Mel Frequency cepstral coefficient, acoustic model training, speech model training, linear predictive analysis), real-time speech transcription technology (handshake stage, real-time communication stage), speech signal acquisition (A/D conversion, sampling, quantization, encoding);

2) System development technology (webpage production, background development);

3) Caption generation (TextView, Observab monitor speech recognition results);

4) TCM terminology database (compared with the national TCM terminology standard, Chinese Pharmacopoeia TCM Chinese Medicine Network);

5) Knowledge map (analyze TCM keywords, construct knowledge combing map);

6) Bilingual translation technology (Alibaba open source library FastJSon, Baidu machine translation engine, -multilingual translation term base (Chinese medicine term translation comparison standard, traditional Chinese medicine term translation comparison dictionary), ViewModel).

7) Use neo4j technology to retrieve the relationship between keywords and terms in the database.

In one embodiment of the present invention, as shown in Figure 3, this embodiment provides the working principle of the system. The Chinese medicine terminology database is obtained through speech recognition and analyzed to obtain keywords, and the knowledge map technology is used for retrieval. The final result improves the knowledge graph and dynamically expands the TCM terminology database, and the TCM terminology database tests the rationality of the knowledge graph through the accuracy of term recognition.

In one embodiment of the present invention, as shown in Figure 4, this embodiment presents the system's cyclic process, the system extracts the audio and invokes the TCM terminology database, and uses the HKUST Xunfei interface and the Ali open source library FastJSon for speech Recognition, the result after recognition is processed through the logic algorithm of "whether to convert into English", if yes, call the Chinese-English bilingual TCM terminology database and translation interface and finally generate subtitles, otherwise directly generate subtitles (Chinese), and the system will analyze the subtitles The keywords are extracted, and the extracted keywords are used for knowledge map retrieval. In the process of improving the knowledge map, many TCM terms will be generated, which will be used to improve the TCM terminology corpus. In terms of detailed design, the self-training platform combined with the TCM terminology database training model is used for speech recognition.

The above shows and describes the basic principles and main features of the present invention and the advantages of the invention. Those skilled in the art should understand that the present invention is not limited by the above-mentioned embodiments. What are described in the above-mentioned embodiments and description are only to illustrate the principles of the present invention , On the premise of not departing from the spirit and scope of the present invention, there will be various changes and improvements in the present invention, and these changes and improvements all fall within the scope of the claimed invention. The protection scope of the present invention is defined by the appended claims and their equivalents.

Claims (5)

1. The utility model provides an automatic generation system of english bilingual subtitle in traditional chinese medical science video, includes that the front end information based on system architecture obtains module, analysis processing module and back end and compiles the module, its characterized in that: the front-end information acquisition module acquires video data and calls data of a traditional Chinese medicine term library according to the type of the acquired video data; the analysis processing module identifies voice content through a science news flying interface, obtains a text and generates a corresponding subtitle after an analysis result, a judgment program is arranged in the analysis processing module, and the judgment program detects whether the generated subtitle result needs to be translated or not, and the method specifically comprises the following steps: 1) The system calls a Baidu machine translation machine port and a Chinese-English bilingual Chinese medicine term translation library to translate the subtitle result and generate a bilingual subtitle; 2) The caption result is directly exported without translation; the judgment program extracts keywords from the subtitles, the extracted data with the keywords are transmitted to the back-end compiling module through a channel, the back-end compiling module is manufactured through a webpage, and instruction data sent by the front-end information acquisition module and the analysis processing module are acquired through the channel to carry out system manufacturing.

2. The system for automatically generating English bilingual subtitles in traditional Chinese medicine videos according to claim 1, wherein: the audio acquisition method of the front-end information acquisition module comprises voice signal extraction and sound characteristic parameter extraction.

3. The system for automatically generating English bilingual subtitles in a traditional Chinese medicine video according to claim 1, wherein the system comprises: the analysis processing module analyzes the recognized voice data based on an open source library FastJson in Ali.

4. The system for automatically generating English bilingual subtitles in traditional Chinese medicine videos according to claim 1, wherein: and the analysis processing module generates a caption result based on the TextView.

5. The system for automatically generating English bilingual subtitles in a traditional Chinese medicine video according to claim 1, wherein the back-end compiling module has the functions of: 1) Searching by using a knowledge graph technology; 2) Modifying the subtitles; 3) Extracting human voice; 4) Audio transfer; 5) And (5) voice broadcasting.

Images