Machine translation is the process of using algorithms, or probabilistic models, like neural networks or AI to translate text automatically.

Machine translation tools use algorithms, statistical models and artificial intelligence to analyze the structure and patterns of text and translate it. Models can be ‘trained’ to learn the patterns in languages - improving their output.

Neural Machine Translation (NMT) is a technology based on networks of artificial neurons that predict the likelihood of a set of words in sequence. It is currently the most common technology used for quality machine translation.

A Large Language Model (LLM) is a model able to achieve language understanding and generation. They are trained on large volumes of content and are most famously used in generative AI chatbots like ChatGPT but are increasingly used as machine translation tools.

In Machine Translation, Human in the Loop refers to using a human expert at crucial stages in the translation process to improve the output quality. At Acolad we even prefer to say human at core. They might use prompts to improve the output of Generative AI content or resolve AI hallucinations. They might also adjust the output language to help it sound more natural or fit a desired style, add insight and human expertise. Essentially, they add what only humans can truly master. 

Machine Translation Post-Editing is part of the Human in the Loop process. A human language expert will take the Machine Translation output and carry out edits to ensure the final content meets quality standards and, for example, a house style or brand voice. With the development of AI, traditional post-editing has evolved into a more valuable process where linguists now focus on adding significant value, ensuring nuanced accuracy, cultural relevance, and higher-quality outputs in the translation process.

OCR refers to Optical Character Recognition. It’s a process that converts an image of text, for example, a screenshot containing words, into machine-readable text. This is useful for Machine Translation as it allows text within images to be automatically translated, so it does not need to be transcribed. Our innovative Portal can read the text in images, making it much easier to Machine Translate written content within images.

Natural Language Processing (NLP) refers to using artificial intelligence techniques to automatically translate text or speech from one natural language to another. While machine translation focuses on language translation, NLP is a broader field covering tasks like language understanding, generation, and more.