Skip to content
IFAI

Industry

Roblox buys AI voice startup Speechly to clean up voice chat

Published Reach 50

Roblox just acquired Finnish voice tech startup Speechly to tackle toxic behavior in voice chat across its platform. Speechly builds AI tools for real-time speech moderation and transcription. The move will help Roblox protect users as it rolls out avatar-based voice calls, though financial terms remain undisclosed.

Roblox buys AI voice startup Speechly to clean up voice chat
Image: speechly.com

Text filters can easily catch bad words in chat, but once gamers jump on voice, things get messy fast.

Roblox is buying up the tech to fix that. On September 20, voice moderation startup Speechly announced it is being acquired by Roblox. Financial terms of the deal were not disclosed, and Roblox declined to comment on the details.

Cleaning up the audio feed

Helsinki-based Speechly, founded in 2016 and backed by Y Combinator, specializes in real-time voice moderation, speech-to-text transcription, and voice APIs. According to Pitchbook, the Finnish startup had raised $7.53 million from investors like Seedcamp and Cherry Ventures before the exit.

Voice chat is notoriously toxic. Speechly’s own research found that nearly 70% of gamers use voice chat, and 72% of them have experienced abuse while using it.

Roblox has been aggressively pushing deeper into voice over the last two years, shifting away from floating text bubbles to live chatter. Recently, it even introduced facial motion tracking for avatar calls to court older users aged 17 and up. But with 65.5 million daily active users, monitoring billions of spoken words live is a technical nightmare.

Industry outlet Voicebot.ai noted that Speechly’s tech stands out as more sophisticated than rivals like Modulate’s ToxMode or Spectrum Labs. Speechly also recently earned SOC 2 Type II certification, verifying its data privacy and security standards.

Why it matters

Real-time voice moderation is one of the hardest problems in online safety. Unlike text, parsing spoken audio requires handling accents, slang, sarcasm, and latency across massive multiplayer environments.

For Roblox, this acquisition is a defensive must-have. The company is actively trying to grow up with its aging player base by building experiences for teens and young adults. However, one major safety scandal involving predatory or toxic behavior over open mics could derail those ambitions and invite severe regulatory scrutiny.

We think bringing Speechly in-house is far smarter than licensing third-party tools. It lets Roblox bake safety directly into its proprietary engine at scale without paying recurring API tax.

That said, the move walks a fine line. Continuous voice monitoring inevitably sparks surveillance concerns among players, and aggressive AI filters risk killing natural banter if false positives start muting innocent users.

Sources (2)

  1. [1]
    speechly.com(opens in a new tab)Originalspeechly.com
  2. [2]

Best of AI, daily