This site uses cookies.
Some of these cookies are essential to the operation of the site,
while others help to improve your experience by providing insights into how the site is being used.
For more information, please see the ProZ.com privacy policy.
It’s the latest offering from the German tech unicorn
German tech darling DeepL has (finally) launched a voice-to-text service. It’s called DeepL Voice, and it turns audio from live or video conversations into translated text.
DeepL users can now listen to people speaking a language they don’t understand and automatically translate it to one they do — in real-time. The new feature currently supports English, German, Japanese, Korean, Swedish, Dutch, French, Turkish, Polish, Portuguese, Russian, Spanish, and Italian.
What makes the launch of DeepL Voice exciting is that it runs on the same neural networks as the company’s text-to-text offering, which it claims is the “world’s best” AI translator.
As someone who’s just moved to a foreign country, I’m keen to try a voice-to-text translator that actually might work. All the ones I’ve tried so far aren’t real-time — there’s a lag that renders them pretty useless — and the translation quality is pretty poor.
For face-to-face conversations, you can launch DeepL Voice on your mobile and place it between you and the other speaker. It then displays your conversation so each person can follow translations easily on one device.
You can also integrate DeepL Voice into Microsoft Teams and video-conference across language barriers. The translated text appears on a sidebar as captions. It remains to be seen whether DeepL Voice will be available on platforms like Zoom or Google Meet anytime soon.
Google has announced that it has extended AI voice capabilities to over a dozen new African languages across a range of Google services.
Google — which already supports typing with a custom keyboard in Gboard for approximately 200 African languages, and machine translations for over 60 African languages in Google Translate — now supports voice search, talk-to-type on Gboard, and dictation on Google Translate for 15 regional languages.
The development means that the company has more than doubled the number of African languages that enable speech-to-text in Google Translate and has doubled existing voice input support for Gboard and voice search in the region.
Daan van Esch, Technical Program Manager at Google, said that the update “will enable around 300 million more Africans to use their voice to interact with the web.”
Speaking recently to Slator about the challenges and opportunities of the language services market in Africa, Christian Elongue, Managing Director of Kabod Group said, “there is limited training data that many African languages are facing, [and there are multiple initiatives] contributing to creating data sets for various low-resource African languages.”
Simultaneous machine translation (SiMT) aims to deliver real-time translations as a source language, spoken or written. Traditionally, this requires models that control when to “read” more of the source and when to “write” the translation — decisions that rely on intensive model training, complex model designs, and significant computing power.
Now, researchers Libo Zhao, Jing Li, and Ziqian Zeng from Hong Kong Polytechnic University and South China University of Technology have introduced PsFuture, a zero-shot, adaptable read/write policy that enables SiMT models to make real-time translation decisions without additional training.
The researchers said they drew inspiration from human interpreters, who dynamically decide when to listen and when to speak based on evolving contexts. “Interpreters shift from listening to translating upon anticipating that further future words would not impact their current decisions,” they explained.
PsFuture allows translation models to make similar, context-aware decisions, leveraging “the model’s inherent linguistic comprehension and translation proficiency” and eliminating the need for further training.
Simulated Look-Ahead
Rather than relying on a fixed number of source words to determine the right time to start translating, PsFuture allows a model to anticipate what’s coming next. By using pseudo-future information — a simulated, brief “look-ahead” similar to how interpreters anticipate what might come next in a sentence — the model assesses if additional context would change its next translation output. If not, the model proceeds with translating. If more context is needed, it waits to “read” further.
A stop sign in English, French and Inuktut syllabics is seen in Iqaluit, on April 25, 2015. One of the most widely spoken Indigenous languages in this country is now available through Google’s translation service, the first time the tech giant has included a First Nations, Métis or Inuit language spoken in Canada on its platform. THE CANADIAN PRESS/Paul Chiasson
By Brittany Hobson, The Canadian Press
Posted October 17, 2024 9:00 am.
Last Updated October 17, 2024 4:10 pm.
One of the most widely spoken Indigenous languages in this country is now available through Google’s translation service, the first time the tech giant has included a First Nations, Métis or Inuit language spoken in Canada on its platform.
Inuktut, a broad term encompassing different dialects spoken by Inuit in Canada, Greenland and Alaska, has been added to Google Translate, which translates text, documents and websites from one language into another.
The latest addition is part of a Google initiative to develop a single artificial intelligence language model to support 1,000 of the most spoken languages in the world.
There are roughly 40,000 Inuktut speakers in Canada, data from Statistics Canada suggests.
