[Kde-accessibility] Fwd: [SoK 2013] Summary of proposed changes in SImon

[Peter Grasch]

Hi,

On 06/20/2013 02:07 PM, Sanjiban Bairagya wrote:
> Here are the main changes that have been planned to be made to the
> interface so far-
> 1. New users of Simon often don't know the fact that all scenarios are not
> compatible with all base models. So if we do select a scenario and a base
> which
> is not compatible, what happens presently is that it does not work. So we
> can
> think of a way to let the user know about this compatibility from the
> interface
> itself, that is, by using small but visible footnotes, or else we can
> design it such,
> that by default when user chooses a scenario it will select the base model
> with
> which it can work.
Yes, the latter approach sounds good.

Basically, we'll want to abstract this a tiny bit further to present the
user with use cases that pull in the required scenario with a matching
base model.

Please draw a mockup of how you think such a system could look like.
That'll give us something concrete to discuss.

> 2. As far as voice recognition is concerned, I think that, along with the
> volume, the
> signal to noise ratio is also very important. So we can modify the
> recognition
> section in a way, such that when an individual speaks on the mic for
> atleast a
> minimum duration of time X, this action (of speaking) can be automatically
> detected by Simon, and the snr will be calculated from that very moment
> till the
> time there is no voice speaking for some time Y. This will be compared to
> an
> optimal snr and respectively it can be showed directly if it is okay or
> not. By doing
> so, the user wont have to start a training every time he just wants to
> check his
> mic configurations. All he needs to do, is talk and then Simon would say if
> it's
> okay or not.
Yes, we're already showing the volume, why not the SNR.

The way I think this would fit best if we'd show a passive warning in
the main window right next to the volume if we detect a dangerously low SNR.

This should be relatively easy to implement.
You'll find the widget that shows the current loudness (both in the main
window and the training wizard) in
simonlib/simonsound/devicevolumewidget.cpp.
You already get the current loudness in
DeviceVolumeWidget::deviceReportedLevel (and a stream time, actually)
and the minimum SNR we are looking for is accessible through
SoundConfiguration::minimumSNR() (soundconfig.h).

You can look in simonlib/simonsound/wavrecorderclient.cpp how the actual
SNR calculation looks.
In your case (as this is a continuous) stream, you'll have to store
volume levels in devicevolumewidget.cpp, age them appropriately (you
have a stream time, remember?) and compare those.

> 3. I was also thinking of adding a small slider on the right hand side of
> the screen,
> which would show the Simon Manual for quick reference. By "slider" I mean,
> a
> small left-arrow button, which when clicked will show the manual on the
> right of
> the screen taking around 25-30 % of the screen only.
Disagree on this one. Help is handled in a standard way for all
applications. Consistency is important.

Best regards,
Peter Grasch