Debbie Hynes wrote:
> PICers - i'm kicking round a project idea that needs playback of short
> audio messages. Under control of a PIC, naturally! :)
>
> What sort of options are there? There are the ISD chips (ChipCorder).
> What else would you go for? Just casting round for ideas .....
How many seconds? What bandwidth?
*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com
-- http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
Each message ... a few seconds, say 5 sec? BW, say up to 10kHz? Maybe 12 - 20
messages all up.
If a dedicated recorder chip, I'd prefer one that doesn't need its own dev
board or anything. Big $$ :( I can convert analogue sigs to audio files (wav,
au etc) so could T/F the file serially. Or use analogue direct if a chip would
accept it (eg ISD's 1400 series.)
The project's just at the ideas stage right now. has anybody had experience
with ISD's latest offerings?
Debbie
--- Olin Lathrop <spam_OUTolin_piclistTakeThisOuTEMBEDINC.COM> wrote:
>
> How many seconds? What bandwidth?
>
> *****************************************************************
Debbie Hynes wrote:
> Each message ... a few seconds, say 5 sec? BW, say up to 10kHz? Maybe
> 12 - 20 messages all up.
5 seconds/message x 20 messages x 20K samples/sec x 1 byte/sample =
2Mbytes
That's a lot more than can be stored internally in a PIC. A sound
recording chip may be a good choice. If you've got a PIC in the circuit
already, you could just hang the 2Mbytes off the PIC. Watch for the
recording bandwidth if it's flash or EEPROM though.
Why 10KHz bandwidth? You don't need more than 4KHz bandwidth to reproduce
good quality voice.
*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com
-- http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
This guy has a neat little device. Uses an 'F84, voice chip with NVRAM for 60
seconds voice quality. ISD2560
www.winbond.com.tw/c-winbondhtm/partner/PDFresult.asp?Pname=919
(ISD25120 for 2 min) Don't know if you can break up the message into smaller
ones. Haven't studied the datasheet yet. I'm sure there are some other chips
fromWinbond that will. http://www.silcom.com/~pelican2/PicoDopp/MICROHUNT.htm
> Debbie Hynes wrote:
> > Each message ... a few seconds, say 5 sec? BW, say up to 10kHz? Maybe
> > 12 - 20 messages all up.
>
> 5 seconds/message x 20 messages x 20K samples/sec x 1 byte/sample =
> 2Mbytes
>
> That's a lot more than can be stored internally in a PIC. A sound
> recording chip may be a good choice. If you've got a PIC in the circuit
> already, you could just hang the 2Mbytes off the PIC. Watch for the
> recording bandwidth if it's flash or EEPROM though.
>
> Why 10KHz bandwidth? You don't need more than 4KHz bandwidth to reproduce
> good quality voice.
>
-- http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
> > > Each message ... a few seconds, say 5 sec? BW, say up to 10kHz? Maybe
> > > 12 - 20 messages all up.
> >
> > 5 seconds/message x 20 messages x 20K samples/sec x 1 byte/sample =
> > 2Mbytes
Use an ISD unless the quality must be utterly superb
I've used the ISD2590 for years in a communication product for people who
can't talk.
The people who buy it (speech therapists etc) say that my product's speech
quality is as good or better than anything available.
It gives better quality than 'telephone" quality speech.
However - compared to eg 44 kbps / CD quality sound or whatever it's poor.
> > Why 10KHz bandwidth? You don't need more than 4KHz bandwidth to
reproduce
> > good quality voice.
As above.
4 kHz is entirely adequate for most purposes.
But if you want the 'voice on black satin background" quality you get from
CD quality recording then you need something better.
Probably at least a 12 bit codec with perhaps companding for storage.
I suspect you don't need anything like this quality in this application.
use an ISD ! :-)
For real quality use a 16 bit or so sigma-delta converter and save 16 bit
samples.
Russell McMahon
-- http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
part 1 499 bytes content-type:text/plain; (decoded 7bit)
Olin Lathrop wrote:
> Debbie Hynes wrote:
> > Each message ... a few seconds, say 5 sec? BW, say up
> to 10kHz? Maybe 12 - 20 messages all up.
>
> 5 seconds/message x 20 messages x 20K samples/sec x
> 1 byte/sample = 2Mbytes
Just 1K/s is pretty enough.
