A better quality ISD (Winbond) chip..anything new ?

If you need playbackl only, you may want to look at using something like a ST serial flash (M25p series) and a cheap audio DAC (e.g. TI, AKM) It is possible (not trivial, but possible) to stream audio from the flash to the DAC at 31.25K sample rate with a cheap PIC running at 8MHz.

I make audio systems for museum exhibits, interactives, theme parks etc. Depending on the application, I have 2 approaches - if it just has to make a noise, and the source material is crappy (derived from a cassette for instance) I use an ISD4004 (or multiples thereof) at 8 minutes apiece (using the fastest available sampling rate). As long as I apply careful compression to the recording process this is fine for speech (the s/n of conversations between spitfires and dispatcher during the Battle of Britain was never the greatest in the first place). If better quality is called for (like music for instance) I use a VS1001K MP3 decoder, A CF card and a microcontroller - all the quality you need at 320kbps. What I'm looking for is a "middle ground"

- better than the ISD but not as complicated as the VS. Not overkill at all, just enough kill :-) M

Hello Moby,

TI used to give out "retro watches", little experimenter boards with a high power MSP430 on there. I believe one of the applications was audio storage. 60dB would be a stretch though, guess you'd have to hang an external ADC onto it but that should not be a big deal. You also need an external memory.

Regards, Joerg

The ISD chip gives less than 4khz bandwidth, NOT 8 (that is the SAMPLE rate). I want 10kHz bandwidth (ie > 20khz sample rate). These things are often used with phone style handsets and even my ancient ears CAN hear the difference of both the noise floor and HF response here. I want the things to sound substantially better than talking Xmas cards. The producers of the audio tracks put considerable effort into them and it is they who are the final arbiters, not Joe Punter, who doesn't give a rats :-) M

Hello John,

It depends on where it'll be used. Automatic PA announcements in upscale stores, cars, airplane cockpits might make the whole product or location look quite cheesy if it were done at 4kHz and 8 bits. It wood be like Naugahide seats in a Bentley.

Regards, Joerg

I read in sci.electronics.design that Joerg wrote (in ) about 'A better quality ISD (Winbond) chip..anything new ?', on Wed, 14 Sep 2005:

I'd be interested to know why the OP wants the performance specified. It seems like overkill.

--
Regards, John Woodgate, OOO - Own Opinions Only.
If everything has been designed, a god designed evolution by natural selection.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Hello John,

On airplanes only after the engines spooled up. And a Bentley or Rolls Royce shall never be noisy inside, at least not when considering the money that needs to be plunked down for one of these. Even my rather modest SUV is surprisingly quiet inside.

Regards, Joerg

I read in sci.electronics.design that Joerg wrote (in ) about 'A better quality ISD (Winbond) chip..anything new ?', on Wed, 14 Sep 2005:

Yes, although noise from 8 bits might well be a greater defect than 4 kHz. Cars (even top-class) and airplane cockpits are NOISY; 60 dB S/N is quite unnecessary.

--
Regards, John Woodgate, OOO - Own Opinions Only.
If everything has been designed, a god designed evolution by natural selection.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

I read in sci.electronics.design that Joerg wrote (in ) about 'A better quality ISD (Winbond) chip..anything new ?', on Wed, 14 Sep 2005:

You may be surprised, but you may be even more surprised if you put a sound level meter inside. In an exceptionally quiet room in a residential area, you may get down to 35 dB A-wtd. I suspect that your SUV does about 45 dB A unless it's coasting with closed throttle on a really smooth surface. Step on the gas and all bets are off.

Same with aircraft, even in the first-class cabins that you travel in. (;-) The meter does not lie, but your ears (and mine) are easily fooled.

--
Regards, John Woodgate, OOO - Own Opinions Only.
If everything has been designed, a god designed evolution by natural selection.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

I'd even suggest that if he's looking for good intelligibility, that a high-pass at about 300 Hz could help dramatically, at least for human voice. They've been doing that for decades in the Ham radio field. (you throw away the fundamental, so that it doesn't hog all of your transmitter power - all of the "intelligence" is in the harmonics anyway.)

