Unpacking the State-of-the-Art in Handwritten Text Recognition
Text recognition-turned-detective uncovered a missing play by Lope de Vega. But the benefits of better handwriting recognition go beyond the history books.
SlatorCon Zurich is now SOLD OUT — See you on October 4th!
Natural Language Processing
Discover what’s new in the field of natural language processing, from ground-breaking research in academia to Big Tech’s developments.
Language AI Products Are Mushrooming as Barriers to Entry Collapse
As the cost of language AI falls, startups are repackaging GPT-4 for translation, text-to-speech, AI voice generation, and lip-synching tech, to build new consumer-grade products.
Researchers Propose New Approach to Many-to-Many Spoken Language Translation
KAIST and Google DeepMind researchers propose training a single model using Unit-to-Unit Translation to achieve seamless cross-language communication.
How GPT-4 and Text-to-Speech Make Visual Content More Accessible
The GPT-4 powered digital visual assistant from Be My Eyes could transform accessibility for the blind and low-vision community. TTS language support is key to extending its reach.
Why Sign Language Translation Poses a Puzzle for AI
Natural language processing drives increasingly sophisticated AI. But is it ignoring a whole class of natural languages? Sign language technology can help build a more accessible and inclusive world.
Slator Game Localization Report
100-page report on the fast-growing game localization vertical. Includes market size, game development, language tech, GaaS, and in-depth localization process guide.
Realistic Dialogues, No Translation: Google Unveils Dataset for Virtual Assistant Training
A group of researchers at Google found that realistic dialogues with no translation and structured context are needed to optimize virtual assistants’ performance.
Speechless Recognition: Can AI Transcribe a Language It Has Never Heard?
Breakthroughs in multilingual speech models are rapidly expanding language coverage for ASR. But what about endangered languages with little to no training data?
Translators, Interpreters Among Most ‘Vulnerable’ to Large Language Models, OpenAI Says
New research by OpenAI and University of Pennsylvania: GPT-4 and human annotators quantify the overlap between LLM capabilities and tasks for more than 1,000 jobs.
GPT-4 Launch Promises Surprising New Use Cases
OpenAI releases its much anticipated GPT-4 large language model. Here’s a short primer on its language, problem-solving, and multimodal capabilities for real-world applications.
Microsoft’s New Voice Replicator VALL-E X Features Translation, Accent Control
We knew Microsoft's text-to-speech model VALL-E can mimic a voice with little training data. Now VALL-E X adds translation, foreign accent control, emotion maintenance, and more.
The Slator 2023 Language AI 50 Under 50 List
The Slator Language AI 50 Under 50 List puts the spotlight on fifty of the newest players in the language AI and technology landscape, including AI dubbing, AI writing, and voice AI.
Google Sheds More Light on Its 1,000+ Languages Universal Speech Model
As the race in all things language AI heats up, Google discloses more details about its Universal Speech Model that the search giant claims supports over 1,000 languages.
How Does Large Language Model BLOOM Perform for Machine Translation?
Researchers test BLOOM’s capabilities for producing good quality machine translation. They find that training the large language model makes a big difference for all language pairs.
Microsoft Says Large Language Models Show ‘Competitive Machine Translation Quality’
Researchers at Microsoft conduct experiments comparing large language model GPT to other MT systems. One GPT stands out with high-resource languages. None do with low-resource languages.
How Large Language Models Can Get Better at Machine Translation of Rare Words
Meta AI researchers experimented with dictionary data prompting on known LLMs in order to improve MT. Results look promising for rare words and domain transfer.
Revisiting Controlled Language for Better Machine Translation Quality
Researchers revisit controlled language as a means of improving MT quality. A new methodology based on stylistic quality is said to eliminate repeated fine tuning.
Spanish NLP Data Labeling Platform Argilla Raises USD 1.6M in Seed Funding
The open-source platform helps users build and refine datasets without hand-labeling training examples. Funding round co-led by Caixa Capital Risc and Zetta Venture Partners.
Maveria Advisory and Slator Advise Repustate on its Sale to Sprout Social
Maveria Advisory and Slator acted as the joint exclusive sell-side advisors to Repustate, a sentiment analysis and natural language processing company, in its sale to Sprout Social, the leading social media management software.
We Prompted ChatGPT to be a Translation Manager
How we used a script from ChatGPT, an avatar from Synthesia, and captions from Descript to produce an AI-generated video.
We Asked ChatGPT Questions About the Language Industry…It Had a Lot to Say
OpenAI disrupts the AI and LLM space once again with ChatGPT, reaching over one million users in just five days. See what the app told Slator about the language industry.
Your information will not be shared with third parties. No Spam.