Hardware Questions

Do I need a Dialogic card?

Our Speech Engine is hardware-independent, so if the client application can collect the audio and put it into a buffer, the engine can decode it. Which hardware a particular client application needs depends only on the client application.

 

What kind of computer hardware does the Speech Engine need?

The following table presents the results of testing with the Windows version of our Speech Engine. The tests were done on a machine with an Intel Core 2 Duo. The grammar that was used has 500 proper names. The audio was 1.5 seconds in length. Each test ran for a total of 280 seconds, with each interaction taking 14 seconds.

 

Ports

Memory (MB)

Number of Decodes

Processor Utilization %

0

395

0

 

1

399

20

2

4

437

80

6

8

437

160

10

16

440

320

15

24

441

480

25

48

443

960

45

96

447

1920

81

 

Processor utilization is based on the amount of time the processor spends doing decodes. Some of the factors that can affect this value are the length of time of each interaction, which includes call initiation, greeting, prompts, the caller speaking, decode, confirmation (if any) and transfer/release time. The average time of each call is then taken and used to estimate how many calls the available resources can handle. This, in turn, allows us to estimate how efficiently the processor is utilized.

 

Note that the size of the loaded grammars will affect the memory requirements of the system.

 

Using other grammars and audio lengths affects the processor utilization as follows:

 

Domain

Utterance

Amount of Audio

Processor Utilization Modifier

Natural number

One million two hundred thousand and fourteen

2.3 seconds

 -6%

Yes/No

Yes

0.39 seconds

-76%

Date

January fourteenth nineteen ninety eight

3.4 seconds

1%

Currency

Four hundred thousand three hundred seventy two dollars and four cents

4.29 seconds

29%

 


Complete Help Topic List | Speech Engine Product Information