Modulate

HQ: United States
Founded Year: 2017
Company Type: Privately Held
Category: Conv Intelligence
This is CXSpy's primary classification of the Modulate software. The software may fit other categories as well.
Type: platform
CXSpy classifies CX software in one of two types (a platform or a tool). This is our view of the Modulate software.

Description:Modulate is a voice intelligence company based in the Boston area that builds AI models which listen to raw audio instead of reading transcripts. MIT alumni Carter Huffman, now chief executive, and Mike Pappas founded it in 2017. The company first became known for ToxMod, a voice chat moderation product used in games such as Call of Duty, Grand Theft Auto Online, and Rec Room. Its current flagship is Velma, a platform built on what Modulate calls an Ensemble Listening Model, which picks from more than 100 small specialized audio models and blends their results. Velma reads emotion, tone, intent, accents, and synthetic speech, then combines those signals to flag fraud attempts, deepfake callers, unhappy customers, policy violations, and AI voice agent failures, in some cases while the call is still live. For contact centers it sells fraud prevention, AI agent guardrails, and QA and coaching use cases, with integrations for Five9, Genesys, Zendesk, Zoom, and Microsoft Teams. Developers can also buy its transcription, deepfake detection, and redaction models through a pay-as-you-go API. In September 2026 Modulate raised $25 million led by Future Ventures, bringing total funding to $60 million.

Modulate Conversational Intelligence Software

About Modulate

Modulate started in 2017 around a question its founders still use in their pitch: what if computers could understand a conversation itself, and not only the words in it? Carter Huffman and Mike Pappas met at MIT and built the company in the Boston area, where it is still based.

Its first big product was ToxMod, which listens to voice chat in online games and flags harassment and other harmful behavior for moderators. Activision uses it in Call of Duty, Rockstar Games in Grand Theft Auto Online, and Rec Room in its social platform.

The work on games pushed Modulate toward a design that reads audio directly. Velma, its current platform, runs on an Ensemble Listening Model that chooses from more than 100 specialized audio models for each task. The company says this is up to 1,000 times more efficient than sending the same audio to one large model. Velma picks up emotion, tone, accent, intent, and signs of a synthetic voice, and those signals can be combined into higher level events such as a vishing attempt, a customer about to churn, or an AI voice agent going off script.

That has made Modulate a contact center vendor as well as a trust and safety one. It markets fraud and deepfake detection, guardrails for AI agents, QA and coaching, and agent welfare monitoring, and it lists Five9, Genesys, Zendesk, Zoom, and Microsoft Teams among its integrations. On September 28, 2026 it announced $25 million led by Future Ventures, with Hyperplane and Lakestar, for $60 million raised in total.

Modulate Features

1. Velma platform for real-time voice intelligence across contact center and other calls

2. Ensemble Listening Model that combines more than 100 specialized audio models

3. Deepfake and synthetic voice detection for fraud and vishing prevention

4. Guardrails that monitor human and AI voice agents for risky or failed interactions

5. Emotion, accent, language, and audio event detection

6. English and multilingual speech-to-text with diarization

7. PII and PHI redaction from audio and transcripts

8. ToxMod voice moderation for games and social platforms

9. Integrations with Five9, Genesys, Zendesk, Zoom, Microsoft Teams, SIP, and Slack

Modulate Pricing

Modulate publishes self-serve API prices billed per hour of audio. English Fast speech-to-text costs $0.025 per hour in batch and $0.05 per hour streaming, and multilingual speech-to-text costs $0.03 per hour in batch and $0.06 streaming, with optional emotion, accent, and PII tagging add-ons. Deepfake and synthetic voice detection costs $0.25 per hour with a $0.05 minimum per call, emotion detection $0.02 per hour, and music versus speech detection $0.01 per hour. The Velma enterprise platform and ToxMod are sold through sales on custom terms.