Modulate raises $25 million to sell voice-deepfake detection at 25 cents an hour

Modulate charges $0.25 to screen an hour of audio for synthetic speech and $0.03 to transcribe an hour in batch ^[9]. Detection is priced at about 8.3 times transcription ^[2]. That spread is the clearest evidence in the report that anyone treats voice-deepfake screening as a product in its own right. The company says its detector scores 98.9% accuracy with a 1.1% equal error rate, the point where false acceptances and false rejections balance ^[8]. It also claims up to twice the accuracy and seven times fewer false positives than general large language models applied to audio ^[6].

The volume figure puts a rough ceiling on the business. At more than 10 million hours a month ^[7], billing every hour at the detection rate would bring in about $2.5 million a month, or $30 million a year ^[3]. Billing every hour at the transcription rate would bring in about $3.6 million a year ^[4]. Neither figure is revenue. The company built its reputation moderating gaming and social platforms before selling into financial services, healthcare and enterprise compliance ^[10], and the report does not include revenue, a valuation for this round, or a split of those hours by product. The new $25 million equals the list price of 100 million hours of detection, or ten months of current volume ^[5].

The funding totals do not reconcile. Crypto Briefing reports the round at $25 million, the new total at $60 million, and the previous raise at $41 million at a valuation of roughly $170 million ^[1][2][3]. Subtract the round from the total and $35 million is left, $6 million short of the reported prior figure ^[1]. Either the $41 million counts money the $60 million leaves out, or one of the numbers is wrong. Until that is settled, the $170 million mark is the only valuation in the report ^[3].

The stated use of the money is expanding the detection and audio analysis platform ^[1]. Carter Huffman and Mike Pappas co-founded the company in 2017 to sell voice skins that changed how users sounded in real time ^[12]. It now charges customers to catch altered voices.

This could run three ways. Detection holds its $0.25 price and pays for the rest of the platform. Or call-center and compliance vendors build their own detectors and the price drifts toward the $0.03 transcription rate; at that price the all-detection ceiling on current volume falls from $30 million a year to $3.6 million ^[3][4]. Or moderation stays the volume business and deepfakes stay the pitch.

I think the round is a bet on the first outcome. More than 100 models run over the same raw audio ^[5], so Modulate can sell detection alongside transcription and compliance monitoring in one pipeline. Crypto Briefing argues regulated buyers prefer that package to stitching together separate point tools ^[11]. Against that, whoever already carries the call can offer the same bundle. If Modulate discloses that detection makes up a small share of billed hours, this view is wrong.