The number of speakers alone is not enough to determine whether a language can be included in Google Translate, said Isaac Caswell, a senior software engineer with the platform.
There also has to be enough online text data to pull from to create a language model.
Other Indigenous languages in Canada have “had simply too little data to have any usable machine translation model,” said Caswell.
For example, engineers looked at adding Cree, which is spoken by more than 86,000 people in Canada, but there were fewer websites in the language to pull from.
“We don’t want to put anything on the product which just produces broken text or nonsense,” said Caswell.
Chinese e-commerce company Alibaba has invested heavily in its fast-growing international business as growth slows for its China-focused Taobao and Tmall business.
BEIJING — Chinese e-commerce giant Alibaba’s international arm on Wednesday launched an updated version of its artificial intelligence-powered translation tool that, it says, is better than products offered by Google, DeepL and ChatGPT.
Alibaba’s fast-growing international unit released the AI translation product as an update to one unveiled about a year ago, which it says already has 500,000 merchant users. Sellers based in one country can use the translation tool to create product pages in the language of the target market.
The new version is based only on large language models, allowing it to draw on contextual clues such as culture or industry-specific terms, Kaifu Zhang, vice president of Alibaba International Digital Commerce Group and head of the business’ artificial intelligence initiative, told CNBC in an interview Tuesday.
“The idea is that we want this AI tool to help the bottom line of the merchants, because if the merchants are doing well, the platform will be doing well,” he said.
Large language models power artificial intelligence applications such as OpenAI’s ChatGPT, which can also translate text. The models, trained on massive amounts of data, can generate humanlike responses to user prompts.
Language AI tools are transforming the industry, boosting efficiency, cutting costs, and driving growth – with DeepL usage far outpacing Google, Microsoft and more
COLOGNE, Germany, Oct. 9, 2024 /PRNewswire/ — DeepL, a leading global Language AI company, has been named the #1 most-used machine translation (MT) provider among global language service companies in a new 2024 ALC Industry Survey report by the Association of Language Companies (ALC) and Slator. The company’s rise to market leadership, coupled with its exponential growth – DeepL now serves over 100,000 business and government customers worldwide – highlights the growing significance of AI-powered translation solutions in transforming industries, including language services, manufacturing, legal, healthcare and more.
“This exciting milestone highlights the accuracy and reliability of DeepL’s specialized Language AI platform, which is trusted by businesses worldwide for critical translation projects. It also is a testament to our positive impact on their cost savings, efficiency, and growth,” said Jarek Kutylowski, CEO and Founder, DeepL. “As AI in language services gains in popularity, we are honored to be the industry’s preferred Language AI partner and are committed to providing industry-leading, cutting-edge, specialized tools for translation, AI-driven content creation, and more.”
The new ALC report surveyed 127 language service companies (LSCs) from 28 countries*. The results underscore the expanding role of machine translation in the services offered by LSCS to key industries such as healthcare, law, and education.
We haven’t been able to test Meta’s live translation, so we can’t say how seamless it will be. But development of these products is coming fast.
Justin Dawes
The Ray-Ban Meta AI-powered glasses are getting a feature for live voice translation.
Mark Zuckerberg, CEO and founder of Meta, demonstrated the new feature and a slew of other updates during the keynote speech of the Meta Connect developer conference, streaming from the company’s main campus in California.
When talking to someone who speaks Spanish, French, or Italian, the user wearing Ray-Bans should be able to hear a real-time English translation through a speaker in the glasses, he said. The user then replies in English, and a mic in the glasses picks up the voice and transfers it to the user’s connected mobile app. The translation from English to the other language is voiced aloud through the app for the other person to hear.
The implications for travel are obvious, and a press release from Meta about the upgrade said as much: “Not only is this great for traveling, it should help break down language barriers and bring people closer together.”
The company did not say when the update is coming but said it would be soon. The company plans to add more languages in the future.
Meta earlier this year said that it was integrating voice-activated AI into the glasses, which meant it could translate menus and answer questions about landmarks seen through the lenses.
Zuckerberg demonstrated the new translation feature with Brandon Moreno, former two-time UFC Flyweight Champion. Moreno spoke in Spanish to Zuckerberg, and Zuckerberg responded in English to Moreno.
The demo was brief, but the chatbot was able to translate in real time despite some slang and pauses.
Every time I tell someone that I’m a Chinese and Italian Foreign Languages major, I watch them do a slight double take. It bothers me, I’ll admit, but not for the reasons you’d expect. I’m frustrated that I can’t seem to conjure a response to their unspoken questions.
