SlatorPod is the weekly language industry podcast where we discuss the most important news and trends in translation, localization, and language technology. Brought to you by Slator.com.
#141 DeepL Aborts Write Launch, German Translation Employment Dips, Media Loc M&A
DeepL launches then pulls new AI-based writing tool, translator and interpreter employment declines in Germany, job index dips for fourth time in 2022, Blu Digital Group expands further into media localization, and Deluxe buys Post Haste Digital.
#140 How the Brain Processes Language, With MIT Neuroscientist Ev Fedorenko
Joining SlatorPod this week is Ev Fedorenko, Associate Professor of Neuroscience at the Department of Brain and Cognitive Sciences, MIT. Ev also runs EvLab, an MIT language lab that discovers how the human brain creates language.
Ev talks about the different hypotheses concerning the origin of language and how it has likely been a gradual evolution. She shares a number of intriguing research findings on the relationship between language and abstract representations of structure (i.e., complex thought).
Ev discusses how language processing takes place and how we can use brain imaging to compare language with other non-linguistic tasks, such as solving math problems and composing music. She questions whether specific languages can be objectively easy or difficult to learn as an adult.
She also considers what sets polyglots apart when it comes to learning languages and some of the generalizations made in research. Ev talks about how language processing in machines like GPT-3 compares to that in humans. She argues that it would be more fruitful to build language systems that are structured similar to the human brain.
Ev concludes with the collaboration between academia and the booming field of applied AI, despite different goals. She touches on the MIT Quest for Intelligence, which brings together scientists and engineers to build better human-like models for the benefit of society.
#139 AI Dubbing Finds Traction, Translation Buyer’s Guide Published
In this week’s episode, Florian is joined by Anna Wyndham, Senior Research Analyst at Slator. The two discuss the language industry news of the week, with a recap of the Nordic Translation Industry Forum held in Sweden last week. The NTIF covered a broad range of presentations, including a talk from Ikea about how they are building a custom machine translation (MT) engine from scratch.
In startup funding, AI dubbing startup, NeuralGarage, raised USD 1.45m in a seed round led by Exfinity Ventures. While this is the latest in a series of Indian machine dubbing startups to have raised funds, NeuralGarage focuses on visuals. This includes changing the lip and jaw movements of the person on the screen to match the target speech.
In other AI dubbing news, Papercup won a two-year translation and dubbing contract with Bloomberg, signaling greater interest in synthetic voices. The deal will see Papercup mainly localizing Spanish global news coverage, financial market analysis, and documentaries for Latin American and US audiences.
The duo talk about the latest machine translation research carried out by Google examining the sentence-level translation capabilities of their Pathways Language Model (PaLM). The researchers found the translations via PaLM, a large language model, to be more creative and very fluent — but still lagging behind state-of-the-art, supervised MT.
In Belgium, two associations have released a best practices guide for translation services procurement in the public sector. The Belgian Quality Translation Association and the Belgian Chamber of Translators and Interpreters delve into the tasks and tools used during a typical translation production cycle, among other things.
#138 DeepL’s Unicorn Round, $15 Million Translation Client, Bank Builds MT In-House
Florian and Esther discuss the language industry news of the week, with German machine translation (MT) company DeepL said to be in the final stretch of closing a major investment round led by Silicon Valley venture firm, IVP. This will allow DeepL to expand further into the enterprise, as it ramps up hiring of customer success, account management, and related enterprise-focused roles.
The duo talk about how BNP Paribas, Europe’s second-largest bank, has been building its MT engine in-house over the past five years — instead of using publicly available MT tools, such as Google Translate or DeepL.
In M&A news, LanguageLine, the language services sub-unit of French call center giant Teleperformance, acquired UK-based LSP Capita Translation and Interpreting (Capita TI) for an undisclosed amount. Capita TI’s parent company, Capita plc., has been trying to sell off the language services unit and eight other non-core businesses since early 2020.
Down Under, Straker Translations has announced the renewal of its contract with IBM for another three years. The New Zealand-based LSP originally won the two-year contract in 2020, which has already contributed USD 14.8m to its FY 2022 revenue. IBM previously had to use over 100 LSPs to fulfill its translation needs.
The pod rounds off with news of Descript raising a USD 50m series C round led by the OpenAI Startup Fund, following a series B round in January 2021. The audio and video editing platform has raised USD 100m to date.
#137 Intento CEO Konstantin Savenkov on Best-Fit MT, Google Translation Hub, Generative AI
Konstantin Savenkov, CEO and co-founder of Intento, joins SlatorPod to discuss how they enable enterprise clients to choose best-fit machine translation (MT) services and technology.
Konstantin outlines Intento’s services and product offerings, which include an MT Hub for localization and office productivity, an MT studio, as well as MT evaluation and maintenance. He talks about why there is a lack of consolidation when it comes to MT providers, translation management systems (TMS), and translation productivity technology (aka CAT).
The CEO discusses the highlights from Intento’s State of Machine Translation 2022 report, which evaluates 31 MT engines, their best-performing language pairs, and pricing comparison. He shares his views on Google’s Translation Hub and how it raises more interest from investors evaluating the language industry.
Konstantin gives his take on generative AI — such as Stability AI’s Stable Diffusion and Open AI’s Whisper — and how this will not only impact the content generation and authoring industry but the machine translation space as well. He presents the idea of source quality improvement, where the source text is generalized before machine translation to make it a much more post-editable text.
Konstantin concludes the podcast with Intento’s roadmap for 2023, which includes connecting to more enterprises and adding more automated post-editing features.
#136 Deepgram: From Dark Matter to Deep Learning Speech API With Scott Stephenson
Scott Stephenson, CEO of Deepgram, joins SlatorPod to talk about his unique journey to co-founding the deep tech, automatic speech recognition (ASR) company and raising over USD 50m in funding.
Scott recalls how his experience working with dark matter detectors as a particle physicist in China led to him becoming a deep-learning entrepreneur. He discusses some challenges in solving ASR; from labeling data for machine learning to formulating and executing an effective go-to-market strategy.
The CEO gives his thoughts on Whisper, OpenAI’s open-source ASR model, and how it may actually grow the total addressable market for voice AI companies. He shares the difficulties when it comes to translating a transcript versus translating straight from audio into another language.
Scott gives his advice on how to build a successful AI company and appeal to investors. The pod rounds off with Deepgram’s roadmap for the next year, with text to speech, voice cloning, real-time translation, and sentiment analysis being potential step changes in their growth trajectory.
First up, Florian and Esther catch up on the language industry news from the past month, with Google announcing the launch of Translation Hub, its enterprise-scale document translation service.
Esther discusses some of the language highlights from Netflix’s third-quarter earnings call, including the titles of some of the best-performing non-English content. Meanwhile, Zoo Digital’s share price was at a near all-time high as they weighed in at an almost USD 170m market cap.
The duo also talk about funding, where multilingual AI writer Jasper announced it had raised USD 125m in its unicorn-making series A, which valued the startup at USD 1.5bn. And, after a dip in the Slator Language Industry Job Index in September, the LIJI defied expectations of a slowdown as it reached an all-time high in November.
Great coverage about language industry innovation
Florian and Esther are excellent conversation partners and truly know their field. They are curious and engage their guests in a natural, fun and fact finding way.
Informational and Fun!
Florian and Esther do a great job of sharing and analyzing the latest localization industry news and interviewing a variety of loc industry leaders. Well done!