How far can we rely on machine learning for language translation? | Techsauce
Contact us
18

How far can we rely on machine learning for language translation?4 min read

Posted by
Posted date April 17, 2019
Two years ago, machine translation, let alone Google translate, was not very effective at translating complex languages like Japanese or Arabic. Today, automated translation quality has gotten much better. With constant development in artificial intelligence (AI) and machine learning, language translation will certainly undergo a lot of changes.

Since 2007, Andovar has been at the forefront of language technology. Their proprietary technology has helped large companies, some of which are Fortune 500, scale from the US and EU to penetrate into the Asian market and aided Asian companies to expand outside of their continent. From text to audio and video, Andovar is able to effectively and uniformly translate technical and formal language quickly and accurately, whether it is in the e-learning space, gaming or audio-visual industries. In this interview, the CEO and Founder of Andovar, Conor Bracken shared great insights into the current state of the translation industry and what the future holds for machine translation.

What does “glocalization” mean and why is it important for businesses?

Glocalization is the creation of products or services for the global market by adapting them to local cultures. At its core, the business is going global while paying attention to the local. It is costly and complicated to compete globally due to the pronounced differences in local cultures. In different locales, there are linguistic, cultural and technical differences. As an example, the word “localisation” is spelt as “localisation” depending on which side of the Atlantic Ocean you are from. Technologically, some languages are written from right to left, instead of left to right. Even today, companies still struggle to convey their message consistently, across language barriers and cultural understanding, without trading off on cost, timeliness, and quality.

Which industry benefits most from your services?

Professional language translation brings benefits to any industry when the company wants to expand internationally. Translation serves as the enabler for businesses to penetrate a new market where a different language is used. For us, we work a lot with gaming companies coming from West to East and East to West. Localizing the game, testing it on Android or iPhone, and doing voiceover are the things we do. Some of our big customers are Chinese game companies that want to expand globally. In Thailand, only 5% of the games were born in Thailand, with around 50% being Chinese or Japanese games and 50% western games. We help those companies get to their local customers where there are language and cultural barriers that need to be broken down. Beyond gaming, we also do a lot of translation work for more specific industries like movies and e-learning.

How much machine translation is being used at Andovar?

All translation companies nowadays are trying to see where they can use machine translation; it doesn’t replace human. It’s the humans who will edit it before the translated piece goes out. It’s quite difficult because many clients want to use machine translation to reduce cost, but they also have to understand the challenges that come with machine translation. To answer your question, at Andovar, we use machine translation to translate the basics and have human translators do the final touch up.

What are some limitations of machine translation?

  • Certain languages are difficult to translate: French to English or Italian to English is definitely much easier than Japanese to English as there are a lot of similarities between those western languages.

 

  • Difficult for marketing: Marketing is about making customers fall in love with your brand. Keywords are very important and directly translating a sentence using AI will not make sense most of the time.

 

  • User interface: This is very challenging and probably impossible to be done properly by machine as these applications do not have any form of context.

What’s the future of machine translation industry?

In order to engage successfully with global markets, enterprises increasingly need to manage fast-moving streams of multi-lingual content from both within its international offices as well as the wider communities which they wish to engage with.

For many years, people rejected Google translate and machine translation. In the past 2-3 years, the quality of the off-the-shelves system and custom machine translation has got a lot better. Everybody is trying to integrate machine translation as the quality is becoming better and better. However, we don’t see a point that machine translation will completely replace human translation, at least not for the next 20 years. We can definitely see that the efficiency gain has picked up. So right now, machine translation is here to improve productivity rather than completely replace human translation.

Another thing is automatic dubbing with a computer voice. Instead of using human, some companies just use software to do this job. But it’s not coming close to replacing humans. Consumers and people listening to e-learning courses or foreign movies hate listening to robotic voices, which is why we won’t see human dubbing disappearing any time soon.

On the other hand, there are opportunities for live broadcasting in which a live translation is needed. People tend to not want to wait, otherwise, what’s the point of the live shows? Automated translation could help with this process, at least, an automated voice is better than nothing.

 

A journalist who loves writing about business, startup, and tech for Techsauce Global. Catch me hopping around South East Asia and China or reach out to me at vanessa@techsauce.co

Comments

comments

Sign-up for exclusive content. Be the first to hear about ConvertPlug news.
Subscribe