“Why learn languages you have no personal connection to? Ones spoken by so few people so far away? Why spend hours memorizing conjugation tables when you’ve already met the language requirements?” In short: “Why learn a new language at all?”
I’ve been studying Chinese for practically my whole life, Italian for a little over a year and French and Nepali during high school and my gap year. And still, I find that the more I throw myself into learning languages, the further I am from conjuring any answer to that simple question: Why?
If you have ever sat through lunch with native speakers at Oldenborg, agonized over verb agreements in office hours or even tried to converse with locals or extended family members in a foreign country, then you already know how it feels to exist in a sort of linguistic exile.
On good days, I feel a mix of pride and exhaustion. On bad days, I feel like setting myself to this impossible task is almost comically akin to Sisyphus’ struggle, except that, unlike him, I have a choice.
I’ve learned that however fluent you may think you are, there is no finish line in learning a language: The mountain remains perpetually stretched out before you, and you must continue to push that boulder further and further up.
Researchers and people from the deaf community have teamed up to co-create a sign language machine translation (SLMT) app.
The research team designed a theatrical performance in sign language, seen through the eyes of artificial intelligence (AI).
“Historically, deaf people have been excluded from the development of automatic translation technologies,” explains Shaun O’Boyle, Research Fellow in the School of Inclusive and Special Education (Dublin City University DCU).
“This has often caused backlash and resistance from deaf communities, as the projects were designed and developed without any input from the very end-users they intended to serve—resulting in a technology no one wanted to use and a big waste of money,” adds Davy Van Landuyt, Project Manager at the European Union of the Deaf (EUD).
For this research, the team decided to reverse the standard approach, with O’Boyle, Van Landuyt, and the other partners of the European project SignON —including the Vlaams GebarentaalCentrum (Flemish Sign Language Centre) — asking participants “If we were to introduce an AI to Shakespeare texts in Irish Sign Language, which extracts would we choose first?
This engagement with the AI allowed them to connect with the audience and gather their opinions about the technology.
In a July 29, 2024 paper, researchers from Apple and the University of Southern California introduced a new approach to addressing gender bias in machine translation (MT) systems.
As the researchers explained, traditional MT systems often default to the most statistically prevalent gender forms in the training data, which can lead to translations that misrepresent the intended meaning and reinforce societal stereotypes. While context sometimes helps determine the appropriate gender, many situations lack sufficient contextual clues, leading to incorrect gender assignments in translations, they added.
To tackle this issue, the researchers developed a method that identifies gender ambiguity in source texts and offers multiple translation alternatives, covering all possible gender combinations (masculine and feminine) for the ambiguous entities.
“Our work advocates and proposes a solution for enabling users to choose from all equally correct translation alternatives,” the researchers said.
For instance, the sentence “The secretary was angry with the boss.” contains two entities — secretary and boss — and could yield four grammatically correct translations in Spanish, depending on the gender assigned to each role.
The researchers emphasized that offering multiple translation alternatives that reflect all valid gender choices is a “reasonable approach.”
Unlike existing methods that operate at the sentence level, this new approach functions at the entity level, allowing for a more nuanced handling of gender-specific references.
The process begins by analyzing the source sentence to identify entities (such as nouns or pronouns) with ambiguous gender references. Once identified, two separate translations are created: one using masculine forms and another one using feminine forms. The final step integrates these translations into a single output that maintains the grammatical integrity of the target language.
AI video creation platform D-ID is the latest company to ship a tool for translating videos into other languages using AI technologies. However, in this case, D-ID also clones the speaker’s voice and changes their lip movements to match the translated words as part of the AI editing process.
The technology stems from D-ID’s earlier work — which you may recall from the viral trend a few years ago where users were animating their older family photos, and later those photos were able to speak. On the back of that success, the startup closed on $25 million in Series B fundraising in 2022 with an eye on serving its increasing number of enterprise customers in the U.S. who were using its technology to make AI-powered videos.
With the company’s now-launched AI Video Translate tech, currently being offered to D-ID subscribers for free, creators can automatically translate their videos into other languages to help them expand their reach. In total, there are 30 languages currently available, including Arabic, Mandarin, Japanese, Hindi, Spanish and French, among others. A D-ID subscription starts at $56 per year for its cheapest plan and the smallest number of credits to use toward AI features and then goes up to $1,293 per year before shifting to enterprise pricing.
D-ID suggests the new AI video technology could help customers save on localization costs when scaling their campaigns to a global audience in areas like marketing, entertainment, and social media. The technology will compete with other solutions for both dubbing and AI video.
