Speakers Authentication and Fraud Detection

SoniCracker enhances contact center security and customer experience, leveraging advanced identification and verification technologies, thereby reducing losses, by employing highly-effective fraud prevention and detection tools.

The Threats

New voice impersonation technologies such as DeepFake, and technical challenges to retrieve significant information from voice, such as speaker verification and speaker intent makes it easier, faster and cheaper to commit voice fraud, crime and even harm national security.

How do we eliminate such threats?

We employ a disruptive holistic Audio Processing and Voice Intent (as a Service) solution. It targets and retrieves significant audio information, providing relevant insights, detecting events and conduction analytics. This leads to more effecting Fraud Prevention and Detection as well provides appropriate recommendation for action.

How it works?

Audio Processing

The sophisticated SoniCracker AI based algorithm processes the audio sample, retrieving all relevant information including the number of speakers, overlap and-non speech separation, speaker ID, exposure of recorded calls and deepfake voice use.

Audio Intent

The sophisticated SoniCracker NLP algorithm analyzes the speaker content and reveals his/her real genuine intent.

Why choose SoniCracker?


Full Information Extraction via innovative Audio Source Separation
Both non-overlap and overlapping speech
Working in reverberant audio environments
Best Benchmark accuracy of total words comprehension


Free Speech Verification which does not require a fixed pass-phrase
Robust to speaker voice perturbations
Best Benchmark accuracy of correct verifications in Large scale Data Base


Cloud Native application with simple APIs to integrate to any existing infrastructure
Runs on CPU (no need for GPU)
Simple integration with an existing CRM/Fraud systems
Saves up to 50% of storage space
Compatible with cell phones, runs as an application

Typical use cases

Speakers identification and verification

SoniCracker specializes in voice biometric, which supports speaker identification and verification. The voice biometric technology creates a “fingerprint” of the human voice, which is used for speaker identification/verification and for measuring speaker reliability. This product is used in various market verticals such as banks, for improving customer quality of experience as well as for preventing and investigating fraud/crime.

Speaker overlap separation

SoniCracker specializes in speaker source separation. Our algorithm can separate multiple speakers that either one with another or are in sequence, during a conversation, in real-time. This is also known as the "Cocktail-party effect". Our cutting-edge algorithm is language-agnostic and can separate multiple speakers in a noisy and reverberant environment.

We use this algorithm as the key engine for our fraud detection system, but it can also be used as a standalone tool to accelerate investigations, provide better user authentication and speech-to-text quality, and save CAPEX on storage due to regulation.

Product supported Feature List

Speaker Identification and Verification - Voice Biometric
Speakers Overlap Separation and Diarization
Attempt for Impersonation - Deepfake discovery
Background Audio Analysis
Unknown Speaker Personalization information
Speaker Intent, Insights and wisdom
Key words Search
Emotion tracker

Requestdemo access

SoniCracker is ready to perform a demo of our technology per your request. Demo can be provided via both prerecorded data and Live per feature request. During the demo we can also explain the mechanism behind it. To get it please fill in the details.
Contact Details
2164 Nutters Barn Lane,
Ames, Iowa