Name: VoiceGrid X

Text: (c 3)


Voice Biometric Solutions:

VoiceGridTMStream is a speaker identificationapplication aimed at finding audio files containing the
target speakers' voices in a stream of audio data. The system can perform a N:N search of up to
10,000 incoming voice samples per day against a target or watch list database of up to 100 known or
suspect voice records.

The VoiceGridTMX application is designed for speaker identification. The system allows sorting all the
records in accordance with target speakers' voice samples. The standard configuration allows investigators
to process up to 10,000 incoming recordings per day against a list of up to 100 targets.
VoiceGridTMStream voice matching, along with the entire VoiceGridTMproduct line, uses STC exclusive
analysis tools, including processing, segmentation, feature extraction and identification of voice samples.

VoiceGridTMStream is an important supplement to investigative or forensic analysis. It is ideal for the
lawful interception of audio communications to quickly identify suspect voices and subsequently isolate the
conversations of interest in a wide range of critical areas including:
Terrorist threats
Dmg traficking
Gang & organized crime

Domestic vit ' rce
Abuse of emergency or 9 1 1
communication channels
Prank or false report calls

VoiceGridTMStream can be used as a stand-alone tool to enhance future voice identification capabilities or
serve as the foundation for a larger voice recognition system infrastructure.



Matching incoming records to target speaker samples
Automatic 1 :N or N:N voice sample matching
Monitoring or speaker surveillance in near real-time mode
Automatic 2-speaker dialogue separation
Full report generation
Adjustable processing options to expedite throughput
Adjustable thresholds to filter results
Text-independent and language-independent
All types of the recording channels supported: CDMA, GSM.
landline, microphones

High-speed performance
Speaker identificationin near real-time
Detects two speakers within one audio file (in mono or
stereo channels)
Meets the needs of law enforcement community
Can be used as a stand-alone solution and/or integrated
with other intelligencesystems


VoiceGridm Lab stand-atone
applicationfor 'l:Nsearches of up to 100
records per individual investigation.


I -

Suspects voice samples

audio files classified according to suspects voice samples

V~iceGrid" Locaf - single server
network solution for 1N searches,
supports a database of up to 10,000
recordsand up to 10 workstations.

VoieeCriP Naticn - networked l:N
search and voice data management
solution with unlimited database size
and unlimitedweb client worktation
connectivity. Distributed processing
and WAN architecture designed fo~
Large uty, county, state or national
system deployment=

Managing voice ID results:
Selected files are transferred to folders created for each of the target speakers
Reports in CSV and HTML
ID results available for all files
Results sent in XML over TCP (the network)
Message sent to specialized client for real-time pop-up at operator workstation
- -

Recommended hardware:
Operating System


Windows XP, Windows 71 Windows
Server 20031 Windows Server 2008
CPU Intel Core i7

VoiceCridm Stream -a sophisticated
distributed solution for realtime speaker identification in
communication channels, phar
calts, files from mobile devices ana
other sources. Integrates wiih voice
database up to 10,000 target speakers
which can be scaled up to millions of
processedrecords per day.

4 GB
100 GB

Speech Techology Center



4 Krasutskogo street
St. Petersburg, 196084
Tel.: +7 812 331 0665
Fax: +7 812 327 9297

13 Hellwigstrasse
Saarbrueken, 66121
Tel.: +49 (0)681 8590565


VaiceGridm X L a sta
applicationfor speaker identification
in multiple files. Performs a N:N

Document Path: ["brochure512.pdf"]


Click to send permalink to address bar, or right-click to copy permalink.

Un-highlight all Un-highlight selectionu Highlight selectionh