RealityChek, a content-verification platform owned by Hydaway Digital, has added a capability that analyzes live audio streams to flag synthetic or AI-generated speech as it is heard, rather than after a recording has finished.
The feature, Streaming Audio Detection, runs continuous forensic analysis on a live audio stream and returns a verdict in real time as audio comes in. The company says it examines spectral patterns, prosodic markers, compression artifacts, and generative-model signatures characteristic of synthesized speech, and that it is available through the RealityChek API and detection interface. Hydaway is pitching it for call center fraud prevention, live media monitoring, voice authentication systems, and compliance recording analysis.
The company argues that most existing detection tools focus on pre-recorded audio files analyzed after the fact, a workflow it says is too slow for settings where trust decisions have to be made live. Detecting synthetic speech as it happens, rather than in stored files, has become a growing industry focus as voice cloning spreads. OmniSpeech has built deepfake voice detection into Zoom meetings, and, in the live telephony that RealityChek is targeting, Pindrop and BT have partnered to counter AI-driven voice fraud in UK call centers.
Karl Kottmeier, chief executive of Hydaway, called audio deepfakes “the sleeper threat of the AI fraud landscape,” arguing that much of the damage happens in real time, over the phone and in live broadcasts, rather than in static file uploads. The company said voice cloning is being used in phone scams impersonating executives and family members, in call center fraud, and in attempts to bypass voice-authenticated financial and government systems.
Hydaway Digital, based in North Vancouver and listed on the TSX Venture Exchange, operates RealityChek through a wholly owned subsidiary. The platform offers detection across image, video, audio, and text content, along with identity verification and liveness detection.
Sources: Hydaway Digital
–
By Ali Nassar-Smith
