Happy New Year - what now though?

Happy New Year  - but what of the future. What do you think the next 20
or so years will bring to DSP or related disciplines?

Personally I can see hand-held real-time languge (ie speech recognition
and synthesis) translators with beamformers for noise cancellation. The
next few years I feel that FPGAs will play more of a role since the
tools will get better and better.
A bit more distant I can see Quantum computers solving real problems.
DSPs will always get faster and smaller but what of the theory? New
kinds of adaptive filters? A wristwatch-phone with mpg video? Will
everything just get smaller and cheaper?
I would be interested to have your views.

Tom

Tom <somebody@nOpam.com> wrote in message news:<3FF25141.3AA2CF94@nOpam.com>...
> Happy New Year  - but what of the future. What do you think the next 20
> or so years will bring to DSP or related disciplines?
> 
> Personally I can see hand-held real-time languge (ie speech recognition
> and synthesis) translators with beamformers for noise cancellation. The
> next few years I feel that FPGAs will play more of a role since the
> tools will get better and better.
> A bit more distant I can see Quantum computers solving real problems.
> DSPs will always get faster and smaller but what of the theory? New
> kinds of adaptive filters? A wristwatch-phone with mpg video? Will
> everything just get smaller and cheaper?
> I would be interested to have your views.
> 
> Tom

As for applications, your guess is as good as mine. If the HW technology 
keeps shrinking at the same rate it has done up to now, I will not be 
surprised to see hand-held equipment (of the size of current mobile phones)
that do everything current PCs do in terms of entertainment (games, net 
surfing, email and chats). Actually, I think the newest mobile phones are 
almost there as we speak.

What theory is concerned, I believe we will see a backslash not dissimilar 
to the "dot com" spoof a few years ago. As of today, there are lots of 
"visionaries" around who see all sorts of technologies and applications 
coming. At some point these "visionaries" will have to stand and deliver, 
and I believe 99.9% of them will be exposed as, well, not quite up to 
the task. Once this balloon has bursted, the rest of us may hopefully 
proceed doing real world jobs with real world tools meeting real world 
expectations. And get payed sufficiently to lead a real world life.

Rune

Tom wrote:

> Happy New Year  - but what of the future. What do you think the next 20
> or so years will bring to DSP or related disciplines?

Happy New Year to you, too, Tom, and the comp.dsp as well.

I have simple prediction to make: things are going to get more DSP
software-centric (rather than DSP hardware-centric). The reason? The
horsepower required for typical algorithms isn't changing yet the
available horsepower is growing. And since processors are much more
flexible (programmable) than hardware, there usage will be preferred
over hardware implementations. Of course there will always be the
niche HW applications that require high speed, low cost, low power,
or a combination of each.

> Personally I can see hand-held real-time languge (ie speech recognition
> and synthesis) translators with beamformers for noise cancellation. 

I have already witnessed a fantastic four-mike beamformer. The problem
is getting the price down, and I mean by just a few dollars.

> The
> next few years I feel that FPGAs will play more of a role since the
> tools will get better and better.
> A bit more distant I can see Quantum computers solving real problems.
> DSPs will always get faster and smaller but what of the theory? New
> kinds of adaptive filters? A wristwatch-phone with mpg video? 

Look for cell phones within the next year that will do streaming mpeg
video over 2.5G/3G networks (e.g., EDGE).

> Will
> everything just get smaller and cheaper?
> I would be interested to have your views.

Cheap is crucial.
-- 
%  Randy Yates                  % "...the answer lies within your soul
%% Fuquay-Varina, NC            %       'cause no one knows which side
%%% 919-577-9882                %                   the coin will fall."
%%%% <yates@ieee.org>           %  'Big Wheels', *Out of the Blue*, ELO
http://home.earthlink.net/~yatescr

"Tom" <somebody@nOpam.com> wrote in message
news:3FF25141.3AA2CF94@nOpam.com...
> Happy New Year  - but what of the future. What do you think the next 20
> or so years will bring to DSP or related disciplines?
>
> Personally I can see hand-held real-time languge (ie speech recognition
> and synthesis) translators with beamformers for noise cancellation. The
> next few years I feel that FPGAs will play more of a role since the
> tools will get better and better.
> A bit more distant I can see Quantum computers solving real problems.
> DSPs will always get faster and smaller but what of the theory? New
> kinds of adaptive filters? A wristwatch-phone with mpg video? Will
> everything just get smaller and cheaper?
> I would be interested to have your views.
>
> Tom

Integration. Higher speed.

Shorter lifetimes of chips, driven in part by the need of mobile phone
manufacturers
for ever more features. Already seeing 500MHz+ chips in
some of the current phones.
Samsungs 500MHz arm chip.

