Modulate raises $25M for audio-native AI models

Voice AI startup Modulate has raised $25 million to expand access to its audio-native models for developers. The company's Velma platform uses an Ensemble Listening Model architecture that selects from over 100 specialized audio models per task, claiming up to 1,000 times greater efficiency than a single large model. Its tools detect deepfakes, flag fraud, monitor voice agent performance, and screen for harassment. More than 10 million hours of audio run through its models monthly. Future Ventures led the round, with Hyperplane and Lakestar participating. Total funding now stands at $60 million.