Modulate Cuts AI Voice Detection Time to 1.6 Seconds, Adds Twilio Ventures Backing

Modulate says its latest models can identify an AI-generated voice in a median of 1.6 seconds, down from 2.25 seconds, while keeping detection accuracy above 99.1% on its real-world evaluation of AI and human calls. The Boston-area company announced the results alongside new investments from Twilio Ventures and Kayak co-founder Paul English, money it will put toward research, product development, and broader deployment of tools that detect synthetic callers and supervise AI voice agents.

The company has not disclosed the size of the additional investments. The backing follows a $25 million financing announced in September, led by Future Ventures with participation from Hyperplane and Lakestar. Alongside the funding, Modulate is adding more than a dozen audio-native signals to its Velma platform, an offering designed to help enterprises understand both who — or what — is calling and how an AI-led conversation is unfolding.

The stakes are straightforward. Voice automation is spreading through contact centers, and a synthetic caller might be a helpful assistant booking an appointment or a cloned voice attempting impersonation. Detecting generated speech supplies one piece of evidence; deciding what to do with a call requires a broader view of its behavior and context.

Why 1.6 Seconds Matters

Modulate frames the reduction in detection time as an operational threshold. At a 1.6-second median, the company says, businesses can identify synthetic speech early enough to route a call, flag it for review, or apply additional verification while the conversation is still underway. The system examines the underlying audio directly and does not depend on watermarks inserted by individual voice-generation providers — including newer speech-to-speech and full-duplex models that generate audio in both directions simultaneously.

The shift from 2.25 to 2.25 to 1.6 seconds — corrected: from 2.25 seconds to 1.6 seconds — represents roughly a 29% reduction in the company's reported median. It is a detection-time measure, not a guarantee that every call is classified within that interval or that a full security workflow finishes that quickly. The announcement does not provide the evaluation sample size or a breakdown of false positives and false negatives behind the greater-than-99.1% accuracy figure.

"Call operators – human or AI agent – must be able to recognize what's happening in those conversations and respond appropriately while the interaction is still ongoing," said Carter Huffman, CEO and co-founder of Modulate.

For contact centers, the practical benefit is intervention before a conversation ends. A warning delivered after an account change or a sensitive disclosure has already occurred serves a different purpose from a signal available during the interaction itself.

How Does Velma's Architecture Work?

Velma uses what Modulate calls an Ensemble Listening Model, or ELM — an architecture that coordinates specialized models rather than relying on a single general-purpose model to interpret a transcript. The company says the system brings together more than 100 models covering signals such as emotion, tone, intent, cadence, synthetic speech, and conversational behavior.

An orchestration layer combines lower-level signals into higher-level assessments while preserving links to the underlying evidence, according to Modulate's product description. Time-stamped events can indicate where a conversation changed, giving an operations team something more specific to review than an unexplained call-level risk score.

The design addresses a known limitation of transcript analysis. A transcript captures words but does not preserve every pause, interruption, shift in emphasis, or delivery pattern. Modulate's approach adds acoustic information so developers can examine how a conversation proceeds alongside what participants say.

The company's deepfake-detection documentation illustrates the distinction at the API level. The system performs acoustic classification of audio segments, returning time boundaries, verdicts, and confidence scores, and supports both recorded files and live streams. A detection confidence score is evidence for a classification, Modulate notes — not proof of a caller's identity or intentions.

Is Synthetic Speech the Same as Fraud?

Modulate explicitly separates the two questions in its voice-fraud screening guide. Deepfake detection evaluates whether a voice is synthetic; conversation analysis evaluates behavior. A legitimate AI assistant can use generated speech, and a human caller can attempt social engineering without any voice synthesis at all.

The distinction also matters for callers using assistive speech technology. A synthetic-voice flag alone is an insufficient reason to treat a caller as malicious. In the new release, Modulate says Velma can combine that flag with signals associated with intent, manipulation, tone, and conversational behavior to support more informed handling.

The practical goal is a graduated response:

• Ordinary routing for permitted automation
• Additional verification where a transaction warrants it
• Escalation when the conversation presents suspicious behavior

The detection system supplies evidence. The organization's policies determine the response.

What About Monitoring AI Agents Themselves?

The new Velma signals also target a less dramatic but commercially significant problem: an AI agent can sound fluent while delivering a poor service experience. Modulate says the additions help evaluate response timing, audio quality, and conversational misunderstandings that transcript-only analytics can miss.

On its AI guardrails page, the company describes Velma as monitoring participants rather than acting as the conversational agent itself. Customers can configure monitoring around operating procedures and expected behavior, then send signals into existing systems through APIs and webhooks. That separation — the system doing the talking apart from the system assessing the interaction — gives teams a way to investigate where an agent is causing frustration, going outside policy, or failing to respond appropriately, and to use those findings to improve deployment.

What Does the Twilio Connection Signal?

Twilio Ventures focuses on AI, developer tools, and customer-engagement businesses, which places audio intelligence squarely within its investment remit. The relationship predates the check: a May 2025 announcement described compatibility for Modulate's VoiceVault and ToxMod products, including Twilio Media Streams-compatible APIs for bringing live call audio into its systems.

The new investment builds on that history, though the announcement does not establish a new exclusive distribution agreement. Modulate says the capital will support research, product development, developer adoption, partnerships, and broader deployment options — the same priorities that anchored September's expansion beyond the company's earlier voice-safety applications.

With this update, Modulate is pulling detection and supervision closer together: recognizing generated speech, examining the surrounding behavior, and assessing whether an enterprise's own voice agents are doing the job they were built to do.