A startup called Palabra AI<\/a>, which develops an AI-powered speech translation engine, is tackling one of the more difficult aspects of teaching large language models (LLMs) to understand multiple languages.<\/p>\n Today, LLMs make it easier to convert text from one language to another, but speech translation is not as simple, according to Reddit co-founder Alexis Ohanian.<\/p>\n \u201cAI can generate content and translate text. But [Speech] Translation is a unique problem because it requires real-time language switching, and the voice also needs to sound human,\u201d Ohanian said.<\/p>\n Ohanian and his venture firm Seven Seven Six<\/a> (776) believe that Palabra AI<\/a> could be an answer to that problem. That\u2019s why the firm is leading an $8.4 million pre-seed round into the startup. The round also saw participation from Creator Ventures along with individual investors, including Instacart co-founder Max Mullen, former a16z partner Anne Lee Skates, former DeepMind Head of Product at Mehdi Ghissassi, and Namat Bahram.<\/p>\n Seven Seven Six\u2019s Ohanian also noted that Palabra\u2019s combination of product execution and team expertise pushed the firm to invest.<\/p>\n \u201cWith Palabra, the translation layer works very smoothly. The company has a strong AI research team that does high-quality work around speech. Plus, the startup has made great choices in product design and quality of output,\u201d he added.<\/p>\n Palabra was founded by Artem Kukharenko and Alexander Kabakov in 2023. Kukharenko, a former machine learning engineer at Samsung, said that he has lived in various countries as a digital nomad and faced problems with languages. That\u2019s why he wanted to combine his machine learning experience to solve for real-time translation.<\/p>\n Techcrunch event<\/p>\n \n\t\t\t\t\t\t\t\t\tSan Francisco<\/span> \u201cMany other companies have tried to solve the translation problem. But when they combined different pieces of tech, including speech-to-text and text-to-speech APIs while translating, the latency number increased, and the translation didn\u2019t feel real-time. With Palabra, we have been able to bring down the latency to 800ms, making our translation seamless and real-time,\u201d he told TechCrunch over a call.<\/p>\n Kukharenko added that the company has built a custom data pipeline to make it easy for the startup to add support for a new language within weeks. He said that at the end of this pipeline, Palabra places a human interpreter who checks for quality of the output. The startup said that its algorithm also accounts for different scenarios like noisy environments and interruptions.<\/p>\n Palabra AI has both consumer and business-facing translation products. The company offers a desktop app \u2014 for both Mac and Windows \u2014 that works with popular video calling apps, including Google Meet, Zoom, Discord, Slack, and Microsoft Teams. The app can support translation in over 30 languages, which means you can listen to someone speaking in their native tongue in your choice of language. This could be handy for calls that involve participants of various nationalities.<\/p>\n On its website, Palabra says that its desktop app will soon support native translation features for content sites like YouTube, Netflix, Twitch, and Vimeo. <\/p>\n You receive 30 minutes of translation free each month. For a higher limit, you can pay for plans starting at $25 per month for 60 minutes of translation across apps.<\/p>\n The startup provides APIs and SDKs to enterprises to integrate translation into their products. <\/p>\n Currently, its technology is powering video platforms such as Agora for live multilingual streams. Plus, language-service providers like GIS Group are using Palabra\u2019s tools alongside human interpreters. The company said multiple event organizers also utilize its tech to provide multilingual streams.<\/p>\n Palabra has tough competition in the translation market. On the consumer end, startups like Y Combinator-backed EzDubs<\/a> are building apps that work for in-person or call conversations. Earlier this year, Google also launched real-time translation for video calls on Meet<\/a>. On the business end of things, startups like Dubai-based Camb.AI<\/a> are building translation tech to broadcast live events in multiple languages.<\/p>\n To stand out, Palabra is currently working on a new streaming prediction model, which could result in a drastic drop in latency. It is also working on supporting translation for over 10,000 simultaneous audio streams.<\/p>\n We\u2019re always looking to evolve, and by providing some insight into your perspective and feedback into TechCrunch and our coverage and events, you can help us! Fill out\u00a0<\/em>this survey<\/a><\/em>\u00a0to let us know how we\u2019re doing and get the chance to win a prize in return!<\/em><\/p>\n<\/div>\n
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 27-29, 2025<\/span>\n\t\t\t\t\t\t\t<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2025\/08\/App-demoscreen-3.jpg?w=680\")