SWAR CHIHNAYAN

"Ek Swar, Ek Pehchaan"

AI-Based Criminal Voice Identification System
Upload Mode Library Mode
Tips: Minimum 60 seconds of suspect sample is recommended for good accuracy.
Pro Tip: For optimal forensic accuracy, upload clear reference audio (approx. >25 dB)
You may upload one or more samples.

Investigation Parameters

Configure thresholds based on signal quality

Advanced Parameters
Controls how closely audio must match the target's voice print.

Lower: Catches faint voices.
Higher: Only exact matches.
0.55
Gatekeeper threshold.
Any audio segment below this confidence score is instantly discarded to save processing time.
0.28
Determines "Stickiness".
Higher values keep recording during pauses/breaths to maintain natural flow.
Normal
Merge logic.
If two positive clips are separated by less than this time (seconds), they become one clip.
2.0s
Folder Organization.
If the target is absent for longer than this time, upcoming clips are saved into a new folder/session.
5 Min
Advanced Filters

Apply noise reduction to evidence file (Use for static/hiss).

Isolate target voice from background noise.

Processing Audio Signal...