Listen to attached mp3. (12K/12s)
> Olin Lathrop wrote:
> > Debbie Hynes wrote:
> > > Each message ... a few seconds, say 5 sec? BW, say up
> > to 10kHz? Maybe 12 - 20 messages all up.
> >
> > 5 seconds/message x 20 messages x 20K samples/sec x
> > 1 byte/sample = 2Mbytes
>
> Just 1K/s is pretty enough.
> Listen to attached mp3. (12K/12s)
>
> Mike.
>
>
>
> --
> http://www.piclist.com hint: The list server can filter out
> -----Original Message-----
> From: Mike Singer [.....msingerKILLspam.....POLUOSTROV.NET]
> Sent: 04 November 2003 05:02
> To: EraseMEPICLISTspam_OUTTakeThisOuTMITVMA.MIT.EDU
> Subject: Re: [EE:] Sound chips - options?
>
>
> Olin Lathrop wrote:
> > Debbie Hynes wrote:
> > > Each message ... a few seconds, say 5 sec? BW, say up
> > to 10kHz? Maybe 12 - 20 messages all up.
> >
> > 5 seconds/message x 20 messages x 20K samples/sec x
> > 1 byte/sample = 2Mbytes
>
> Just 1K/s is pretty enough.
> Listen to attached mp3. (12K/12s)
>
> Mike.
Next challenge is to decode MP3's in realtime with just a PIC :o)
Mike RJ
=======================================================================
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.
=======================================================================
Any questions about Bookham's E-Mail service should be directed to postmasterspam_OUTbookham.com.
Michael Rigby-Jones wrote:
> > > > Each message ... a few seconds, say 5 sec? BW, say up
> > > to 10kHz? Maybe 12 - 20 messages all up.
> > >
> > > 5 seconds/message x 20 messages x 20K samples/sec x
> > > 1 byte/sample = 2Mbytes
> >
> > Just 1K/s is pretty enough.
> > Listen to attached mp3. (12K/12s)
> >
> > Mike.
>
> Next challenge is to decode MP3's in realtime with just a
> PIC :o)
40MHz -> 10M instructions/s -> 1000 instructions per sample.
For such great coders as Scott, Olin, Dmitry it is not
a challenge, I'm absolutely sure :-)
Denny Esterline wrote:
> Yeah, But who can make sense out of Bob Dylan regardless
> of sampling frequency?
Have not caught your point, sorry. (Perhaps due to my weak
English)
Did not you see that 12K file lasted 12 seconds. Or weren't
you unable to look through file's properties: 8k/s quality,
11K frequency?
Sure, and I believe Mike has some nice libraries that *encode* and decode
MP3's in realtime on a PIC....
Actually, Debbie, ADPCM might be worth looking into. 16F source is available
from Microchip, I'm working on an 18F implementation right now. Will
definitely be real-time on an 18F252 (which is what I'm playing with, and
I'm at about ~40 KHz!), and at the sampling rate's you'll need (8 KHz?),
it'll leave plenty of juice to do lots more. Of course, it's a relatively
large undertaking, compared to just getting a chip that does most of it for
you... :-) .
If you do decide to go the ADPCM route, what to use for storage is a thorny
issue. 8 KHz ADPCM is ~4 kbytes/sec, might want to look into FRAM, which I
believe is cheaper than comparable NVRAM.
The ultimate option would of course be to use GSM or another of the
speech-compression algorithms (that's what commercial answering machines,
etc. use). I believe GSM requires FP ops though, so you'd need to see if you
could implement it on a PIC.
> -----Original Message-----
> From: Mike Singer [KILLspammsingerKILLspamPOLUOSTROV.NET]
> Sent: 04 November 2003 08:28
> To: RemoveMEPICLISTTakeThisOuTMITVMA.MIT.EDU
> Subject: Re: [EE:] Sound chips - options?
>
>
> Michael Rigby-Jones wrote:
> > > > > Each message ... a few seconds, say 5 sec? BW, say up
> > > > to 10kHz? Maybe 12 - 20 messages all up.
> > > >
> > > > 5 seconds/message x 20 messages x 20K samples/sec x
> > > > 1 byte/sample = 2Mbytes
> > >
> > > Just 1K/s is pretty enough.
> > > Listen to attached mp3. (12K/12s)
> > >
> > > Mike.
> >
> > Next challenge is to decode MP3's in realtime with just a
> > PIC :o)
>
>
> 40MHz -> 10M instructions/s -> 1000 instructions per sample.
