Boston startup Modulate raises $25 million for voice AI models — TechCrunch

Boston-based U.S. startup Modulate has raised $25 million in new funding to develop its artificial intelligence-powered voice analysis platform. The company offers businesses tools for transcribing conversations, analyzing emotions, detecting deepfakes and AI-generated music, as well as monitoring compliance by voice agents in regulated industries, TechCrunch reports.

Investors and company development

The funding round was led by Future Ventures, with Hyperplane and Lakestar also participating. According to PitchBook, Modulate had raised $41 million before this round, and its valuation stood at $170 million.

The company was founded in 2017 by Mike Pappas and Carter Huffman, who met while studying in the physics department at the Massachusetts Institute of Technology. Initially, Modulate developed voice-changing technologies for the gaming industry, later focusing on a voice moderation tool. The company is now also working on detecting various types of AI-generated audio and analyzing speakers' intentions.

More current news is available on the UA.News Telegram channel Telegram.

Models for call analysis

Modulate uses more than 100 models divided into two groups. The first extracts signals from audio and identifies emotions, tone, language, and the synthetic origin of a voice. The second analyzes intentions: for example, what exactly a customer is trying to say, whether a caller is violating rules, and whether they may be trying to deceive the person they are speaking with.

According to Huffman, using smaller models does not require specialized hardware or significant computing resources. The company can train models for new capabilities, add them to the system, and engage them as needed through an orchestrator.

The startup specializes in detecting voice deepfakes and warning organizations, including call centers, about potential fraud. Modulate's technologies are also used to assess AI agents' responses to customers, monitor compliance with regulatory requirements, and track cyberattacks through voice calls. The company employs between 40 and 45 people; it plans to hire another ten specialists for model development and is expanding its capabilities for on-premises deployment of systems and running them on devices to enhance privacy.