Modulate raises $25M to advance deepfake detection with audio-native AI platform

Modulate, a voice intelligence company founded by MIT alumni, has closed a $25 million funding round to expand its deepfake detection and audio analysis platform. The round was led by Future Ventures, with Hyperplane and Lakestar also participating.

The fresh capital brings Modulate’s total funding to $60 million, a notable jump from its previous $41 million raised at a valuation of roughly $170 million.

What Modulate actually does

At the core of Modulate’s business is Velma, an audio-native AI platform that doesn’t just transcribe speech. It dissects it. The system deploys more than 100 specialized models to analyze raw audio across multiple dimensions: emotion, tone, intent, whether the voice is synthetic, and the subtler conversational patterns that humans pick up instinctively but machines have historically struggled with.

The company claims Velma delivers up to 2x the accuracy and 7x fewer false positives compared to traditional large language models applied to audio tasks. Those aren’t trivial improvements when you’re talking about flagging fraudulent calls in real time. A false positive in a call center means a legitimate customer gets blocked.

Modulate currently processes more than 10 million hours of audio every month, with a lifetime total exceeding 600 million hours.

The deepfake problem keeps getting worse

Modulate ranks first on the Hugging Face Open ASR Leaderboard for transcription as of mid-2026, and its deepfake speech detection achieves a 98.9% accuracy rate with a 1.1% equal error rate. That last metric matters because it represents the point where false acceptances and false rejections are balanced. A 1.1% equal error rate is competitive enough to be deployed in production environments where mistakes carry real financial consequences.

The platform handles both real-time and batch processing, which means it can screen live calls as they happen or sift through recorded audio archives after the fact. Use cases span call centers, healthcare services, gaming platforms, social media, and any industry where voice-based interactions need to be trusted or regulated.

On the pricing front, Modulate’s deepfake detection API runs $0.25 per hour of audio analyzed, while its transcription API costs $0.03 per hour for batch processing.

From voice skins to voice security

Modulate was co-founded in 2017 by Carter Huffman and Mike Pappas, both MIT graduates. The company’s original product was voice modulation technology, essentially voice skins that let users change how they sounded in real time.

The company is based in Boston and has been building its reputation in the gaming and social platform moderation space before expanding into financial services, healthcare, and enterprise compliance.

What the funding signals

Future Ventures leading this round is notable. The firm, co-founded by Steve Jurvetson, tends to back companies working on foundational technology shifts rather than incremental improvements.

For regulated industries like healthcare and financial services, the compliance angle is particularly compelling. Organizations in those sectors face increasing pressure to verify the identity of callers and ensure that recorded interactions are authentic. A platform that can handle detection, transcription, and compliance monitoring in a single pipeline has obvious appeal over stitching together multiple point solutions.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.