>
> For such great coders as Scott, Olin, Dmitry it is not
> a challenge, I'm absolutely sure :-)
>
> Best Regards,
>
> Mike,
Right...now we just need a PIC that runs OK at 40MHz :o)
Mike RJ
=======================================================================
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.
=======================================================================
Any questions about Bookham's E-Mail service should be directed to spamBeGonepostmasterspamBeGonebookham.com.
> > 40MHz -> 10M instructions/s -> 1000 instructions per sample.
>
> that might look like a lot, but it will be consumed pretty quickly by
> just a little floating or fixed-point math!
>
> Wouter van Ooijen
Jinx, I think you missed out some zeros there...50MHz Scenix ->
50,000,000 instructions per second :)
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
Josh misunderstood:
> Jinx, I think you missed out some zeros there...50MHz Scenix ->
> 50,000,000 instructions per second :)
>
Jinx was refering to Instructions per sample, not second.
Regards
Mike RJ
=======================================================================
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.
=======================================================================
Any questions about Bookham's E-Mail service should be directed to TakeThisOuTpostmasterEraseMEspam_OUTbookham.com.
Michael Rigby-Jones wrote:
> > 40MHz -> 10M instructions/s -> 1000 instructions per sample.
> >
> > For such great coders as Scott, Olin, Dmitry it is not
> > a challenge, I'm absolutely sure :-)
>
> Right...now we just need a PIC that runs OK at 40MHz :o)
Ask Olin, he is supplied now with good PICs just to stop
his critics against MChip :-)
Jinx wrote:
> > Mike,
> > Who was forced give up PIC coding.
>
> Oh-oh. Court order ? What've you been up to ?
Oh-oh, no, no court order, much, much worse, -
it's a life. Chained to Mchip, you are prisoned
for the life to kinky architecture, max 16 bit
devices(still in the future), strange set of docs
and tools and so on.
So I feel better with MS.
("Delenda est Carthago" Herbert, I agree :-)
> 40MHz -> 10M instructions/s -> 1000 instructions per sample.
>
> For such great coders as Scott, Olin, Dmitry it is not
> a challenge, I'm absolutely sure :-)
The problem, as I understand it, isn't so much the speed at which things
must be done, but the sheer volume of memory such a process takes up.
However, with that said, considering the math involved the speed of the
process is also most certainly a big problem. TTYL
----- Original Message -----
From: "Mike Singer" <msingerEraseME.....POLUOSTROV.NET>
> Oh-oh, no, no court order, much, much worse, -
> it's a life. Chained to Mchip, you are prisoned
> for the life to kinky architecture, max 16 bit
> devices(still in the future), strange set of docs
> and tools and so on.
>
> So I feel better with ***MS***
Microsoft? Incredible! I guess then you were fed up with embedded devices in
general....?
> I've looked through archive, almost all your posts contain
> somebody's e-mail address. Could you please cut it off in
> the future, at least my address.
Too lazy....pray I switch to Thunderbird soon :-P
> > Sure, and I believe Mike has some nice libraries that
> > *encode* and decode MP3's in realtime on a PIC....
That was sarcasm....:-)
Bottom line, MPEG compression requires a horrible amount of floating point
ops; it is simply impossible to do it in real-time on a fixed-point 18F,
even at 40 MHz. Take a look at how fast (the extremely efficient, I must
say) FFT algo on piclist.com runs, and then figure out how to do 1000x that
(for the DCT/iDCT) in *real-time* on a PIC......
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
Jinx wrote:
>
> > Jinx, I think you missed out some zeros there...50MHz Scenix ->
> > 50,000,000 instructions per second :)
>
> Wasn't really my day yesterday was it ? "ips" meant to refer to
> instructions per sample re Mike's 1000 "ips" of the 10MHz PIC
Thankz to all the folks who input on this one! Been reading some data sheets -
looks like ISD/Winbond is the way to go if you wanta keep it simple. ADPCM etc
... hmm, sounds a bit hairy. :| Sound quality doesn't have to be music/CD level
but needs a wider BW than speech. i'll roll a few ideas past PICers when i
sorta get settled on a scheme ... hah, i *also* reckon BD's a dork! :) Rich
dork, though. :|
thankz again - Debbie :)
> Use an ISD unless the quality must be utterly superb
> I've used the ISD2590 for years in a communication product for people who
EMBEDINC.COM> wrote:
