10 results about "Parallel translation" patented technology

The invention provides a real-time translation method and interaction system based on streaming voice segmentation and semantic verification. The method comprises the following steps: receiving a conference audio stream, identifying effective audio frames based on dual-channel voice activity detection, and pressing the effective audio frames into a dynamic buffer area; according to a preset dynamic segmentation strategy, outputting an initial voice segment from the dynamic buffer area for voice recognition, and obtaining a corresponding initial text segment; performing multi-stage reliability verification on the initial text fragment, and dynamically correcting or complementing the initial text fragment according to a verification result to obtain a reliable recognition result; and after the reliable identification result is obtained, triggering an asynchronous parallel translation task. According to the method, a semantic correction mechanism cooperating with the adaptive truncation depth is designed, so that the semantic fragmentation problem of the long-sequence audio during streaming truncation is solved, and the translation accuracy of a complex word order language is greatly improved while low delay is ensured.

The invention discloses a multi-device and multi-language synchronous translation device and method based on a block chain and double AIs, and relates to the technical field of intelligent translation and edge computing, the device comprises a translation machine terminal, an edge computing node and a cloud server which form an end-edge-cloud three-level collaborative architecture, a translator terminal collects voice and lip shape information through a multi-mode sensing module, performs preliminary recognition by means of an edge lightweight AI unit, and realizes safe and rapid networking based on a block chain identity unit and an NFC touch networking unit. The edge computing node deploys a block chain group management module, a semantic collaborative optimization module and a local cache retransmission module, and is responsible for multi-language translation, semantic optimization and reliable distribution. The cloud server realizes global model training and term library management; according to the method, low-delay and high-accuracy multi-device synchronous interactive translation is realized through device authentication, voice recognition, edge optimization, multi-language parallel translation and adaptive transmission, and the translation efficiency and security in a multi-user and multi-language scene are remarkably improved.

Provided is a pseudo parallel translation data generation apparatus for generating pseudo parallel translation data for accurately performing machine translation in an adaptation target domain even when there exists no parallel translation data for the adaptation target domain. Using other-domains parallel translation data D0(L1-L2), other-domains first language data D0(L1), other-domains second language D0(L2), adaptation target domain first language data D0(R1), and adaptation target domain second language data D0(R2), the pseudo parallel translation data generation apparatus 100 performs optimization processing for a cross-lingual language model including an input data embedding unit 2 and an XLM processing unit 3, and performs parameter optimization processing for a pseudo parallel translation data generation NMT model including the input data embedding unit after the optimization processing and a machine translation processing unit 5. Performing processing using the pseudo parallel translation data generation machine translation model obtained by the parameter optimization processing allows for obtaining pseudo parallel translation data for the adaptation target domain for which no parallel translation data sets exist.

The invention provides a multi-format translated file processing method and system, electronic equipment and a storage medium, and the method comprises the steps: carrying out the structural analysis of a translated file when determining that the format of the received translated file meets a format requirement, and obtaining a translation key value; based on the translation key value, processing a data structure and a format of translation content in the translation file to obtain a target translation file; performing multi-language translation on the target translation file, and performing structured storage based on a translation result of each language; and when the intelligent translation function is triggered, performing multi-language translation on the target translation file to obtain a translation result. Parallel translation processing of multiple languages is innovatively realized, so that the translation efficiency and the translation precision are improved. The translation efficiency is improved by adopting a key algorithm of source language intelligent detection and batch translation mapping. According to the technology, the labor cost is remarkably reduced, and the human translation requirement is reduced. In addition, the maintenance cost is reduced through a unified management mechanism.

The invention discloses a cross-platform multi-terminal multilingual real-time translation device and method based on an edge block chain, and relates to the technical field of intelligent translation and cross-platform collaboration.The device comprises a mobile cross-platform layer, a translator terminal, an edge computing node and a cloud server. A smart phone is used as a core hub, and cross-platform equipment rapid networking and collaboration among an ios system, an Android system and a translation machine are achieved. The method comprises the following steps: completing cross-platform identity authentication and security networking based on a block chain; the mobile terminal collects audio and completes lightweight AI recognition and sentiment analysis; semantic optimization and multilingual parallel translation are carried out on the edge nodes; accurate distribution and personalized playing of translations are realized based on a block chain account book; according to the method, emotion recognition, quantum encryption and federated learning technologies are fused, low-delay, high-accuracy and high-security cross-language real-time interaction is realized, and the efficiency and experience of multi-user multi-language communication in a mobile scene are remarkably improved.

