VoiceLock is a hardware version (DSP Chip Ported)
of VoiceCop (Speaker Verification Solution) which has
been targeted for security device for both household
and business fields. VoiceLock, stand-alone or together
with common key and keypad locking device, uses his/her
own human voice for its key to authenticate access to
precious property.
VoiceLock can be adapted to various field such as household
or office appliance for safe, drawl, automobile key.
Moreover, it would be an ideal product which can be
used for concurrently prevailing Cyber Apartments which
equipped with intelligent computer system. |
 |
Especially, compared to other security systems
using the finger print or the iris, this voice security
system with speaker verification stands high in public estimation
since there is a lot less danger of loss, forgetting of
password and its price efficiency.
Easy Voice Password Registration
With only 3 times voice password registration, VoiceLock
can learn one's own voice completely; therefore, makes it
possible users to apply voice verification solution by themselves
with no technical fear at all.
High Recognition Rate
Not only adapting conventional simple voice signal comparison
model but also plural voice comparison model which using
sentence-independent phoneme model which composed of 52
phonemes regardless of gender and age, VoiceLock has significantly
improved authentication rate.
Intelligent Verification Engine - Speaker Adaptation
To maintain high voice verification rate, VoiceLock automatically
reflects one¡¯s periodic (7 to 10 days) voice changes
into its voice password DB. As a result, VoiceLock completely
solves fatal verification rate decline problem with time.
Perfect Voice Password
VoiceLock continuously measures the differences of one¡¯s
registered voices taking into consideration of every circumstances
and user¡¯s unintentional voice registration errors.
Eventually, VoiceLock maximizes the capability of building
precise voice password.
High Recognition Speed
Equipped with comparison process using sentence-independent
phoneme model that composed of 52 phonemes regardless of
gender and age, VoiceLock can make all the possible voice
password without missing any pronunciation. Moreover, VoiceLock
avoids conventional vocabulary DB comparison which requires
some amount of time, significantly improving the voice verification
speed.
System Environment
Processor
Fixed-Point DSP 40Mips or more
Memory
Program Memory : 128 KByte
Data Memory (Data Working Space): 64 KByte
Flash Memory (Speaker Marking Vector Storage):
7 KByte/Person
Hardware Module
Mic./Speaker Interface: Microphone & Speaker
I/O Interface
AD/DA Converter: Analogue & Digital Converting
Device
MPU: Main Processor Unit
ROM: Program & Database Storage Memory
RAM: Program Execution Memory
Flash Memory: Personal DB Storage Memory for
Registered Speaker
Adaptation
Voice Lock for Door/Safe
Cyber Apartment Door System
