e021036.pdf

DEVELOPMENT KIT

Voice Extreme

Toolkit

When it comes to speech recognition products the California based com-

pany Sensory ( www.sensoryinc.com ) have been at the forefront for some

time now. Their latest offering is the Voice Extreme™ speech recogni-

tion kit.

This kit contains a complete development

environment allowing applications to be pro-

grammed using VE-C, a high-level language

very similar to ANSI-C. As you would expect

this language gives a great boost to pro-

grammer productivity compared with Assem-

bler based software environments and allows

the development of speech-controlled appli-

cations in a Windows environment. The resul-

tant code is compiled and downloaded to the

Voice Extreme module flash memory. The

module can then be run in the development

board or removed and plugged into the target

system. The kit is distributed in the UK by:

application software.

A prototyping area is also avail-

able on the board to mount addi-

tional application specific hardware.

All the I/O signals of the module are

available along with the supply

voltage rails. A connector is also

included if you prefer to build the

additional hardware on an external

card.

The heart of the development kit is

the Voice Extreme module. It includes

Milford Instruments

Leeds. LS25 5AQ

Tel:01977 683665

www.milinst.com

The development kit contains a versatile

hardware development card containing the

speech module together with a software Inte-

grated Development Environment (IDE)

including an editor with source code debug

and the Quick Synthesis™ program which

allows any sound files with the .wav exten-

sion to be used in any of the Sensory RSC

applications.

One board – many possibilities

The development kit hardware includes all

the electronics, a microphone, speaker and

sockets for connection to external equip-

ment. An RS232 cable and mains adapter is

also included. On the card there are five

pushbuttons, one is for system reset another

for program download/program starting

while the remaining three, along with three

status LEDs are free to be used by your

Elektor Electronics

1/2002

DEVELOPMENT KIT

the RSC-300/364 microcontroller, crys-

tal oscillator, ﬁrmware ROM and a

2 Mbyte ﬂash memory to store your

own application software. This mod-

ule plugs into the development board

(and eventually into your target appli-

cation) via a 34 pin header. For pro-

duction purposes the Voice Extreme

module is available in a QFP package

or in die form. In-circuit reprogram-

ming is also supported.

Software

The bundled software includes an

Integrated Development Environ-

ment (IDE). This program suite con-

tains all the tools necessary to write

application specific software and

includes a C editor, compiler, linker

together with help functions and

downloader. The editor supports C-

Highlighting which improves pro-

gram readability.

The IDE is supplied on a CD ROM

together with the Quick Synthesis™

program that converts and com-

presses .wav ﬁles into a compatible

file format used by the hardware.

Also included on the CD are exam-

ples and extensive documentation.

In order to write speech recogni-

tion programs easily, it is of course

necessary to make use of the library

routines provided. Thanks to the

numerous examples included on the

CD together with the on-line help

facility we found it a simple process

to understand and make use of these

library functions.

In the software documentation a

distinction is made between

‘speaker independent’ and ‘speaker

dependent’ speech recognition

technology. In the first case the

software will simply recognise a

signal but in the second case it will

recognise that signal only from a

particular speaker. The difference

between the two is a question of

the accuracy with which the soft-

ware evaluates the signal. The

higher the accuracy then the more

speaker specific the result will be.

The software allows the recognition

accuracy to be adjusted in steps.

As well as the speech recognition

routines are also software sound

routines that can be used to gener-

ate synthetic music and allow you

to play with samples and sound

fragments.

In practice

All of the included example pro-

grams were tested in the Elektor

Electronics laboratory. The speech

recognition functioned outstandingly

well despite the low quality micro-

phone fitted to the card. The manu-

facturer suggests substituting a high

quality microphone to further

increase software recognition per-

formance.

The examples demonstrate just

some of the possibilities of this pow-

erful tool. The accompanying source

code gives you a good insight as to

how the card can be used in many dif-

ferent situations and is an ideal starting point for

you to begin writing your own applications.

In addition to the speech recognition

examples there are also interesting music

generation demos and compression of tone

samples that work with the library supplied.

The kit is ideal for anyone interested in the

field of speech recognition and especially if

you already have an application in mind. All

you need is a PC running Windows (95+)

with 16 Mbytes RAM, 15 Mbytes free hard

disk space, an unused RS232 port together

with a little C programming experience and

of course the purchase price of the kit

£109+VAT.

(010086-1)

1/2002

Elektor Electronics

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: