To summarise AI translation, it is a simple substitution of words from one language to another and it works on recognition of phrases and sentences to restructure according to the new language.
AI translation can also be referred to as machine translation (MT), automated translation, automatic or instant translation. Our sophisticated technology is specifically tailored to each client, allowing yourself to build up a bank of personalised translations in a secure network.
There are two types of AI translation systems:
- Rule-based systems use a combination of language, grammar rules and dictionaries for common words. Specialist dictionaries are created to focus on certain industries or disciplines to ensure that the terminology remains consistent as well as relevant for each industry. Rule-based systems typically deliver consistent translations with accurate terminology when paired with these specialist dictionaries but are rigid in its process.
- Statistical systems do not have ‘knowledge’ of language rules. Instead, the systems ‘learn’ to translate by analysing large amounts of data for each language pair. As it goes on, the AI begins to learn more about the language pair. Statistical systems can be trained for specific industries or disciplines as well by using additional data relevant to the sector needed. Typically, statistical systems deliver more fluent-sounding but less consistent translation – especially when it comes to terminology.
At Wolfestone, we have a bespoke hybrid AI translation engine. We combine these two systems to get the best of both worlds. All of this work comes on our usual secure servers as well, locked in a cloud to guarantee data protection for yourselves.
- Translating content quickly to get an understanding of information in another language – preferably used more internally rather than external documentation as it is simply for knowledge purposes.
- Increasing productivity of human translators by providing a first pass translation that can be quickly edited by a human in a system called post-editing. Productivity is increased up to 6,000 words per day.
- Translating instant content such as chat, email or customer support communications.
- Content that may be useful in another language, but where there is insufficient budget or the business case is not yet strong enough to warrant the investment of a full human translation.
- Google Translate can only translate 5,000 words at a time whereas we can do over 25,000 words per day.
Who should use AI translation?
Everyone can benefit from AI translation if used correctly:
- Individuals can benefit from the cost-saving of our AI translation for documents like certificates.
- Enterprises can use it to build a customised translation bank to see costs go down over time. Companies may have a volume and deadline that human translators cannot do either. For example for internal information use, it benefits in translating the key points for meetings across different branches around the world.
- Law firms have a lot of terminology recycling which means their translation banks are primed with repeat phrases.
- Governments use AI translation as part of their intelligence gathering and monitoring activities. In the times of tight purse-strings, it can save money with limiting budgets and be used to increase the productivity of human translators.
How fast is AI translation?
It depends on the service level. As a guideline, our bespoke AI translation engine can translate 25,000 words per day – more if it is all in a single document. Longer timeframes are needed when post-editing is required. Post-editing is a valuable service and quicker than traditional translation, but it can lower the speed to 4,000-6,000 words per day.
As the engine builds a bank of translation specific to each client, the process becomes much quicker and cost-effective.
Please note, these estimates do not take into consideration any pre-formatting or post-formatting.
What document types are supported?
Currently, the current formats are supported: doc, docx, xls, xlsx, ppt, pptx, xml and html. If your file is not one of these types, we are able to convert it and notify you to discuss your options.
Can AI translation handle complex formatting?
This is similar to normal translation service. It is text based rather than formatting related. The final document shape depends on post-translation DTP (desktop publishing) to ensure that they coincide.
Can’t I just use Google Translate or other free tools?
Yes, but you will not receive a quality nor accurate translation. Those tools are flawed and free for a reason. The search engines analyse all web content which leads to translation accuracy really suffering. There is no optimisation for the service but rather algorithms making guesses based on the web’s popularity. Popularity, though, is not always correct; especially on the internet, where shorthand is rife.
The biggest issue is that your data is then used to benefit everyone, not just yourselves. With us, a professional translation agency, we can quality check to spot the issues and notify you accordingly with a plan of action.
These tools share your data, it is not at all protected. When dealing with sensitive issues and confidentiality, it is not worth the risk.
What common errors can we expect from AI translation?
Word order in sentences can become confused depending on the target language. Idioms are another issue, similar to literal translations or perhaps a word with several meanings in the wrong context leading to meaning misinterpretation. All of this can be fixed by post-editors, should you opt for them.
Our AI Translation service levels
We have updated our services to be more comprehensive of your requirements. That is why we can offer our AI translation service in several levels to accommodate your needs, budgets and, most importantly, the final quality.
Below is a summary of our four service levels:
This service allows for roughly 25,000 words per day to be translated and possibly more if it is all in one file. The service can accommodate tight deadlines.
This is the most efficient service, but due to no human editing and the complexity of languages, the translation will not be perfect. We recommend this for internal use only when information needs to be understood and not published externally.
With monolingual post-editing, we can translate around 4,000-6,000 words per day. It can be scaled up for particularly tight deadlines.
Files will be translated using AI translation and then the target language will be edited by a linguist to make it more fluid. Monolingual post-editing is great for fast turnarounds of large volumes of content that has to be translated where the quality is not the most important criteria.
We recommend this to be used for in-house communication rather than anything external, legal or similar. It is much more fluid than simple AI translation, but it is far from perfect due to only monolingual post-editing. It is ideal for small tasks like instruction manuals under 500 words, clothes’ tags and products with small, quickly edited content.
Through this method, the service will approximately translate around 4,000 words per day, but after 1 million words have been translated, for example, the speed will increase to 6,000-8,000 per day and continue down that route.
This service is similar to monolingual but we have a post-editing translator work on both languages rather than just the target language. We recommend this over monolingual post-editing if you want a more fluid, natural translation that will be used externally to represent the company. The bonus of this service is that the quality of translation is higher yet the turnaround is still sharp. Although it won’t be perfect like a human translation, this is still a process that is for those tight deadlines or large volumes that need a more consistent translation.
As the translation bank builds, the turnaround becomes faster and cheaper. This then can build individual, customised AI for yourselves that really is something worth looking into if you require translation frequently.
This is the most comprehensive service, one which we believe to be the biggest and best in AI translation currently available in the market. Per project, there will be a variation on the timescale, with an average of 25,000 words per day in the end.
We custom build you an engine that will see your costs shrink over the course of our collaboration. This could lead to savings of 70-80% within a year or two – depending on the amount of work. It is something we recommend if you translate a high volume and it is better to centralise your translation with one company than to spread it out or use people in-house.
If you have already previously used us, we will input all previous work into our hybrid engines to begin building your bank with translations by human translators which will hugely improve the AI immediately. Although you may not have an immediate bank, with each passing project, the savings will be passed onto yourselves.