The utility model relates to a container type gas generator set shifting device which comprises a container body with one end provided with an opening, and two parallel translation guide rails are installed above the bottom wall of the container body. During maintenance, the box door is opened to slide the lifting guide rail towards the direction of the translation guide rail, so that the lifting guide rail is aligned with the adjacent end of the translation guide rail, and the lifting guide rail and the translation guide rail are connected with the sliding seat. At the moment, the electric push rod is started to extend to push the generator set main body mounted on the sliding seat to move to the position above the horizontal section of the lifting track along the translation guide rail, and the bottom of the generator set main body is supported to a proper height through the lifting track; and the bottom of the generator set main body can be overhauled without being hoisted by external hoisting equipment, so that a user can use the generator set conveniently.

Provided is a machine translation processing system that enables highly accurate machine translation of source sentences including markup language tags while retaining information about the markup language tags, without the need to prepare a large amount of parallel translation data including tags. As described above, in the machine translation processing system 1000, the training data generation device 1 performs training data generation processing to detect the start / end correspondence codes and replace the detected start / end correspondence codes with alternative codes in parallel translation sentences that do not include markup language tags, thereby allowing for easily generating a large amount of data equivalent to parallel translation data into which markup language tags have been inserted. Using the parallel translation data obtained through the training data generation processing by the training data generation device 1 in the machine translation processing system 1000 as training data for learning processing of the machine translation model achieves the same advantageous effect as when the machine translation model learning processing is performed using the parallel translation sentences with markup language tags as training data.

The application discloses a kind of enterprise multilingual document translation method, device, equipment and storage medium, it is related to computer technical field, comprising: judging the document type corresponding to the document to be translated, and based on the judgment result to the document to be translated is parsed, to obtain corresponding text to be translated;Based on the text to be translated and utilize translation engine pool to carry out parallel translation, to obtain each initial translation result, the double check is carried out to the initial translation result, to obtain corresponding check result;Based on check result whether the initial translation result satisfies preset standard condition is judged, the obtained determination result is used to determine target translation result satisfying the preset standard condition;The target translation result is filled in the target area of the document to be translated using the document type, and based on the target translation result, the layout of the document to be translated is adjusted, to complete corresponding document translation operation.The accuracy and reliability of document translation are improved.

The invention belongs to the field of angle steel punching machines, and discloses a large angle steel production punching machine which comprises a lower supporting frame, a first supporting piece, a second supporting piece and a hydraulic assembly. The first supporting piece is perpendicular to the side wall of the lower supporting frame. When the punching machine is switched to apply pressure to the other side of the angle steel, damping liquid of the hydraulic assembly on the original non-stress side flows back, the locking state of the self-locking piece is relieved, the abutting force of the second supporting piece is weakened, and when the angle steel is pressed to move horizontally, the triggering assembly drives the hydraulic assembly and synchronously activates the second supporting piece on the opposite side to abut against the side wall of the angle steel to restrain outward movement. Meanwhile, the self-locking piece of the first supporting piece on the same side as the stress side rotates to form a rigid stop, the angle steel moving path is limited to be parallel translation perpendicular to the stress direction, interlayer dislocation caused by downward sliding of the upper-layer angle steel is effectively restrained, and sliding of the upper-layer angle steel is limited, so that when a flexible layer in the middle of the double-layer angle steel deforms, the deformation of the flexible layer is avoided; and the coaxiality of the upper-layer angle steel punching hole and the lower-layer angle steel punching hole can still be kept.

The application provides a continuous real-time adjustable focus endoscopic OCT probe, which comprises an imaging module, a moving module and a driving module; the driving module comprises a rotating motor, two parallel translation modules, a first connecting piece and a second connecting piece, the rotating motor is installed on one of the translation modules, the first connecting piece is connected with the rotating motor; the second connecting piece is connected with the other translation module; the moving module comprises a coaxial cylinder, a bearing, an optical fiber and an elastic component, the coaxial cylinder is formed by coaxially sleeving a plurality of pipe bodies with different diameters, the bearing is sleeved into the coaxial cylinder, the optical fiber is sleeved into the bearing, the moving module is connected with the first connecting piece through the elastic component; the bearing is connected with the second connecting piece through a third connecting piece; the imaging module is installed at the front end of the optical fiber; the rotating motor drives the first connecting piece to rotate and drives the optical fiber to rotate, one of the translation modules drives the second connecting piece to move, so that the distance between the optical fiber and the imaging module is adjusted to adjust the focal length.