Select your language:
Deutsch English Italiano Türkçe Français Español
Home
Projekt Ibi
Projectteam
Project partner
Project concept
Simon-touch AAL
Project concept
Projectteam
Products
simon
sam
ssc
simone
BENEFIT/07-2011
Projektteam
Projektpartner
Projektkonzept
Dokumente
Videos
ASTRO/03-2012
Concept
Documents
Videos
For Endusers
Hardware
Installation
Application package
Function Description
Tips & Tricks
Workshops
Documentation
For Developers
Downloads
Wiki
Forum
Bugtracker
Blog
Organisation
Publications
Press
constitution
Registration
Board
Projectteam
Imprint
Sponsoren
2011
2010
2009
2008
2007

Analysis of available Software

After we made some tests with available software we came to the point, that none of this programs fit to our agenda:

  • High cognitive rate
  • Platform independent
  • Modularity
  • Being up-to-date
  • User-firendly


Criteria

criteria

IBM Via Voice

Dragon Naturally Speaking

Sphinx

Julius

Recognition Rate (60%)

5

8

6.5

8

Platform independent (15%)

6

2

8

9

Costs (5%)

7

4

10

10

Modularity (15%)

5

1

9

10

Up-to-date (5%)

 

8

5

7

Overall result

5.0

5.85

7.2

8.5

The recognition rate was the only actual important criterion for us, and thats why it is 60% of the total points.

To be Up-to-date was not that important because it is possible that even a program, which is not developed for years, still can be efficient.

The platform independence had a small weighting because you can adjust the platform and the last point, modularity, got only 15% because also not changeable programs can be good or even “perfect”.


point allocation:

For each Program and each criterion we hat 10 points. Where 0 points are the worst and 10 points are the best.

To deliver you an insight, we describe you why we gave the points for each criterion.

 

Recognition rate:

We got these values from test reports or similar and compared them.

At this point we had no own opinion

 

Platform independence:

Via Voice is a program for Windows and got only that much points, because there was a working version for Linux.

Dragon Naturally Speaking got 2 points because it runs with the program Wine on Linux, even if it is a program for Windows, but it is very complex to get it running.

While Julius, as an Engine, works on every platform, Sphinx is due to its Java-roots heavier to integrate it in the system.

 

Costs:

Even there we had no influence. Sphinx and Julius are Open-Source and that is why they got 10 points each. Via Voice is available for around € 60,- whereas Dragon Naturally Speaking costs, depending on its version, between € 200,- and € 1000,-.

 

Modularity:

Dragon Naturally Speaking is hardly changeable, that why so few points. Via Voice is in principle modular, but quite limited because it is Closed Source. In fact that Julius is only an recognition engine, it is very modular and completely adjustable – nearly like Sphinx, which has only a few restrictions

 

Up-to-date:

Via Voice will not be developed any more. We got all other information again from test reports or similar and compared them.

 

We got two winners. Due to the table we had to take Sphinx and Julius, which we had not done because we wanted a solution for windows and a platform independent solution.

 

So there were two concepts left:

  • Dragon Naturally Speaking
  • Julius


Workout of 2 solutions:

Because both concepts varied in their focus, we decided to draw both up an present them to our client. He decided in accordance with our recommendation for the concept with Julius (called “simon”)

 

Dragon Naturally Speaking:

+ High cognitive rate

+ System integration

+ direct Recognition of commands and dictation

- not cross platform compatible

- individual adjustments nearly not possible

Julius:

+ High cognitive rate

+ Modularity

+ Open Source

+ easy adjustable/expandable

+ cross platform compatible

+ up-to-date

- no language models available