More programmable analog devices
like Cypress psoc but next step up.

Bit late with the phone with mpeg video prediction.
Already on sale in Japan (not wrist watch size).
I'll try and dig up a link.

Also there is the mpeg player that uses codec to fit a dvd on a largish
flash card
for play back on a gameboy adavance sized player
and also back of car set player.
Was an article in one of the ieee mags, spectrum I think,
sometime in the last 6 months.
not sure if this link will work, may have to be a member
http://www.spectrum.ieee.org/WEBONLY/resource/oct03/tool.html

http://www.zvue.com/
http://zvue.com/mp4info_01.html

Can even get an fpga addon for gameboy advanced
 www.charmedlabs.com
http://www.charmedlabs.com/xportmain.htm

also wireless network access
http://www.x-trafun.com/home.htm

Alex Gibson

"Alex Gibson" <me@privacy.net> wrote in message
news:bt0iuf$23uuq$1@ID-140944.news.uni-berlin.de...
>
> "Tom" <somebody@nOpam.com> wrote in message
> news:3FF25141.3AA2CF94@nOpam.com...
> > Happy New Year  - but what of the future. What do you think the next 20
> > or so years will bring to DSP or related disciplines?
> >
> > Personally I can see hand-held real-time languge (ie speech recognition
> > and synthesis) translators with beamformers for noise cancellation. The
> > next few years I feel that FPGAs will play more of a role since the
> > tools will get better and better.
> > A bit more distant I can see Quantum computers solving real problems.
> > DSPs will always get faster and smaller but what of the theory? New
> > kinds of adaptive filters? A wristwatch-phone with mpg video? Will
> > everything just get smaller and cheaper?
> > I would be interested to have your views.
> >
> > Tom
>
> Integration. Higher speed.
>
> Shorter lifetimes of chips, driven in part by the need of mobile phone
> manufacturers
> for ever more features. Already seeing 500MHz+ chips in
> some of the current phones.
> Samsungs 500MHz arm chip.
>
> More programmable analog devices
> like Cypress psoc but next step up.
>
> Bit late with the phone with mpeg video prediction.
> Already on sale in Japan (not wrist watch size).
> I'll try and dig up a link.
>
> Also there is the mpeg player that uses codec to fit a dvd on a largish
> flash card
> for play back on a gameboy adavance sized player
> and also back of car set player.
> Was an article in one of the ieee mags, spectrum I think,
> sometime in the last 6 months.
> not sure if this link will work, may have to be a member
> http://www.spectrum.ieee.org/WEBONLY/resource/oct03/tool.html
>
> http://www.zvue.com/
> http://zvue.com/mp4info_01.html
>
> Can even get an fpga addon for gameboy advanced
>  www.charmedlabs.com
> http://www.charmedlabs.com/xportmain.htm
>
> also wireless network access
> http://www.x-trafun.com/home.htm
>
> Alex Gibson


voice interfaces added to everything.
talk to your fridge, toilet etc

couple I missed

older one - microsoft smart watch, arm 7 + 512MB in a watch
http://www.spectrum.ieee.org/WEBONLY/publicfeature/jan04/0104comp2.html

blue semiconductor lasers on the way
http://www.spectrum.ieee.org/WEBONLY/publicfeature/jan04/0104semi1.html

superconducting motors for US navy.
finally working
http://www.spectrum.ieee.org/WEBONLY/publicfeature/jan04/0104tran1.html

Alex

Tom wrote:
> Happy New Year  - but what of the future. What do you think the next 20
> or so years will bring to DSP or related disciplines?

I recently heard an interview on the radio with guy from university
who is involved in wearable computing. He was asked the same question,
and his reply was that technical prognosis up to about ten years into
the future usually over-estimates the actual evolution, whereas after
that we usually under-estimate. I guess it has to do with the human
mind intuitively only being capable of linear extrapolation. I'll give
it a go nevertheless:

As for wearable computing, I think that miniaturisation and other
developments (eg. flexible displays, HUD capable glasses, &c.) will
bring some cool clothes and accessoires. Read Stephenson's "Snow
Crash" for a readable account of what this might look like.

I also expect an increasing grade of computerization in cars (lots of
development money there). Radar, head-up displays, image processing,
object recognition, voice recognition, noise reduction (active and
passive), GPS and perhaps direct satellite links are going to play key
DSP roles in that segement in the near future.

