AI translation startup Palabra secures investment from Reddit co-founder Alexis Ohanian’s venture firm to expand its global language solutions.
An AI-powered speech translation engine is being developed by a startup called Palabra AI, which is currently addressing one of the more challenging aspects of teaching large language models (LLMs) to comprehend multiple languages.
Currently, LLMs facilitate text conversion from one language to another; however, Alexis Ohanian, the co-founder of Reddit, asserts that speech translation is not as straightforward.
“Artificial intelligence can produce content and translate text.” However, [Speech] Translation is a distinctive issue due to the necessity of real-time language switching and the requirement that the voice sound human, according to Ohanian.
Ohanian and his venture firm Seven Seven Six (776) think Palabra AI may solve this issue. This is why the company is steering an $8.4 million pre-seed round into the startup.
Creator Ventures and individual investors, such as Instacart co-founder Max Mullen, former a16z partner Anne Lee Skates, former DeepMind Head of Product at Mehdi Ghissassi, and Namat Bahram, also participated in the round.
Seven seven six’s The firm was also motivated to invest by Palabra’s combination of product execution and team expertise, as Ohanian observed.
“The translation layer functions seamlessly with Palabra.” The company’s robust AI research team conducts high-quality work in the field of speech.
Additionally, he stated that the startup has made excellent decisions regarding the quality of its output and product design.
In 2023, Alexander Kabakov and Artem Kukharenko established Palabra. Kukharenko, a former machine learning engineer at Samsung, stated that he has resided as a digital nomad in various countries and has encountered language barriers.
That is why he sought to integrate his machine learning expertise to address the issue of real-time translation.
“Numerous other organizations have attempted to resolve the translation issue.” However, the latency number increased, and the translation did not feel real-time when they incorporated various technologies, such as speech-to-text and text-to-speech APIs, during the translation process.
He informed TechCrunch during a call that Palabra has enabled us to reduce the latency to 800ms, ensuring that our translations are real-time and seamless.
Kukharenko further stated that the organization has established a personalized data infrastructure to facilitate incorporating a new language into the startup’s platform within a matter of weeks.
He stated that Palabra places a human interpreter at the end of this pipeline to verify the quality of the output. The startup said its algorithm also considers various scenarios, such as interruptions and chaotic environments.
Palabra AI offers translation products that are designed for both consumers and businesses. The organization provides a desktop application compatible with numerous video conferencing applications, including Google Meet, Zoom, Discord, Slack, and Microsoft Teams. This application is available for both Mac and Windows operating systems.
The app can support translation in more than 30 languages, allowing you to listen to an individual speak their native language in the language of your choosing. This could be beneficial for conversations that involve participants of different nationalities.
Palabra’s website indicates that its desktop application will shortly incorporate native translation capabilities for content websites such as Twitch, Vimeo, Netflix, and YouTube.
Each month, you are granted 30 minutes of transition-free time. To increase the limit, purchasing programs that commence at $25 per month for 60 minutes of translation across multiple applications is possible.
The startup provides APIs and SDKs to enterprises to integrate translation into their products.
Its technology is propelling video platforms like Agora to provide live multilingual broadcasts. Palabra’s tools are also being employed by language-service providers such as GIS Group in conjunction with human interpreters.
The company said numerous event organizers also utilize its tech to provide multilingual streams.
Competition in the translation market is fierce for Palabra. On the consumer side, startups such as EzDubs, which Y Combinator funds, are developing applications compatible with in-person or phone conversations.
Google also implemented real-time translation for video conversations on Meet earlier this year. On the commercial side, startups such as Dubai-based Camb.AI are developing translation technology to broadcast live events in multiple languages.
Palabra is developing a new streaming prediction model to differentiate itself, potentially reducing latency significantly. Additionally, it is in the process of providing translation support for more than 10,000 concurrent audio transmissions.
