Breaking

Modulate Raises $25 Million to Expand Voice AI Models for Fraud and Deepfake Detection

September 28, 2026 at 10:09 EDT

Security
Funding & M&A
Sales & Business

Modulate has raised $25 million to expand a voice-intelligence platform that analyzes conversations for transcription, emotion, deepfakes, AI-generated music and policy compliance in regulated enterprise settings. The Boston startup is pitching a compact, audio-native model stack as companies try to detect fraud and supervise voice agents while calls are still happening.

Sept. 28, 2026 | Modulate

$25M for real-time voice defense

A compact-model voice stack is being funded to spot deepfakes, scams, policy breaks, and troubled AI-agent calls while conversations are still live.

$25M

new funding

100+

voice models

98.9%

deepfake accuracy

$0.25/hr

API detection price

Benchmark error scale

Estimated mistakes per 1,000 calls at the equal-error point. Lower columns are better.

~11

Modulate

~21

Hiya

~26

Resemble

~31

Whispeak

How the voice stack acts

Audio signal

tone, timing, speakers

Specialized models

synthetic voice, emotion, risk

Orchestrated insight

confidence and evidence

Operational action

route, review, verify

Why teams care

Short-audio verdicts and streaming APIs can move risky calls into review before money, access or customer trust is lost.

What still needs care

Synthetic speech is not automatically fraud, and real deployments still need calibrated thresholds, privacy controls and human review paths.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.