For audio (my business), the availability of increasing processing
power is going to spawn a whole new generation of algorithms, for
processing and audio generation. I've seen an exponential growth of
DSP power in products in the eight years time I worked for a hardware
audio processor manufacturer (I made a nice chart of this once for an
AES exhibition, but it unfortunately got stuck at the customs), and
this increase was only partly due to higher sampling rates. The other
part allowed for more sophisticated implementation of classical
processing algorithms, and in part, for implementation of completely
new algorithms which were not possible on older hardware. I don't see
an immediate end to that trend.

On the theoretical side, I'm expecting some useful output of the AI
camp, regarding the application of neural networks in DSP, or
database-oriented expert systems. A nice applicaton would be real-time
language translation, another might be blind signal seperation, or at
least blind signal extraction.

Generally, the availability of miniature high-speed DSPs and memories
can bring devices for relief and support of vision and hearing
impaired people. I can think, for example, of language recognizing
(and also translating?) hearing aids, or newspaper reading hand-held
scanners.

In the spirit of Freeman Dyson, I'm also hoping that technological
development can help bridge our lack of social development. Small,
cheap, satellite-based Internet access devices (all-in-one type,
specialized for communication and not general computing) to bridge
political, religous and economical boundaries could be an idea.

> A bit more distant I can see Quantum computers solving real problems.

Hard to predict development in that field. The problem still is: once
we got them, what do we do with them? If their birth is imminent,
nobody is going to use factoring based cryptography anymore, and what
else can they really do (apart from factoring and unsorted list
search) that is useful?

> DSPs will always get faster and smaller but what of the theory? New
> kinds of adaptive filters? A wristwatch-phone with mpg video? 

I want one like Michael Knight! Speaking of which, I also want Kitt!

> I would be interested to have your views.

It's always fun to speculate :).

Regards,
Andor

Andor wrote:

>
>
> It's always fun to speculate :).
>
> Regards,
> Andor

Whatever happened to Genetic algorithms? I realise there are thousands of
papers on the subject but some 10 years ago it was to be the save-all of
technology. We could design anything mearly by 'evolutionary' methods.
I don't see Genetic algorithms taking over from least-squares techniques for
filter design for example?


Tom

Tom wrote:

> Happy New Year  - but what of the future. What do you think the next 20
> or so years will bring to DSP or related disciplines?
> 

I've no idea.

HOWEVER, your question and resulting posts raised a question.

Would current technology make feasible a USB device that would:
1. have a microphone input
2. keep the last 1 to 2 seconds of input ( speech presumed ) locally
3. output the FFT of the previous X msec of data every Y msec [ 
presuming (X/Y)>~10

I have a "GUT" feeling that current commercial speech recognition has 
conflicting "loyalties' so to speak. "Which allophone" vs "which word 
makes sense in 'context' ". Humm, did you notice I did NOT mention 
PHONEME recognition ;?

Would speech recognition benefit by divide and conquer?

Tom <somebody@nOpam.com> wrote in message news:<3FF48CA1.77E4525D@nOpam.com>...
> Andor wrote:
> 
> >
> >
> > It's always fun to speculate :).
> >
> > Regards,
> > Andor
> 
> Whatever happened to Genetic algorithms? I realise there are thousands of
> papers on the subject but some 10 years ago it was to be the save-all of
> technology. We could design anything mearly by 'evolutionary' methods.
> I don't see Genetic algorithms taking over from least-squares techniques for
> filter design for example?
> 
> 
> Tom

I tried the GAs with a student, and others used GAs in the type of data 
analysis I was involved with at the time. 

The problem with GAs is that they are axtremely expensive, computationally 
speaking, and they give no guarantees. You gan apply a GA to a reasonably 
behaved problem (finite and small search space) and never find the true
solution. There are no indicators provided for the quality of the provided
solution either. To me, GAs are the last resort, only to be used when 
absolutely everything else has failed.

Rune

Richard Owlett wrote:

> Tom wrote:
>
> > Happy New Year  - but what of the future. What do you think the next 20
> > or so years will bring to DSP or related disciplines?
> >
>
> I've no idea.
>
> HOWEVER, your question and resulting posts raised a question.
>
> Would current technology make feasible a USB device that would:
> 1. have a microphone input
> 2. keep the last 1 to 2 seconds of input ( speech presumed ) locally
> 3. output the FFT of the previous X msec of data every Y msec [
> presuming (X/Y)>~10
>
> I have a "GUT" feeling that current commercial speech recognition has
> conflicting "loyalties' so to speak. "Which allophone" vs "which word
> makes sense in 'context' ". Humm, did you notice I did NOT mention
> PHONEME recognition ;?
>
> Would speech recognition benefit by divide and conquer?

Humans seem to manage real-time speech translation (with a delay of half a
sentence or so) so I expect it is possible.

Tom