Post-doc in Acoustic Event Detection (US citizens only)

dave_gelbart@yahoo.com wrote:
> Rune and Jerry,
>
> If I understand your complaint correctly, I think you are reading the
> ad as saying that the detection task is to be accomplished perfectly in
> all possible conditions?

No, I am not. I am reading the ad as to say that whatever task the
applicant is supposed to work with, is to be attmpted solved by
comparing a measured sound by all other possibly measured
signals.

This is a ridiculous strategy for any task regarding DSP. To whoever
accused me of being humourus about the ad: No, this is no joke.
I have worked on projects that were the solution was attempted
solved by this very strategy.

Think what you may about my other writings, but believe me on
this if nothing else: This is no joke. People actually try these
things and expect them top work. Carreers go bust -- not the
carreers of the people in charge, mind you -- , companies go
bancrupt, various security issues are blown wide open.

This is no joke.

> I did not write the ad myself but I am sure that perfect performance is
> not what the author meant.  As Rune said, that would be a rather
> unrealistic expectation.

If the ad really is for real, I suggest you go back to whoever wrote
the ad and ask them to think carefully about what people they
attempt to attract by the ad, and who they would like to be repelled
by it. 

Rune

dave_gelbart@yahoo.com wrote:
> Rune and Jerry,
>
> If I understand your complaint correctly, I think you are reading the
> ad as saying that the detection task is to be accomplished perfectly in
> all possible conditions?

No, I am not. I am reading the ad as to say that whatever task the
applicant is supposed to work with, is to be attmpted solved by
comparing a measured sound by all other possibly measured
signals.

This is a ridiculous strategy for any task regarding DSP. To whoever
accused me of being humourus about the ad: No, this is no joke.
I have worked on projects that were the solution was attempted
solved by this very strategy.

Think what you may about my other writings, but believe me on
this if nothing else: This is no joke. People actually try these
things and expect them top work. Carreers go bust -- not the
carreers of the people in charge, mind you -- , companies go
bancrupt, various security issues are blown wide open.

This is no joke.

> I did not write the ad myself but I am sure that perfect performance is
> not what the author meant.  As Rune said, that would be a rather
> unrealistic expectation.

If the ad really is for real, I suggest you go back to whoever wrote
the ad and ask them to think carefully about what people they
attempt to attract by the ad, and who they would like to be repelled
by it. 

Rune

dave_gelbart@yahoo.com wrote:
> Rune and Jerry,
>
> If I understand your complaint correctly, I think you are reading the
> ad as saying that the detection task is to be accomplished perfectly in
> all possible conditions?

No, I am not. I am reading the ad as to say that whatever task the
applicant is supposed to work with, is to be attmpted solved by
comparing a measured sound by all other possibly measured
signals.

This is a ridiculous strategy for any task regarding DSP. To whoever
accused me of being humourus about the ad: No, this is no joke.
I have worked on projects that were the solution was attempted
solved by this very strategy.

Think what you may about my other writings, but believe me on
this if nothing else: This is no joke. People actually try these
things and expect them top work. Carreers go bust -- not the
carreers of the people in charge, mind you -- , companies go
bancrupt, various security issues are blown wide open.

This is no joke.

> I did not write the ad myself but I am sure that perfect performance is
> not what the author meant.  As Rune said, that would be a rather
> unrealistic expectation.

If the ad really is for real, I suggest you go back to whoever wrote
the ad and ask them to think carefully about what people they
attempt to attract by the ad, and who they would like to be repelled
by it. 

Rune

"Rune Allnor" <allnor@tele.ntnu.no> writes:

> dave_gelbart@yahoo.com wrote:
> > Rune and Jerry,
> >
> > If I understand your complaint correctly, I think you are reading the
> > ad as saying that the detection task is to be accomplished perfectly in
> > all possible conditions?
> 
> No, I am not. I am reading the ad as to say that whatever task the
> applicant is supposed to work with, is to be attmpted solved by
> comparing a measured sound by all other possibly measured
> signals.
> 
> This is a ridiculous strategy for any task regarding DSP.

The original wording was "the detection of one particular target word
and/or sound in the background of all other possible words or any other
realistic sounds".  For the case of words, in speech recognition
terminology the task referred to is known as keyword spotting.  It's a
long standing area of research and has many important practical
applications, including command and control, as mentioned earlier.

It's important to realise that statistical techniques are used.  You can
make realistic assumptions about the variability of the "target word/or
sound" and realistic assumptions about the distribution of "all other
possible words or any other realistic sounds".  As such it comes down to
a statistical pattern matching problem where there is considerable scope
for novel models and algorithms.

In addition, the International Computer Science Institute is an
excellent place to work.


