Talk:Digital speaker

A couple of things that have been preying on my mind regarding this article. One is the "unweildy size" issue, which starts with the arbitrary "a reasonable 2 square inch (13 cm²) driver". Considering that these drive units will all need a bandwidth exceeding 20 kHz, and none need to move much air, a 1 cm² area would be perfectly sufficient - leading to a roughly 6.5 m² total area. Large, but not too large for lavish homes. The other is the anti-aliasing issue. After first reading it, I tried to read up on it and understand it here on WP, and it seems to me that it does not apply - and even if it did, the artifacts would be mirrored around the upper cutoff frequency, and only present in the air, not the drivers. I may be wrong on the second point, but on the first there is no question. Huw Powell 23:11, 12 September 2007 (UTC)

Some key issues with this article
A good start has been made, but I think it needs work to address some fairly key issues.

1. A minor point, but becomes significant in following paras: An area of 0.5 cm^2 would be perfectly practical for the smallest speaker element. 2 sq inches is plainly not a sensible LSB size to choose for such an array.

2n. The array size may be massively reduced by varying the throw of each bit as well as area. This can give more than an order of magnitude of space saving.

3. The real problem with this system is actually that reasonable amounts of area and throw are required to produce satisfactory bass, and these physically large bits can not be driven at ultrasonic frequencies using moving coil technology. To get such areas working above a.f. it is necessary to either break them up into lots of small drivers or else turn to more exotic speaker technologies that can produce high magnitude ultrasonic output.

4. One variation option that addresses the above is to let the biggest bit be a standard woofer, which moves at a.f., thus produces unwanted a.f. output each time it moves, and use microphone plus nfb to get the other smaller bits to counteract and cancel this af output. (How well they would do so is an issue.)

5. It is incorrect to state the ultrasonics can't be filtered from the audio output. Placing an uncoupled diaphragm over the whole array would act as a low pass filter. This would not compare well in performance with electronic filters, but (acoustic) filtering is certainly possible. with good design and multiple passive diaphragms it would even be possible to remove most of the ultrasonics.

6. The number of bits of the speaker system has more or less nothing to do with CD compatibility. CD music can be replayed on anything from 4 bits upward. For speaker arrays of less than 16 bits, the least significant bits of cd data are simply ignored, and quality is a little less. This is no big problem, for example the once popular compact cassette typically achieves apx 6 bit resolution. For speaker arrays of >16 bits, the least significant bits of the speaker are not driven. Note that this does not degrade the sound quality in any way, unlike the previous case.

So yes, the article needs work. Tabby (talk) 14:12, 23 January 2008 (UTC)

"Drastic modification" needed?
The following paragraph was at the top of the article. I moved it here. 210.50.223.65 (talk) 12:08, 24 July 2008 (UTC)

''This article needs drastic modification, since digitally driven speakers without any form of an analogue signal path are now, indeed, to become reality. It is no longer true, as it previously stated in this article, that all modern speakers are necessarily analogue. For more information on this, refer to http://www.patentstorm.us/patents/6492761/description.html''

PC speaker
After reading this article and the article PC speaker, I think, that the PC Speaker is a 1-bit digital speaker. --MrBurns (talk) 18:01, 8 September 2010 (UTC)

Original research?
There seems to be absolutely no references in this article, beyond the external article referencing the existence of digital speakers as a concept, and it seems to me like it is original research WP:NOR - virtually none of the content in the article is present in the reference.

I would like to add that some of the "problems" listed, such as the required size of the are quite likely wrong, such as the required size of the drivers - for instance, there are ribbon tweeters in commercial products that are capable of ultrasonic output well into the range required for 44.1 KHz sample rate, with sufficient SPLs for listening purposes. See http://www.adelaidespeakers.com/monitors.html and/or http://www.visaton.com/en/chassis_zubehoer/ht_baendchen/mht12_8.html for examples of drivers with response beyond 40,000 Hz. 16 of the MHT ribbon tweeters would, sans the mounting brackets, in a 4x4 configuration be 240x180 mm large - pretty precisely the size of an iPad. Such a "large-box" system would then simply use different voltages for each of the 16 bits

With power handling up to 100 W continuously and an assumed sensitivity of 90 dB/W/m, the MSB driver would be capable of a max 110 dB SPL (not counting any losses induced by the acoustic/mechanic low-pass filter, which is plenty for most domestic applications.   Arve (talk) 23:33, 1 October 2013 (UTC)