Cheers! Rich

Hello John,

It doesn't become much more noisy if you step on the accelerator, just for the brief moment the computer needs to adjust the mixture.

Probably true. However, I can notice a marked difference between cell phone (on the car speaker) and AM radio quality, then another from AM to FM and if it's a really quiet ride even between FM and CD.

Regards, Joerg

Succinct answer: I hate ISD too. There is cause enough for this merely in their (unstated) datasheet errata, let alone the actual chip specs.

I've been asked recently to do some similar projects (vending machine applications, mostly). I've chosen to use an SD card (in MMC/SPI mode) to store raw PCM samples. I use an Atmel ATmega8 with an R-2R on one of the ports to do the DACing. Using a parallel 16-bit DAC seems to be an intractable problem on an 8-bit micro because of the impossibility of updating both bytes at the same time.

I come from a toy background (Winbond's other, larger line... among other vendors) and the 8-bit 12kHz samples sound pretty good to me.

I read in sci.electronics.design that " snipped-for-privacy@kcbbs.gen.nz" wrote (in ) about 'A better quality ISD (Winbond) chip..anything new ?', on Wed, 14 Sep 2005:

No, really, you won't get any perceived quality improvement by going from 8 kHz to 10 kHz, especially if you aren't spending $$$$ on loudspeakers. And think about the acoustic background noise in your space. It's probably around 45 to 50 dB A. So unless you are putting out

100 dB or so *average*, not peaks, 60 dB S/N is pointless; the acoustic noise masks it.

--
Regards, John Woodgate, OOO - Own Opinions Only.
If everything has been designed, a god designed evolution by natural selection.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

I read in sci.electronics.design that Joerg wrote (in ) about 'A better quality ISD (Winbond) chip..anything new ?', on Wed, 14 Sep 2005:

1 dB you wouldn't notice. 3 dB you'd just notice. Beg, borrow or acquire a sound level meter and measure what happens. I'd guess +6 dB.

But you should probably get a 'sports muffler' (i.e. loud), to give you a bit of a thrill for all the gas you are burning. (;-)

--
Regards, John Woodgate, OOO - Own Opinions Only.
If everything has been designed, a god designed evolution by natural selection.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Hello John,

After a kaboom while at the army I guess it's more like 6dB for me.

The Mitsubishi doesn't burn much. 25mpg on California gas, 28mpg on Nevada gas. I think you guys calculate in liters per 100km and I believe it would come in under 10. For a car that weighs almost two tons that's not so bad. I think the automatic version would be worse.

Regards, Joerg

Opps, I misread you, sorry Yes 8k b/w would probably do it..

Funny you should mention that - In one particular room I have over a dozen units and the only one that has ever given any problems (a couple of failures in 4 years) is right under a spotlight. What I find weird is that in battery powered locations, they speed up at the end of battery life (squeech!). M

I've found the Mega8 fast enough that you can bit-bang your 16 bit samples out fast enough into a PCM56P (must try out some of those new tiny serial DACs that AD and Maxim are putting out these days). The ISD data sheets are a masterpiece of misdirection but at least they basically work, unlike the Invox clones that came out a few years ago (then vanished). The CS4231/AD1848 codec chips do 16 bit mono, stereo etc and have a byte wide interface (have used these in combination with

30 pin SIMMS for digital delays etc). M M

I read in sci.electronics.design that Joerg wrote (in ) about 'A better quality ISD (Winbond) chip..anything new ?', on Wed, 14 Sep 2005:

Not in UK, and we have bigger gallons. Are those figures average, and do you do a lot of journeys over 20 miles?

--
Regards, John Woodgate, OOO - Own Opinions Only.
If everything has been designed, a god designed evolution by natural selection.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk