Speaker verification
Detect when the same participant has enrolled twice in a trial using advanced voice technology
97% accuracy in detecting the same participant enrolled twice in a trial
97%+ Accuracy rate when using the de-duplication technologyÂ
Revolutionary Voice Technology
Advanced AI-powered acoustic analysis that safeguards clinical trial integrity while saving time and resources.
Powered by Winterlight Labs’ innovative speech analysis platform, our speaker identification solution leverages machine learning to accurately extract and evaluate key acoustic properties, enabling the identification of duplicate patients.Â
AI-powered Quality Tools
Our advanced machine learning algorithms analyze pitch, tone, cadence, and acoustic properties to detect duplicate speakers with the highest accuracy.
Lightning-Fast Processing
Catch duplications earlier in the screening process to save significant time and money. Real-time analysis delivers immediate results.
Security & Compliance
Built with security and consent at the foundation. Fully compliant with GDPR and HIPAA regulations for complete peace of mind.
Easy Integration
Functions as a standalone solution or integrates seamlessly within existing workflows and systems without disruption.
How the solution Works
Audio Capture & Analysis
The system analyzes existing audio recordings captured from participants during trials, focusing on acoustic characteristics rather than linguistic content.
Acoustic Fingerprinting
Our proprietary technology examines intricate acoustic characteristics, focusing on elements such as pitch, tone, and cadence to create unique voice fingerprints.
Systematic Comparison
The detailed acoustic fingerprint is systematically compared against the acoustic profiles of all other enrolled participants' pre-randomization audio data.
Rapid Results
When a duplicate is detected we work with the sponsor and the sites to limit the participant to a singular enrolment with immediate corrective action.
How it works
Step one: Initiate recording
The rater starts the audio recording at the beginning of the session to capture all voices, including the participant, rater, and any additional speakers.
Step two: Transcribe & label
Speakers are identified, transcribed, and subsections within the scale are marked (example; “Word recall” and “Commands” for ADAS-Cog).
Step three: Quality check
Transcript and audio are scanned for scale-specific quality indicators, ensuring that prompts were read correctly, in the correct sequence, and that appropriate follow-ups were made.
Step four: Generate report
Produce a quality report for each scale, specifying the detected quality indicators and their locations within the assessment.
Our speaker identification solution addresses critical challenges across multiple scenarios in clinical trial management.
- Participant Duplication Detection
- At-Home Assessment Authentication
- Rater Verification & Authentication
Get in touch
Don’t let participant duplication and high costs derail your important work. Trust us to safeguard your trials from both.
Contact us today to learn how we can revolutionize your approach to CNS clinical trials.