In a July 30, 2024 research paper, Otso Haavisto and Robin Welsch from Aalto University presented a web application designed to simplify the process of adapting questionnaires for different languages and cultures.
This tool aims to assist researchers conducting cross-cultural studies, enhancing the quality and efficiency of questionnaire adaptation, while promoting equitable research practices.
Haavisto and Welsch highlighted that translating questionnaires is often costly and “resource-intensive,” requiring multiple independent translators and extensive validation processes. According to the authors, this complexity has led to inequalities in research, particularly in non-English-speaking and low-income regions where access to quality questionnaires is limited.
In questionnaire translation, maintaining semantic similarity is crucial to ensure that the translated version retains the same meaning as the original. As the authors noted, “semantic similarity is more important than word-by-word match.” According to the authors, cultural nuances and colloquial expressions can further complicate this process, making it difficult to achieve accurate translations.
To address these challenges, they developed a web application that allows users to translate questionnaires, edit translations, backtranslate to the source language for comparisons against the original, and receive evaluations of translation quality generated by a large language model (LLM).
In May 2024, researchers emphasized the crucial role that emotions play in human communication and introduced a new dataset designed to enhance speech-to-text and speech-to-speech translation by integrating emotional context into the translation process.
In July 2024, Alibaba incorporated speech emotion recognition (SER) into its FunAudioLLM to retain original emotions in AI-powered interpreting.
Building on this, an August 6, 2024, paper by Charles Brazier and Jean-Luc Rouas from the University of Bordeaux demonstrated how to integrate emotional context into large language models (LLMs) to condition translation and improve quality.
They argue that “conditioning the translation with a specific emotion would use a suitable vocabulary in the translation.”
This research builds on the authors’ previous work, which was the first to explore combining machine translation (MT) models with emotion information. Their earlier study demonstrated that adding emotion-related data to input sentences could enhance translation quality. In this latest study, Brazier and Rouas take the concept further by replacing the MT model used in their prior work with a fine-tuned LLM.
Although event software company Canapii officially launched in 2020, its roots in the events sector stretch back two decades. Initially part of Canalys, a market research company in the IT industry, Canapii’s journey began with the need to automate event processes to minimise human error.
“We started developing an app 18 years ago to streamline our events, initially for internal use,” recalls CEO Rita Chaher. “However, as our sponsors, including tech giants like Lenovo and Microsoft, saw its potential, they requested to use it for their events.”
The pivotal moment came for the Reading, UK-based company in 2019 when Canapii secured a contract to manage several major events. The onset of the pandemic in 2020 necessitated a rapid pivot to virtual events.
“We redeployed our resources, recoding everything to support virtual formats. This shift revealed the tool’s potential beyond the IT industry,” Chaher explains. This adaptation was crucial in establishing Canapii as a standalone entity, focused on both on-site and virtual events.
The relentless pace of software companies rolling out translation AI as a feature continues unabated. Video hosting and sharing services provider, Vimeo, has announced the release of AI-enabled video translation and voice cloning features to its video platform.
The company’s new AI-enabled translation functionality promises to “leverage generative AI to translate video, audio and captions into dozens of languages – while replicating the original speakers’ voices.”
The functionality, which also allows users to add and edit automatically generated captions, is the latest in a series of AI enhancements designed to boost the video-for-business enterprise user base. As part of the release, Vimeo allows enterprise users to generate a free, 30-second sample of the translations, after which it charges on a per-minute basis.
Users can upload videos with multiple speakers and can rate the quality of the translation by selecting a thumbs up or thumbs down icon for each language.
Ashraf Alkarmi, Chief Product Officer at Vimeo said that the company’s AI translation solution “goes beyond simple transcription.” He added that the tool allows users to “maintain the original speaker’s authentic nuances and tone […] in any language.”
The company did not reveal, however, if it is working with a third-party commercial AI translation provider to power the technology.
Trados, the industry-leading translation platform by RWS, has recently launched an innovative ‘AI Essentials’ add-on, simplifying the integration of AI into translation workflows. AI Essentials combines two groundbreaking capabilities introduced earlier this year – Generative Translation and Smart Review – designed to enhance translation quality, speed up time-to-market, and cut translation costs. With no separate LLM subscription required, this add-on empowers organizations to customize their Trados solutions in line with their unique business needs, unlocking new possibilities to translate everything efficiently.
Additionally, we have also released a collection of new enhancements to The Trados platform. Here are just a few of our latest innovations:
Improve quality and reduce turnaround times withbetter translation management
PerfectMatch for efficient translation: We have now introduced PerfectMatch in the browser as a new step in the workflow, bringing this powerful capability to the entire Trados portfolio.
