Searching \ for '[PIC] Speech recognition, PIC or otherwise' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/microchip/devices.htm?key=pic
Search entire site for: 'Speech recognition, PIC or otherwise'.

Exact match. Not showing close matches.
PICList Thread
'[PIC] Speech recognition, PIC or otherwise'
2007\09\07@053258 by Jinx

face picon face
I've been asked if there's any way I can put together a demo for next
Sunday. Yes, absolutely no pressure, that's more than a week away,
and nothing will go wrong of course !

Been Googling around and found examples for dsPICs. Which would
be OK if I had more time. Maybe I can squeeze some experiments
in but I have to get the chips first. And being asked last thing on a
Friday when the PIC shop is closed for the weekend is not helpful

I haven't found anything for lesser PICs. Many examples of PICs
interfacing with the HM2007, but HMC seem to have been eaten up
by Elan and although Elan have some speech-output ISD-type chips
I don't see a speech-input chip like the HM2007 on their site. Some
retailers found but none in NZ

I'm sure someone somewhere has tried this for an older PIC

Another option, if I can't do this quickly in h/w, would be to find
PC software that can perhaps recognise words or numbers and send
that out (I couldn't hazard a guess in what form or for what reason -
robotic or automation application perhaps ?) of a serial / parallel port

Any help greatly appreciated

Free downloads especially. Even crippled or limited-duration will do,
it needs to run just long enough to prove a point

TIA

2007\09\07@061426 by Jake Anderson

flavicon
face
Vista's speech recognition is reputed not to suck.
Stuff comes in as if via a keyboard. you could hack that up in a few
minutes.
Alternately dragon voice perhaps might still be available?

Jinx wrote:
{Quote hidden}

2007\09\07@061906 by Ruben Jönsson

flavicon
face
Hello,

Vista has speak recognition (I don't know if all versions have it though).

Perhaps it can be set up to recognize labels of buttons in an application (if
you only need a limited set) so the application can send data over a serial
port to your PIC when the buttons are pressed or spoken.

I don't know if this is possible but the speak recognition is supposed to be
good enough to allow Vista to verbally execute commands from a played soundfile
and this could apparently even be used as a hacking tool to get access to your
computer.

/Ruben



{Quote hidden}

> -

2007\09\07@063439 by Tamas Rudnai

face picon face
Hi,

It could be, but highly unlikely as every speech recognition needs a bit of
training so that the operator 'teaches' his/her voice to the system
otherwise it would not easy to recognise words. Like the standard Vista
knows a kind of American accent so that if somebody in London would talk to
his computer it would not understand that pronunciation. But AFAIK there are
other issues with voice recognition. With mobile phone you only could teach
couple of words to ring someone or take a call. But you have to teach it
again. I do not know too much details about it but as far as i can remember
you need to recognise fraction of sounds and then you will have the sequence
of a word which you can find in a table.

Tamas



On 9/7/07, Ruben Jönsson <spam_OUTrubenTakeThisOuTspampp.sbbs.se> wrote:
{Quote hidden}

> > --

2007\09\07@064257 by Tamas Rudnai

face picon face
Hi again,

Just a quick search on Google shows up loads of info, like these:

www.faqs.org/docs/Linux-HOWTO/Speech-Recognition-HOWTO.html
http://www.generation5.org/content/2002/howsrworks.asp

This is also very interesting:
http://www.ee.columbia.edu/~stanchen/e6884/slides/lecture12.avsr.pdf

The Google result:
http://www.google.ie/search?hl=en&q=voice+recognition+theory&btnG=Google+Search&meta=

Tamas


On 9/7/07, Tamas Rudnai <.....tamas.rudnaiKILLspamspam@spam@gmail.com> wrote:
{Quote hidden}

>

2007\09\07@070810 by Jinx

face picon face

> Vista's speech recognition is reputed not to suck.

Jake/Tamas, I've come across references to Vista and other PC
software and how effective they are. Now I realise I've got a
second problem. Even if I did a crash course in these systems, I'm
still probably incapable of sending the recognition result out of a
port, because I simply don't know the language in which they're
written. That's why I was hoping for a straw to clutch in the
robotics area

The dsPIC still looks attractive, mainly because I know a lot
more about micros, and how to program them to send signals
out of pins

It is still a daunting task to be up and running and passably
competent in a week, but I'll give it a go if I have to. A possible
complication is that Microchip include this chip as a requirement
to pre-process the audio

www.silabs.com/tgwWebApp/public/web_content/products/Wireline/Voice_C
odec/en/Si3000.htm

An agent is in NZ but, like the required dsPIC, I'll have to wait
until Monday to find out if stock is available PD toute suite

2007\09\07@072907 by Gerhard Fiedler

picon face
Jinx wrote:

>> Vista's speech recognition is reputed not to suck.
>
> Jake/Tamas, I've come across references to Vista and other PC software
> and how effective they are. Now I realise I've got a second problem.
> Even if I did a crash course in these systems, I'm still probably
> incapable of sending the recognition result out of a port, because I
> simply don't know the language in which they're written. That's why I
> was hoping for a straw to clutch in the robotics area

It could work somehow if you put a terminal program into the foreground.
One of the functions of these voice recognition systems is to just send
keystrokes, simulating you typing the words you're speaking. The terminal
then sends that out through serial.

Doesn't have to be Vista; XP has that also.

Gerhard

2007\09\07@073129 by Xiaofan Chen

face picon face
On 9/7/07, Jinx <.....joecolquittKILLspamspam.....clear.net.nz> wrote:
>
> The dsPIC still looks attractive, mainly because I know a lot
> more about micros, and how to program them to send signals
> out of pins

Interestingly there is a thread going on in Microchip forum.
You might want to take a look.
http://forum.microchip.com/tm.aspx?m=166133

Last year I attened the 16bit seminar in Singapore and
the voice quality of the demo was ok.

Xiaofan

2007\09\07@091900 by Jinx

face picon face
> Doesn't have to be Vista; XP has that also.

>From the XP Pro Help
For speech recognition (SR)
1. Open Speech in Control Panel, and then click the Speech Recognition tab

OK, problem. No Speech Recognition tab. I can select Microsoft
Sam as the voice of the computer though

Yay, wahoo ;-)

If he doesn't sound like Yosemite Sam, not interested

I'm still looking at the dsPICs, but have a worried feeling all the
examples are in C. I don't do C

2007\09\07@093553 by Michael Rigby-Jones

picon face


{Quote hidden}

Joe,

Probably not suitable due to your immediate time constraints, but SensoryInc <http://www.sensoryinc.com/products/integrated_circuits.html> make integrated circuits and modules for speech recognition.

To install the speech recognition components on XP, take a look at the following:
<http://support.microsoft.com/kb/306537>

Cheers

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2007\09\07@094804 by Gerhard Fiedler

picon face
part 1 1522 bytes content-type:text/plain; charset="big5" (decoded quoted-printable)

Jinx wrote:

>> Doesn't have to be Vista; XP has that also.
> >>From the XP Pro Help
> For speech recognition (SR)
> 1. Open Speech in Control Panel, and then click the Speech Recognition tab
> > OK, problem. No Speech Recognition tab.
Well, it's not all built-in by default, you may have to specifically
install it :)
>From <http://tinyurl.com/2233d>:

----------------------------
There is no Speech Recognition Engine (SRE) built directly in to Windows
XP. You need to install a compatible engine, and in most cases, you'll
install it from one of two [ge: actually three] sources.

¡E        The Microsoft SRE for Windows XP is available in Office XP applications,
such as Microsoft Word 2002, or Microsoft Excel 2002. If you have Office
XP, or one of these programs on your computer, then you probably already
have access to the SRE. You simply need to install the SRE as explained
below.
¡E        If you do not have one of these programs, then a less expensive option
for installing the SRE is to purchase the Windows XP PLUS! Pack. This is
available for less than $30 from Microsoft and other online vendors.
¡E        A third option is available for advanced users. The SRE is provided for
free as part of the Microsoft Speech Software Development Kit 5.1.
Microsoft provides no technical support for this software and it is not
generally recommended for end users. ----------------------------

See also <http://tinyurl.com/2nhak6>

Gerhard



part 2 35 bytes content-type:text/plain; charset="us-ascii"
(decoded 7bit)

2007\09\07@095725 by Dario Greggio

face picon face
Gerhard Fiedler wrote:

> Well, it's not all built-in by default, you may have to specifically
> install it :)

That is the main difference with Vista  :)

I have it installed on Windows 2000 too (for home automation).
It *may* work, problem is that you have to be close to the Mic, and this
does not make much sense to switch on the lights or the washing machine... !

--
Ciao, Dario

2007\09\07@100223 by Jinx

face picon face
> SensoryInc <www.sensoryinc.com/products/integrated_circuits.html>
> make integrated circuits and modules for speech recognition

Yup, I saw those and traced them through to Arrow. I slumped because
that usually means indent and MOQ. And I suspect that may be the case
with dsPICs too. Arrow NZ can supply many things but a lot of it is ex-
Australia

> To install the speech recognition components on XP, take a look at the
> following:
> <http://support.microsoft.com/kb/306537>

Thanks

For a proof-of-concept demo a PC will do. And that will give me some
breathing space to try and embed it with a PIC. I'll be seeing the person
who sprung this on me and suggest he not spring things on me

2007\09\07@165515 by Marcel Duchamp

picon face
Jinx wrote:
> I'm sure someone somewhere has tried this for an older PIC
>
> Any help greatly appreciated

Sometime back, an idea was floated on the piclist but I don't know how
it worked out.

The plan was this.  Feed a microphone input through a  pre-amp and then
to two separate op-amps setup as lowpass and highpass circuits.  Then
each of these to comparators and into the pic of your choice.  Some form
of agc or perhaps manual control with vu meters may be required for
obtaining usable dynamic range.

The idea being to separate the users voice into low and high bands and
produce waveforms that the pic could sort out for simple phrases such as
 on, off, up, down, 1, 2, 3,... and so on.  The user must undergo some
training.

I suppose if this idea is viable, at least for a small number of simple
phrases enunciated well, that the pic could also modify the  front end
analog characteristics perhaps (ie tune it) to help out a bit.  Plus,
one could have more than 2 channels.

That's all I remember about the idea but it looked possibly feasible to
me.  Whether it's feasible for you in the short time you have is
something that could entertain us all!

2007\09\07@191851 by Jinx

face picon face
> Whether it's feasible for you in the short time you have is
> something that could entertain us all !

I love you too ;-)

You could be right though. I think Mr Demo-springer has bitten
off more than I can chew

XP speech recognition might be the short-term answer if it works

2007\09\08@144644 by Cedric Chang

flavicon
face
>
>
> You could be right though. I think Mr Demo-springer has bitten
> off more than I can chew
>
> XP speech recognition might be the short-term answer if it works

If anyone has ever used an embedded ( small size,  not PC size )  
solution for voice recognition that they thought actually worked  
well, please tell me about it.  Voice Dependent is okay with me.    
Cedric


2007\09\08@153123 by peter green

flavicon
face

> If anyone has ever used an embedded ( small size,  not PC size )  
> solution for voice recognition that they thought actually worked  
> well, please tell me about it.  Voice Dependent is okay with me.    
> Cedric
>
>
>  
How small does it need to be, would something like an epia PX booting
off a CF card be usable?

2007\09\08@193231 by Jinx

face picon face

> If anyone has ever used an embedded ( small size,  not PC size )
> solution for voice recognition that they thought actually worked
> well, please tell me about it.  Voice Dependent is okay with me.
> Cedric

I was sent this link off-list

instruct1.cit.cornell.edu/courses/ee476/FinalProjects/s2006/avh8_css3
4/avh8_css34/index.html

(the 'voice control car budget' panel is quite funny)

2007\09\08@215023 by Cedric Chang

flavicon
face
>
> On Sep 8, 2007, at 1:31 PM, peter green wrote:
>
>
>> If anyone has ever used an embedded ( small size,  not PC size )
>> solution for voice recognition that they thought actually worked
>> well, please tell me about it.  Voice Dependent is okay with me.
>> Cedric
>>
>>
>>
> How small does it need to be, would something like an epia PX booting
> off a CF card be usable?

That would be okay.....

2007\09\09@032147 by Russell McMahon

face
flavicon
face
>>> If anyone has ever used an embedded ( small size,  not PC size )
>>> solution for voice recognition that they thought actually worked
>>> well, please tell me about it.  Voice Dependent is okay with me.

Strictly finite vocabulary or large vocab?
If limited, what sort of dictionary size?


       Russell


2007\09\09@145120 by Cedric Chang

flavicon
face
>
>
>
> On Sep 9, 2007, at 1:22 AM, Russell McMahon wrote:
>
>>>> If anyone has ever used an embedded ( small size,  not PC size )
>>>> solution for voice recognition that they thought actually worked
>>>> well, please tell me about it.  Voice Dependent is okay with me.
>
> Strictly finite vocabulary or large vocab?
> If limited, what sort of dictionary size?
>
>
>         Russell

finite    maybe 200 words   maybe less
Cedric


2007\09\10@135431 by Jinx

face picon face
<http://www.sensoryinc.com/products/integrated_circuits.html>

make integrated circuits and modules for speech recognition

Reply I got from Sensory Inc via Adilam via Arrow. Samples of
ICs might be possible, but there's a whole lot of development
paraphernalia you need (it's like ordering PICs when you haven't
got a programmer or editor)

===============

I can  arrange for a quote to be sent from Sensory Inc. but because
this is very much a "volume" business, I will need the following details.
(I still need these details even if it is a revision of an existing
project.)

