Posts tagged with "One Hour Translation"

One Hour Translation: Google Wins the Battle of Real Time Voice Translators

OHT used expert in-house linguists to compare the performance of Skype Translator, Google Assistant and Apple’s Siri in translating business and tourism expressions from English into Japanese, French, German and Spanish and vice versa

On average across languages, Google scored the highest – 4.54 out of 6, Skype second (4.32) and Siri third (4.09). Google was the best in Japanese (4.01), German (4.5) and Spanish (4.8), while Siri led in French (4.87) 

Google Assistant is the top performing real time voice translator, according to a benchmark conducted by One Hour Translation (OHT), an online platform which provides translations in more than 100 languages and 3,000 language pairs.

With demand for real time voice translation on the rise, OHT decided to test out the leading services: Skype Translator (run by Microsoft Translate), Google Assistant and Apple’s Siri, and rank them for accuracy.

With the help of expert in-house linguists OHT took 16 business and 10 tourism expressions and translated them from English into Japanese, French, German and Spanish and vice versa. The same sentences were then given to real time voice translator devices, apps and digital assistants to see just how they performed. The results were rated by the linguists on a scale of 0 to 6.

On average across languages, Google scored the highest – 4.54 out of 6, Skype second (4.32) and Siri third (4.09). Google was the best in three out of the four languages – Japanese (4.01), German (4.5) and Spanish (4.8), while Siri led in French (4.87).  Overall Japanese was the hardest language to translate with an average score of 3.7.  French was the easiest language for the instant voice translator devices to translate with an average of 4.75, followed by Spanish (4.54) and German (4.41).

“The real time voice translators were more accurate in translating tourism related experssions in comparison to business expressions” said Yaron Kaufman, chief marketing officer and co-founder of OHT. He attributed this to the use of a lot of business-related abbreviations which are not easily recognized by real time voice devices. Kaufman added that “despite the recent improvements in all of the assistants we tested, real time voice translations still cannot be relied on for business related content.”

Some examples of the sentences: “Stay on budget for this campaign, we can’t have it affecting our ROI”; “R&D are cutting too many corners, the product is undeployable”;  “Schedule a meeting between your CMO and our product manager”; “Do you have any allergies? This dish contains peanuts and avocado”; “I need to find the fastest way to the airport, my plane is leaving soon”; “My travel insurance should cover that bill.”

There are also new developments on the horizon in the field of real time voice translation. Among other advancements, Amazon is planning to release a DIY toolkit for creating translation apps and Xiaomi has released a new and advanced physical device for real time translations.

About One Hour Translation

One Hour Translation (OHT) believes that businesses should be able to reach any customer, anywhere, anytime, with no language barriers.

One Hour Translation’s AI powered cloud-based translation management platform, HALO,  helps enterprise customers reduce overhead by automating their translation process and workflow. HALO combines automated workflows, Neural Machine Translation (NMT) and professional translation services, to process all of the enterprise content quickly and easily via API/WEB. A dedicated NMT engine is automatically trained as the translations proceed and as a result the project’s cost keeps decreasing while translation speed improves. HALO is easy to implement and use, encrypted, secured and allows the enterprise to use any mix of its translators and reviewers with those of OHT, as well as any mix of NMTs for optimal quality and cost. The platform also allows OHT to manage a company’s resources and in-house budgets earmarked for translation in order to obtain the best possible results.

OHT is the leader in translations for enterprise customers, currently serving over 60 percent of the Fortune 500 companies, including Coca-Cola, Deutsche Bank, Microsoft, Amazon, IBM, HP, Xerox, Acer, Shell, Deloitte, HSBC, Procter & Gamble, IKEA, 3M, McCann, Allianz, Xiaomi and many other organizations.

One Hour Translation specializes in translation for 30 expert domains, including law, technology, marketing, website translation, applications, software and more.

Company Website

Twitter

Facebook

ONEs – OHT NMT Evaluation Score

Dramatic Improvement in Neutral Machine Translation

There has been a dramatic drop in the cost of translation services for major corporations in the past year and the savings amounts to at least 30%, according to One Hour Translation (OHT). The company, the world’s largest online translation agency, attributes the drop to the improved capabilities of NMT (Neural Machine Translation) technology.

The downward trend in translation costs is expected to continue as NMT engines are being trained to identify thousands of new words and phrases using the latest machine learning technology. That is good news for global e-commerce companies like Booking.com, Hotels.com, Agoda, Trivago, Amazon, Alibaba and others who depend almost daily on quality translation of large volumes of content for their millions of customers worldwide. 

“Around third of corporate content that we assess can already be translated using only NMT with no need for human input in the process,” said Ofer Shoshan, CEO and co-founder of OHT. He adds that the percentage is increasing all the time as the capabilities of the various translation engines continue to improve. The remainder is translated using hybrid translation that combines NMT and human post-editing, a process pioneered by OHT and utilizes human translators. 

OHT has developed a content assessment process for its corporate customers that is conducted by five or six experts who review the content. They quantify the content into three distinct categories that are defined and based on the amount of human input that will be necessary. The range is from no need for human input with the translation solely based on NMT engines, hybrid translation which combines NMT and human post-editing and a final category where human translation is required. As part of the process, OHT also recommends to its corporate customers the NMT engine that is best suited for a specific translation. 

“The translation services industry is being revolutionized with the cost of translation declining at a time when demand for rapid quality translations is skyrocketing,” noted Shoshan. The technological advances in the field, he believes, will lead to a further savings as the percentage of content translated by NMT or by hybrid translation approaches 50% or more. That process is enhanced as the NMT engines are trained and constantly updated to identify more and more words and phrases. In the coming years the trend is expected to completely change the global translation service market and sharply reduce the number of human translators.