Swift, accurate and cost-effective Neural Machine Translation (NMT) Service with human post-editing
Neural Machine Translation (NMT) represents a cutting-edge AI-powered translation service which involves an end-to-end approach for automated translation, overcoming the weaknesses of classic, machine-based translation.
In comparison with previous Machine Translation solutions, NMT is a state-of-the-art approach, taking advantage of deep learning methods and resulting in a cleaner and more semantically correct translation output.
It represents the latest form of MT that utilizes a neural network closely based on the human brain, allowing it to categorize data into various groups and layers.
The core of NMT consists of existing bilingual databases and automated learning processes that contribute to the constant improvement of its efficiency, speed and quality.
How Phrase-Based Machine Translation (PBMT) differs from Neural Machine Translation (NMT)
While both NMT and PBMT at their core are forms of Machine Translation, how they perform differs from one another drastically.
PBMT is part of an older statistical method of Machine Translation, which involves predictive algorithms that teach a computer how a specific text should be translated.
PBMT takes its data from various previously-translated bilingual texts which highly influence the final quality.
Please note that NMT has several limitations in comparison to a standard linguist-based TEP process:
NMT is available only in a select number of languages due to a need to "train" the system beforehand with large quantities of data. The number of languages NMT is available in is constantly growing.
Only one linguist participates in the Machine Translation Post Editing stage of the NMT process.
NMT is not a replacement for a standard human-based TEP process, it is an alternative to it. NMT quality will always be lower in comparison to TEP.
NMT offers the following benefits for your translation workflows:
Cost effective solution
High translation speed
The closest result to human-based translation
Higher quality translation in comparison to PBMT
Significantly faster learning period in comparison to PBMT
Final NMT translation stages require less post editing in comparison to PBMT
... and many more!
The source content is analysed in terms of the subject matter, target language(s) and the expected quality of the target content in order to determine the applicability of NMT.
The translatable content is analysed and prepared for translation.
The NMT engine is used to translate the content.
The NMT output is revised by a linguist. Depending on the desired quality of the target content, two options are possible:
- Light MTPE (Machine Translation Post Editing), which produces understandable target text, without focusing on style, consistency or fluency of the target text.
- Full MTPE, which produces a more accurate translation than Light MTPE and is closer to the output produced by a human translator.
For our PREMIUM translation and localization projects, we use our specialized translation, editing and proofreading (TEP) workflow, which results in translations of the highest quality for all your projects. Below is a simplified version of the steps involved for each translation project.
AD VERBUM works with more than 150 languages, covering a wide range of content types and industries with our certified linguist pool of more than 3,500 experienced subject matter experts.
We offer multiple pricing options for all of our translation services.
Translation service quality certified company.
Quality management system certified company.
Information security standard certified company.
+371 6 7229 430
©2002-2023, AD VERBUM Ltd.