Project name:
Brief Project Description:
Project Start date:
Project Status:  ie evaluating/prototyping/production.
Estimated Annual Usage:
Project Engineer:
Target Price: if you have one

Please note that the Supplier will not send samples or quotes without
these details, and also it means we can obtain best price and delivery

2007\09\10@155104 by Cedric Chang

flavicon
face
>
> On Sep 9, 2007, at 8:02 PM, Jinx wrote:
>
> <http://www.sensoryinc.com/products/integrated_circuits.html>
>
> make integrated circuits and modules for speech recognition
>

Jinx
If the quote is for my benefit, thank you very much.
What I really want to see is a demo of 200 speech dependent
words working really well.  ( Like 99.5 % )
Thanks

Cedric

BTW   Last time I checked out sensory ( 4 years ago ).  Their demo  
s**ked.

2007\09\10@162103 by Brent Brown

picon face
On 10 Sep 2007 at 14:02, Jinx wrote:

> <www.sensoryinc.com/products/integrated_circuits.html>
>
> make integrated circuits and modules for speech recognition
>
> Reply I got from Sensory Inc via Adilam via Arrow. Samples of
> ICs might be possible, but there's a whole lot of development
> paraphernalia you need (it's like ordering PICs when you haven't
> got a programmer or editor)

Hi Jinx,

Oddly enough I received a NZ priced catalog from Digikey in my letterbox yesterday
(didn't ask for one, didn't know they did NZ). When quickly loooking through it
Sensory Inc voice recognition caught my eye. NZ$58 for a VR Stamp module,
NZ511 for a "tool kit". Check out nz.digikey.com or phone 0800 449 837.

--
Brent Brown, Electronic Design Solutions
16 English Street, St Andrews,
Hamilton 3200, New Zealand
Ph: +64 7 849 0069
Fax: +64 7 849 0071
Cell: +64 27 433 4069
eMail:  @spam@brent.brownKILLspamspamclear.net.nz


2007\09\10@182932 by Jinx

face picon face
> Oddly enough I received a NZ priced catalog from Digikey in my
> letterbox yesterday (didn't ask for one, didn't know they did NZ)

You're the 2nd NZ person to tell me that. Maybe I'll get one too

> When quickly loooking through it Sensory Inc voice recognition
> caught my eye. NZ$58 for a VR Stamp module, NZ511 for a
> "tool kit". Check out nz.digikey.com or phone 0800 449 837

Arrow do have a little Sensory Inc stock ex-Melbourne. One price
I could get was $399 for the RSC-4128 Demo. I'm still not sure
exactly what you need as a minimum to be up and running. I think
it's comparable to what tools you'd select from Microchip, because
the Sensory Inc documentation goes into assembly language at the
microcontroller level

We're also seriously considering PC-based Nuance. The reason
being that a local company who do voice-activated phone systems
use Nuance as their engine (said company's systems BTW start
at $500,000, which is a little pricey. I mean, I've got that on me ;-))
but Nuance itself is a lot cheaper)

dsPIC not out of the frame either (I would personally like to explore
that, although Sensory Inc are already there, as well as PIC-based
CVSD to replace an obsoleted National part,), but for ease of
implementation in future developments I think the project will be on
a server

2007\09\11@111833 by Tony Smith

picon face
> We're also seriously considering PC-based Nuance. The reason
> being that a local company who do voice-activated phone
> systems use Nuance as their engine (said company's systems
> BTW start at $500,000, which is a little pricey. I mean, I've
> got that on me ;-)) but Nuance itself is a lot cheaper)


Nuance works quite well, their angle is no training.  They were doing
Australian & Scottish accents last time I came across them,  but that was
many years ago.  $500,000 is cheap for this sort stuff.

Under the 'otherwise' option, how about this:
<http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=300148162545>?

(It was 10 pounds last time it was listed, he must have realised how
valuable it really is.  Ha.  He's still lacking a few zeros though.)

120 port IVR, recognises a few words, can be trained for more.  Practically
bullet-proof, redundant EVERYTHING, you can even pull the CPU's out, and
it'll pick up where it left off when you put it back it.  No, it won't sing
'Daisy'.  It pulls that trick off by writing all of its registers, etc to
disk.

Software doesn't stop either, dividing by zero is perfectly ok, it'll just
ignore you.

If it worked, and if the terminal was still there (unlikely), I'd buy it.
Start my own 'phone service company.  I'm surprised the mgf didn't buy it,
they'd know it's history.  Hmmmm.