Previously only available in the desktop application, PerfectMatch is a form of context match that compares source files to existing bilingual files rather than to a translation memory. By identifying matching segments from the previous document and considering the surrounding context, it can provide a perfect match.
The Rask AI Platform: A Pioneer in AI Video Translation
In today’s world, as it globalizes, video content creators are under increasing pressure to make their creations available to an ever more diverse international audience. It is now much easier to translate videos with the advent of technology mayflies and Artificial Intelligence (AI) rising. We no longer speak of it as being a difficult or hard task. Innovative tools powered by AI, like the Rask AI Platform, are leading the way in this changed scene.
Video creators can, therefore, extend social media by giving their material actual global reach. Through such platforms, a lot of different features are supported: real-time translation, many video platforms, and localizing videos into multiple languages. This happens most strikingly through these functions, which enable creators to overcome language barriers and reach out to a variety of linguistic and cultural communities around the world.
Best Video Translation Apps
Here is the original list of top AI-driven video translation tools, which are aimed at helping content creators bring their video content to the world. These online video software options are based on the latest new techniques in neural networks and designed right from the metal up to do a full translation that is authentic and fast.
COCONUT CREEK, FL. – MotionPoint, the leading website translation and localization platform, announces the official date of AdaptiveCon 2024! The ultimate event for exploring the latest breakthroughs in translation technology for websites will take place virtually for free on August 7th, 2024, at 12 PM ET.
Why Attend AdaptiveCon 2024:
Cutting-Edge Technology Revealed: MotionPoint will introduce groundbreaking advancements in website translation technology. Don’t miss the opportunity to be among the first to witness these innovations.
Industry Insights: Our event is designed for marketing leaders and web developers eager to stay at the forefront of website translation technology. If you’re looking to harness the latest AI advancements to optimize your translation budget, streamline processes, and enhance quality, this event is for you.
Key Learnings:
Maximize your translation budget by identifying which pages require post-edit translation with groundbreaking AI.
Leverage Brand-Voice AI to improve style-guide adherence, and glossary adherence.
How multinational companies can balance marketing messaging in different countries from Subway’s globalization leader after doing so in over 100 countries.
DeepL SE, a well-funded translation software startup that leverages customized artificial intelligence models for improved accuracy over traditional platforms, has announced the debut of its most powerful AI model yet.
The startup’s next-generation language model is said to be designed specifically for translation and editing tasks. It’s based on a highly specialized large language model that’s fine-tuned on enormous amounts of proprietary language data. The company says the LLM provides more “human-like translations” with a reduced risk of hallucinations and misinformation.
The new model’s skills were enhanced via a human model tutoring process that involved thousands of handpicked language experts, who were hired to “tutor” the model to ensure the accuracy of its translations.
Germany-based DeepL has emerged as a rival to better-known translation systems such as Google Translate and general-purpose AI models such as ChatGPT, which can perform translations as part of a much wider repertoire of skills. Because DeepL’s systems are laser-focused on translating and creating business content, the company claims its models provide much more accurate and precise translations for enterprises.
Red Cross volunteers getting ready to deliver the mission. Photo illustration by Juan Carlos Molina Padilla/American Red Cross
The American Red Cross Multicultural Communications team announces the launch of its innovative Translator Hub. This cutting-edge platform is designed to bridge the language gap and ensure effective communication with non-English speaking audiences. By utilizing a custom machine translation system complemented by rigorous human reviews, the hub offers accurate and culturally sensitive translations for all text and materials.
The Translator Hub is a significant milestone for the American Red Cross, reflecting its commitment to inclusivity and accessibility. Users within the organization can now translate a wide range of documents, ensuring vital information reaches diverse communities. Whether it’s press releases, social media content, or educational materials, the hub ensures these messages are understood by everyone, regardless of language barriers.
The key to the hub’s success is its dual approach to translation. Initially, a custom machine translation system generates a preliminary document. This is then meticulously reviewed by human translators to guarantee accuracy, contextual relevance and cultural appropriateness. This two-step process ensures that the final output is not only linguistically correct, but also resonates with the target audience.
The translation news daily digest is my daily 'signal' to stop work and find out what's going on in the world of translation before heading back into the world at large! It provides a great overview that I could never get on my own.
susan rose (X)
შეერთებული შტატები
I receive the daily digest, some interesting and many times useful articles!
Share an article »
ProZ.com blog »
