Modulate
Audio analysis models and APIs that identify emotion, intent, synthetic speech and risky behavior in conversations, helping developers monitor voice agents, detect fraud and moderate voice chat.
Product
Audio signals that trigger useful action
Modulate’s Velma analyzes audio directly, using an ensemble of specialized models to identify tone, emotion, intent, synthetic speech and conversational behavior. These signals combine into events such as suspected fraud, harassment, customer frustration or an AI agent failing to follow policy. Developers can use real-time results to intervene during a call, or examine recorded conversations afterward. The platform also offers transcription, deepfake detection and sensitive-data redaction APIs.
Customers define the behaviors they want to detect, including through uploaded procedures and playbooks, then route flagged moments into case management, risk or coaching systems through APIs and webhooks. The company has established deployment experience in voice moderation: its published ToxMod case studies describe use by Ubisoft, Activision and Rockstar Games. Its broader developer platform serves voice-agent supervision, customer experience and security applications. The September announcement reports more than 10 million audio hours analyzed monthly; additional APIs, SDKs and deployment options are planned.
Funding history
1 round- $25M