Pau Todó’s Post

🌍 Ok Google, Translate in Awadhi! 🌍 Exciting news from Google Translate! 📢 Adding support for 110 new languages, nearly doubling the current offering of 133 languages. This major update is powered by Google’s PaLM 2 AI, which is a language-learning genius. 🧠💡 🤩 Now, Google Translate can understand languages closely related to others, like Awadhi and Marwari (close to Hindi) and various French Creoles (close to Seychellois and Mauritian Creole). 🎉 And guess what? Cantonese, one of the most requested languages, is finally here! 🥳 Looks like Google Translate just became our new polyglot best friend. 🌐🤖 #Google #Googletranslate #Technology #AI #language

Google Translate has added Inuktut, the first Indigenous language in Canada available on the platform. This initiative, developed in collaboration with Inuit Tapiriit Kanatami, supports language revitalization and provides digital accessibility for Inuit communities. Read more: https://buff.ly/3YnTEh6 #LanguageTechnology #Inuktut #IndigenousLanguages #AI #Reconciliation #MultiLingualMedia

MTPE – What they promise: - A faster translation - A cheaper translation - A high-quality translation And all this would be true if the reviewer simply had to scan the text and patch up a few obvious mistranslations or misunderstandings. This is how many people see the process. But it's not the reality. The reality is that machine-translated texts stick too closely to the source syntax. This means that they sound strange and unnatural. When faced with these kinds of texts, reviewers have a choice. They can either ignore the dodgy syntax and stick to what they are paid to do, ignoring all problems beyond the most glaring translation mistakes. Or they can take their time to fix up the text and produce a good translation, sometimes rewriting entire sections from scratch, essentially doing a proper job at a fraction of the cost. So the client may get a faster and a cheaper translation, but they won't get a quality one. And if they do, it's because the work has been slower and more expensive not for them, but for the linguist. #mtpe #translation #machinetranslationpostediting #localization #language #artificialintelligence #ai #chatgpt #humancreativity #humandesign

MTPE – What they promise: - A faster translation - A cheaper translation - A high-quality translation And all this would be true if the reviewer simply had to scan the text and patch up a few obvious mistranslations or misunderstandings. This is how many people see the process. But it's not the reality. The reality is that machine-translated texts stick too closely to the source syntax. This means that they sound strange and unnatural. When faced with these kinds of texts, reviewers have a choice. They can either ignore the dodgy syntax and stick to what they are paid to do, ignoring all problems beyond the most glaring translation mistakes. Or they can take their time to fix up the text and produce a good translation, sometimes rewriting entire sections from scratch, essentially doing a proper job at a fraction of the cost. So the client may get a faster and a cheaper translation, but they won't get a quality one. And if they do, it's because the work has been slower and more expensive not for them, but for the linguist. #mtpe #translation #machinetranslationpostediting #localization #language #artificialintelligence #ai #chatgpt #humancreativity #humandesign

我識講廣東話...! (Ngo5 sik1 gong2 gwong2 dung1 waa2!) 😊 Google Translate adds Cantonese and 109 other languages with AI advancements • Google Translate has expanded its language support to include Cantonese, along with 109 other languages, through advancements in generative Artificial Intelligence (AI). This marks the largest-ever expansion of the app's language capabilities, made possible by Google's PaLM 2 large language model. • Cantonese, widely spoken in Chinese communities worldwide and in Hong Kong, has been a highly requested language for translation. However, due to its frequent overlap with Mandarin in written form, finding data and training models for Cantonese has been challenging. • Google aims to support the 1,000 most spoken languages globally and has also added Punjabi and several African languages, including Fon, Kikongo, Luo, Ga, Swati, Venda, and Wolof. The company plans to continue expanding its language offerings and improving translation accuracy by collaborating with expert linguists and native speakers. Good Read: Article about the topic; Full Story https://lnkd.in/gtwRYyGE

Maybe one day an article like this would be written about a company that collaborates with local artists and writers to produce these books, and a home grown machine translation company, rather than Chat GPT and Google Translate 🤔 I think of the children's stories I heard growing up in Ethiopia, and how specific they were to the context. And when my elders tell me stories about various villages in Eritrea, they are super super specific. Why not preserve that and pass it on, and give people the ability to do that for each of their cultures and contexts. My 2 cents? We should empower local artists and writers and story tellers and build tools to do that, rather than using technology built on stolen artists' work and exploited labor right from the continent. Like what Keoni Mahelona & Te Hiku Media did: which you can read about in the AI colonialism series, https://lnkd.in/eizSEymD

Analytics India Magazine writes "AI4Bharat Introduces BhasaAnuvaad, Speech Translation Dataset of 13 Indian Languages with 44,400 Hours of Data - AI4Bharat also developed Indic-Spontaneous-Synth, a synthetic evaluation set to highlight how current models, though effective on datasets like FLEURS, tend to underperform in realistic, spontaneous language translation scenarios, underscoring the need for more robust datasets. " https://lnkd.in/eT6wXx8k. #ai4bharat #indianlanguages #speechtranslaten #trainingdata #artificialintelligence #analyticsindiamagazine

Cantonese's inclusion in Google Translate has indeed been a long-awaited development. The challenge lay in the fact that Cantonese often overlaps with Mandarin in writing, making it tricky to find sufficient data and train accurate models. However, thanks to advances in generative artificial intelligence (GenAI), Google has now added Cantonese to its translation service, marking its largest-ever expansion with support for 110 new languages. 🌐🗣️ The accuracy of the Cantonese translation in Google Translate has improved significantly, but it’s essential to recognise that machine translation still has limitations. While it can handle common phrases and sentences reasonably well, nuances, context, and idiomatic expressions may occasionally be lost. For critical or complex translations, it’s advisable to consult a human translator

With Duolingo, we know AI can teach us new languages, but can it revive endangered ones? USC Viterbi School of Engineering's Jared Coleman and Professor Bhaskar Krishnamachari are leveraging AI to breathe new life into Owens Valley Paiute, an endangered Indigenous language. Their innovative approach combines rule-based translators with large language models, providing a practical tool for real-world use. “The tool is smart enough, given a few hints, to be able to do a lot of the translation on its own,” says Krishnamachari. This could be a game-changer for preserving endangered languages like Hawaiian, Welsh, and Cherokee. What other endangered languages do you think AI could help save?

Exciting news for language lovers! Google Translate just added 110 new languages, powered by AI! This means smoother communication for millions worldwide. Read more about the new languages and how it works 👉 https://ow.ly/6QFV50Sxehn #LanguageLovers #GoogleTranslate #AICommunication