Tony

2007\09\11@211558 by Jinx

face picon face
> Nuance works quite well, their angle is no training.  They were
> doing Australian & Scottish accents last time I came across them,
>  but that was many years ago.  $500,000 is cheap for this sort
> stuff.

It does appear to be a good product for the front end. The back
end (user GUI and output) doesn't seem too difficult to implement,
being compatible with C++ and VB

Although I'm working with non-technical people who start many
sentences with "All you have to do is ..." or "Can't you just ..... ?"

> Under the 'otherwise' option, how about this:
> <cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=300148162545>?
>
> (It was 10 pounds last time it was listed, he must have realised how
> valuable it really is.  Ha.  He's still lacking a few zeros though.)

When Pavarotti got to the Pearly Gates, St Peter called out to God
and said "Here's that tenner I promised you"

Too soon ?

I saw it finished at 30UKP, with no bids. That's appalling for the
seller, but it looks like they've made no attempt to get a realistic
price, so TS for them. Good grief, the hardware alone at that price
is more than worth it to a tinkerer. I would have listed it with a
sizeable reserve or tried selling it back to Telsis

2007\09\12@041948 by Alan B. Pearce

face picon face
>Although I'm working with non-technical people who start many
>sentences with "All you have to do is ..." or "Can't you just ..... ?"

<VBG> one of my colleagues, recently retired, banned certain words from such
conversations. 'Should', 'just', 'only' were on the list ...

2007\09\12@063445 by Tony Smith

picon face
> I saw it finished at 30UKP, with no bids. That's appalling
> for the seller, but it looks like they've made no attempt to
> get a realistic price, so TS for them. Good grief, the
> hardware alone at that price is more than worth it to a
> tinkerer. I would have listed it with a sizeable reserve or
> tried selling it back to Telsis


Too right.  Brand new that would have cost someone a few million.

For the UK, Oz & probably NZ as well, if you've ever seen (or rung) a number
that's appeared on TV, in the paper etc, of the "ring & win a prize", "ring
& we'll fax you...", "call now, big boy..." type, then that box is what's at
the other end.

Even as a 2nd-hand rack (standard sizes, + castors) it's worth a few quid.
Depending on configuration, you'd also get a couple of battery backed 48v
power suppliers, 12/5v switchers, 5-fan rack, a 68000 CPU (ha), and assorted
telephony related bits & pieces.

That's a DX-120 (also in -60, -30 & -15 versions), and can handle 120
simultaneous telephone calls, either voice, fax or data (inbound and/or
outbound).  Network to TCP-IP & RS232 as well.  The reason for the big price
tag is they don't slow down under load, unlike PC-based stuff.  Very very
reliable.  Painful to program at times, the speed of BASIC combined with the
clarity of assembler.  Weird when an instruction to set a variable takes the
same amount of time as one to play a file, send a network command etc.  1
instruction = 1 tick, no exceptions.

All you need is an ISDN line or two...

Tony


Look, it's easy, here's a chunk of code to record a voice file:

ACRecord:  A = @836         \set item
   U = @808                \set answer code
\
   Wait                    \'til file plays
\
   Q @831 = 0, ACBeep
\
   A *, ACBeep             \set up abort
   @0 = @891 + @890        \file to play
\
ACWaitForStar:  Skip @827 = 0   \play prompt?
   P @0                    \yes
   B 1                     \short beep
   H 8                     \pause
   Wait
   J ACWaitForStar         \keep looping
\
\
ACBeep:  B 4                \after the beep..
\
   Block                   \keep start time
   @840 = %T Mod 360
   @840 = @840 * 60
   @840 = @840 + %S
\
   Skip @824 = 0           \* abort enabled?
   A *, ACStar             \set * abort
   Skip @825 = 0           \# abort enabled?
   M #, ACHash             \set # abort
\
   R = @804                \set length
   T = @806                \& timeout
\
   VT @835, A              \record file/item
\
   @837 = 0                \rec finished

...hey, it's not machine code.

2007\09\12@075627 by Howard Winter

face
flavicon
picon face
Jinx,

On Wed, 12 Sep 2007 13:15:49 +1200, Jinx wrote:

> Although I'm working with non-technical people who start many
> sentences with "All you have to do is ..." or "Can't you just ..... ?"

ROFL!  Been there...

I used to have a boss (Programming manager) whose favourite phrases when dealing with user departments asking for things were:

"Easy enough"
"A mere bagatelle"
"No real problem"

Having heard this, the users tended to be very upset when told later that the "mere bagatelle" they'd asked for would take six months to achieve...

I'd much rather under-promise and over-deliver than the other way round!

Cheers,


Howard Winter
St.Albans, England


More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...