Tony

tony.nospam@nospam.tonyRobinson.com wrote:
> "Rune Allnor" <allnor@tele.ntnu.no> writes:
>
> > dave_gelbart@yahoo.com wrote:
> > > Rune and Jerry,
> > >
> > > If I understand your complaint correctly, I think you are reading the
> > > ad as saying that the detection task is to be accomplished perfectly in
> > > all possible conditions?
> >
> > No, I am not. I am reading the ad as to say that whatever task the
> > applicant is supposed to work with, is to be attmpted solved by
> > comparing a measured sound by all other possibly measured
> > signals.
> >
> > This is a ridiculous strategy for any task regarding DSP.
>
> The original wording was "the detection of one particular target word
> and/or sound in the background of all other possible words or any other
> realistic sounds".  For the case of words, in speech recognition
> terminology the task referred to is known as keyword spotting.

The term "all possible" is insane from both an engineering and a R&D
point of view. How do you know that you have covered **all**possible**
variants of anything? How do you know that some person from half-way
around the world doesn't throw in the odd word of his native language?
There are some languages of the bushmen in the Namib desert that
sound like clicks to me and most probably to almost everybody else
in the world. Do you count these "clicks" among the "all possible
words" of yours? If not, on what basis do you discriminate between
"possible" and "impossible" words?

Paying attention to detail is the key characteristic of the
professional,
irrespective of trade or craft.

> It's a
> long standing area of research and has many important practical
> applications, including command and control, as mentioned earlier.

Maybe this is a good project and a worthy application, I don't know.
The ad certainly leaves a somewhat twisted impression.

> It's important to realise that statistical techniques are used.  You can
> make realistic assumptions about the variability of the "target word/or
> sound"

Agreed.

> and realistic assumptions about the distribution of "all other
> possible words or any other realistic sounds".

No. You can make assumptions about "other words and sounds".
Once you make claims about "**all** other **possible** whatever"
you are in deep trouble.

> As such it comes down to
> a statistical pattern matching problem where there is considerable scope
> for novel models and algorithms.

Right. For a clever writer among laymen at the funding instiitution,
there is always that new parameter to check, that new algorithm to
design. The one factor that always lacks, is the well-defined
scope with measurable deliveries.

> In addition, the International Computer Science Institute is an
> excellent place to work.

I am sure it is. Any place that issues a recruitment ad as the one
above, has never faced a project scope that states measurable
deliveries inside a time frame and a budget.

Rune

Rune Allnor wrote:
...
> > In addition, the International Computer Science Institute is an
> > excellent place to work.
>
> I am sure it is. Any place that issues a recruitment ad as the one
> above, has never faced a project scope that states measurable
> deliveries inside a time frame and a budget.

Sheesh, are you in a bad mood today.

:-)

Rune Allnor wrote:
...
> > In addition, the International Computer Science Institute is an
> > excellent place to work.
>
> I am sure it is. Any place that issues a recruitment ad as the one
> above, has never faced a project scope that states measurable
> deliveries inside a time frame and a budget.

Sheesh, are you in a bad mood today.

:-)

Andor wrote:
> Rune Allnor wrote:
> ...
> > > In addition, the International Computer Science Institute is an
> > > excellent place to work.
> >
> > I am sure it is. Any place that issues a recruitment ad as the one
> > above, has never faced a project scope that states measurable
> > deliveries inside a time frame and a budget.
>
> Sheesh, are you in a bad mood today.
>
> :-)

On the contrary. Right now we are in port for the first time in nearly
three weeks.

As I said in another post, statemenst such as those found in the
ad must be taken at face value; the people who write the ads
have do. Without having the slightest clue about the implications.

Again, this is for real. Been there, done that. Survided, but just
barely. 

Rune

Rune,

I wrote a long reply which on re-reading of it has turned into somewhat
of a rant, so I'm not going to post it and see instead if we can agree
on what's causing you so much upset.

I think you are saying that you view the proposed task as a comparison
over all possible signals where you view the pattern space as vast and
impossible to enumerate and so you feel there will always be some part
of the pattern space that is poorly modelled.  I view the task as one of
a comparison over pattern space for which it is possible to come up with
realistic model of the distribution.  The space may still be vast and
impossible to enumerate but over the distribution considered sensible
models may be formulated and evaluated.

Do we agree on this?  Otherwise you finished your last post with an
indefensible libellous on ICSI, so I shall don my flameproof suit...


Tony

hi Allnor,
            i didnt mean that ad is humourous or you..i apologizeif it
meant the other way.i also worked in speech recognition project and i
understand the complications and toughness of the project and as you
told ad needs to be corrected. And particularly emphasizing a
particular sound or word out of many is damn tough.

regards
particle