Roblox Acquires Voice AI Moderation Startup Speechly
Digital world platform Roblox has acquired voice AI startup Speechly for an undisclosed sum. Speechly is a leader in leveraging speech recognition and natural language understanding for moderating conversations in online games and other digital spaces. Roblox will have the advantage of that tech and Speechly’s team as Roblox implements a generative AI voice assistant to build digital experiences and spaces, including handling voice moderation to catch real-time violations of its rules.
Speechly’s voice API is designed to incorporate voice interactions into games and other experiences, encouraging more natural user interactions as well as reducing harmful behavior in online communities. The low-latency speech recognition serves as a real-time response service for voice commands and speeds up analysis of conversation to both start carrying out commands before the user finishes speaking and alert moderators about issues almost at the same moment they occur. That makes it faster than the traditional method, where voice assistants don’t start processing commands until the user finishes talking.
Speechly is arguably the most sophisticated technical solution in the market with a considerable stable of intellectual property. Potential rivals like Modulate’s ToxMod and Spectrum Labs are lacking in one or more facets of Speechly’s products. The Finnish startup has raised around $7M in total from investors, including Y Combinator, Cherry Ventures, TQ Ventures, Seedcamp, SNÖ, and the founders of Airbnb, Lyft, Tinder, Fitbit, and Docker.
“We founded Speechly in 2016 with the belief that voice was the future of interaction online and our mission was to enable more delightful computer voice interactions and to empower and enhance communication between people,” Speechly CEO Hannes Heikinheimo, who previously worked on localizing Apple’s Siri voice assistant to European languages, explained in a blog post. “We are proud of the speech recognition product and solutions our team created that enabled fast, real-time and cost-efficient voice moderation helping developers reduce toxic behavior in online communities.”
Roblox can leverage Speechly’s real-time speech recognition software to moderate spoken conversations at scale across its immersive 3D environment. The acquisition aims to support new voice chat capabilities being rolled out to the platform’s 65.5 million daily active users. Roblox has a need for Speechly’s technology as it designs moderation practices to suit the demands of voice-based communication among its largely young user base. Speechly’s voice moderation algorithms will enable Roblox to monitor spoken conversations across its platform to identify and take action on any harmful behavior enabled by unsupervised voice features.
“With the addition of new voice features, including voice chat, Roblox is solving new challenges—moderating spoken language in real time. Safety and civility are foundational to Roblox. We are excited to be joining a company dedicated to safety and civility and to use our AI expertise to evolve traditional methods of moderation to meet the scale, real-time and dynamic needs of a user generated content (UGC) platform,” Heikinheimo wrote. “It’s the same focus we have had at Speechly. We share Roblox’s vision for safe and civil immersive communication and bringing more dynamic and nuanced interactivity to the platform through safe and civil voice features.”