The unfulfilled promise of MTPE
Localization professionals are no strangers to using technology to streamline the translation process. It’s standard practice, for example. For translators to use software to break down a text into manageable segments, apply previous translations. And perform quality assurance checks. In addition, translators and companies have on machine translation (MT) to generate approximate translations for years. Improvements in MT have also led to increased interest in MTPE (or machine translation post-editing)that. MT output that has been by a human.
Yet many believe that MTPE produces inferior results compared to having a human translate content and another human t it. As a result, quality-conscious companies have been reluctant to adopt MT and MTPE. Although more expensive, the traditional human-driven approach has traditionally been favored, especially by companies in highly regulated industries such as pharmaceuticals, life sciences, legal, and finance.
In short, there was no generally replacement for a human-driven translation workflow — until now.
There’s a tectonic shift underway in the translation and localization industry. The maturation of neural machine translation (NMT), advances in artificial general intelligence (AI), and the introduction of large language models (LLMs) have made things possible that were previously unthinkable.
Revolutionize your translation mix: human translation with artificial intelligence
AI-based human translation ( AIHT) bridges the quality gap between MTPE and human translation. Our AIHT solution delivers. A remarkable human-equivalent MQM score of 98+ while reducing costs per word by up to 50% and cutting turnaround time in half c to traditional human translation.
AIHT seamlessly integrates the strengths of AI. large language models (LLMs), machine translation, and human expertise into aworkflow, enhancing every step with AI-driven efficiency.
Furthermore, advances in AI are also enabling machine translation to produce excellent quality scores of up to MQM 93 for a fraction of a cent per word!
Laying the foundation
Traditionally, the workflow starts with applying translation memory, a saved record of all your previous translations. New content is compared to this record, and when there is a match of a certain confidence level. The translation is instead of having to translate from scratch. Translation memory can save companies 30-70% on their cost per word.
The difference with Smartling is that belize whatsapp number data 5 million we use AI to expand the coverage of our translation memory and improve cost efficiency by improving lower confidence matches to make them more accurate. We perform a process “fuzzy match repair” using our in-house Smartling MT engines to improve the fit of these low confidence matches.
Content that the translation memory was unable to address is sent to our AI translation stage. Our machine learning quality assessment tool evaluates the output of multiple machine translation engines and selects the highest quality option to use.
Post-processing
With the initial translations in place, it’s time for post-processing, a critical step to ensuring ready-to-use, brand-compliant translations. We start by applying MT and LLM-optimized glossaries (your list of important terms and how they images and visual effects should be translated), which go beyond substitution. The distinction is in ensuring that these substitutions fit the context of the string and are grammatically correct, rather than simply replacing the term.
We also automate content and format cleanup, addressing issues like whitespace, missing or extra tags, and placeholders. This formatting step is important because sometimes MT engines change the formatting, causing errors when strings are re-e into the platform or affecting translation memory cleanup.
Contextualized linguistic review
Finally, the translation is in context by an expert linguist handpicked for the project. This is what we call human-in-the-loop validation.
Our technology has already done the heavy lifting: translating the text and ensuring grammatical accuracy while making adjustments job data to align with brand guidelines. This frees up the professional linguist to focus on high-level validation and refinement.
All work is completed in the Smartling platform, so every change is and saved in real time, eliminating version control clutter. If questions arise, the linguist can communicate with you directly in Smartling.
Human quality at half the cost
This all sounds good, but it only works if human quality can actually be
In our latest report , we share our approach to measuring translation quality and the results we achieve across all workflows. Smartling’s rigorous process includes random sampling across multiple languages each month and a thorough review according to the Multidimensional Quality Metrics (MQM) framework—the industry standard for quality assessment.
Our research found
that all Smartling translation workflows consistently achieve high MQM scores, including AIHT.
Industry benchmarks for human translation range from 95% to 97%. With an average MQM score of 98, AI-based human translation is actually exceeding traditional human translation results from many language service providers. And it does so while reducing cost per word by 50% and improving time to market by 2x.
AIHT is having a huge impact on our customers.
For example, a large company with an annual translation volume of over 20 million words revolutionized its localization strategy by using Smartling’s AI- Human Translation. This cutting- approach substantial cost savings, turnaround times, andexceptional human translation quality. Before switching to Smartling, its translation were met through traditional human translation services.
Want to learn more about how AI-powered human translation can improve your localization program? Get in touch.