BACK TO SELECTED WORK
PROJECT / AUDIO INTELLIGENCE / DEEPFAKE DETECTION

Impact.

Detects AI-generated speech in five languages using WavLM and calibrated predictions.

Impact analyzes speech with specialized acoustic classifiers and can report uncertainty when evidence is inconclusive.

PythonPyTorchWavLMFastAPILibrosa
01 / PROJECT FLOW

How it fits
together.

Drag a card to arrange the flow. The arrows stay connected.

01

Prepare audio for inference

02

Analyze local and global acoustic patterns

03

Return a calibrated human, AI, or uncertain result

02 / INTERACTIVE ARCHITECTURE

Explore the system.

Drag cards · arrows stay connected
01 / INPUT
02 / PROCESS
03 / ANALYZE
04 / OUTPUT
Select a card to read what it does. Drag cards to rearrange the flow.
Back to workGet in touch