Ensemble speaker verification achieving 97% accuracy - Intelligent fusion of MFCC+DTW (92%) and Resemblyzer CNN (94%) for voice authentication
-
Updated
Nov 13, 2025 - Jupyter Notebook
Ensemble speaker verification achieving 97% accuracy - Intelligent fusion of MFCC+DTW (92%) and Resemblyzer CNN (94%) for voice authentication
Real-time AI voice cloning and deepfake detection for live communication using dual ML experts (WavLM + LFCC-LCNN), real-time audio analysis, risk scoring, and forensic LLM reasoning.
Research prototype: React Native client that warns users during calls about AI voice impersonation and phone scams. Fail-open, consent-based. Companion to shieldcall-core.
Momenta's Speech DF Arena benchmark results: per-utterance score files for all 14 datasets, Arena protocol files, and a script to reproduce the 0.89% pooled EER / 1.01% average EER.
VoxShield: منصة لكشف الهجمات الصوتية والتزييف الصوتي، تميّز الصوت الحي من إعادة التشغيل والصوت المولد بالذكاء الاصطناعي عبر تحليل البصمة الصوتية ونموذج Random Forest، مع تقرير مفصل وتنبيهات وبيئة محاكاة تدريبية (Python, Flask, scikit-learn)
🛡️ AudioShield AI is an AI-powered voice security platform that analyzes audio conversations to detect deepfake indicators, scam patterns, impersonation, social engineering, and attacker intent. It combines voice, conversation, and behavioral signals to generate an explainable Threat Score and identify dangerous interactions in real time.
The missing trust layer for the open business communications stack. Antifraud toolkit for EU SMEs · EUPL-1.2 · NLnet NGI Zero Commons proposal.
Pindrop — independent third-party profile of a public API surface, by API Evangelist. Pindrop is a voice and audio security company providing a continuous identity verification and deepfake-detection platform for enterprises. Its technology authenticates callers and video-meeting participants, detects synthetic (AI-generated) speech and fraudulent
Streaming dual-stream fraud + deepfake detection engine for telephone conditions (STRF, SDTG, CSCF, PMA)
Hiya — independent third-party profile of a public API surface, by API Evangelist. Hiya is an AI-native voice security company that blocks spam, fraud, and deepfake calls at network scale for mobile operators, enterprises, and consumers. Its developer platform exposes several REST APIs: Hiya Connect for branded outbound calling (registering origina
To associate your repository with the voice-security topic, visit your repo's landing page and select "manage topics."