Have you compared the audibility of this compared to the alternate? If the result is the same it should sound the same but seems like a lot of unnecessary processing?
I favor separate amplification + standalone dsp for a number of reasons.
For experimentation, plate amps are too much work to fit in until you know you have a final design that's a keeper.
For keepers, they are still hard to fit in, hard to seal, and hard to make sure of proper ventilation.
All eggs are in one basket. DSP fails, no sound, with no way to replace dsp. Amp fails, no sound, with no way to replace amp.
They need to be sized channel wise and power-per-channel wise, to a particular design.
Extra weight for the speaker.
AC power has to be available to every speaker, along with ability to turn them on and off easily. And of course they still need a line level feed too.
Not sure how good vibration is for them.
Can't sell speaker without selling plate amp also.
>(plate) amplifiers with integrated DSP
I suppose that means you dont choose your DAC, they do. If analog input, that means they choose your ADC too. I'd think these plate amps would suffer the "MiniDSP V1 syndrome", where the final sound depends on their choice of ADC and DAC - and all the supportive peripheral circuitry - which are going to have a character you'll hear, on top of or confounded with the sound of your primary DAC - if digitally sourced.
A plate amp with a digital input gets rid of one bridge, but you still are subject to the plate amp manufacturers choice of "DAC" - the signal has got to get back into analog to get to the speakers at some point! With all the fuss in audio over DACs, I'd think that would be a huge issue with DSP plate amps, what with considerations such as is the clock jitter well tempered?
I suppose that means you dont choose your DAC, they do. If analog input, that means they choose your ADC too. I'd think these plate amps would suffer the "MiniDSP V1 syndrome", where the final sound depends on their choice of ADC and DAC - and all the supportive peripheral circuitry - which are going to have a character you'll hear, on top of or confounded with the sound of your primary DAC - if digitally sourced.
A plate amp with a digital input gets rid of one bridge, but you still are subject to the plate amp manufacturers choice of "DAC" - the signal has got to get back into analog to get to the speakers at some point! With all the fuss in audio over DACs, I'd think that would be a huge issue with DSP plate amps, what with considerations such as is the clock jitter well tempered?
Last edited:
Admittedly, there are many comments about his preferences / tastes in music.
I think classical music in general is an excellent reference point.
It is also true that the vast majority of music today is a far cry from the golden age of classical, jazz, etc. in terms of recording and production techniques > krivium could probably write a book about it.
Linkwitz designs are somewhat more WAF-friendly compared tho the big BD horns, though I personally much prefer the looks of the latter.
Core aspects are both simplicity and sensitivity (headroom). With those 2 basic elements properly implemented, only the quality of the recording is the bottleneck.
Ha ha. You deserved way more than one "like." Bunch-a Philistines for failing to see the wisdom in your suggestion if you ask me.
For reference here is an example of a DIY amplifier with integrated DSP.
In this case you get to use Sigma Studio. Unfortunately, it is more difficult to use than the consumer friendly software included by device manufacturers. I've seen comments in this thread regarding limitations on EQ slots and I think, "not a problem in Sigma Studio." Also comments about input vs output. You can control things like that once you're in Sigma Studio because you aren't limited by the consumer friendly software.
Personally, I'd lean toward separate DSP + amplification because I lean toward Bluetooth & Wireless. Not necessarily in practice but in principle. At least, that's how I think it's going to roll in the future. We'll have a central DSP processing point (e.g. a personal computer/cloud service over 5G) then broadcast a wireless signal to the speakers. Power the multi-channel speakers with either a lithium ion battery or 24VDC lines with 12V step downs.
In practice, I'm integrating DSP into the speakers. I have a two-way CBT column that requires 6 channels of DSP. With a central DSP I have to run 6 separate speaker cables to the speaker. If I made a two-way 5 bank CBT I'd need 10 speaker cables running to the speaker. If I integrate the DSP into the speaker then I can run one power cable and one speaker cable to each speaker. Downside is that makes EQ adjustments more difficult -- unless there is a way to update the DSP chips wirelessly.
Now that the Dutch&Dutch 8Cs, KiiThrees, and BeoLab90s are in the market more DIYers will be looking for more powerful DSP upgrades and conveniences to replicate those devices. I suspect some people will want to do CBT array Beolab mashups, too.
Yeah, I figured it wouldnt take that much -
- Negotiate with the content owners for said tracks comprising a song
- Develop a technology to keep consumers from stealing artists work, at least while still in the digital domain...
- It's only bandwidth, which Amazon has in spades already.
Oh Well. I can imagine all the discussion threads and replies referring to so-n-so's mix of some particular song, defined in some file attachment anyone could download and listen to by importing it into Amazon's exclusive player.
"Next Big Thing" they wanted. I tried. But at Amazon, you dont "try", you do. The place really is wide open for people to create and define their own job - as long as 100 other employees havent thought of the same thing first.
don't wanna hyjack the discussion, but I'd like to know peoples' thinking on the philosophical conundrum: 1. "central processing & lots of (sometimes stupidly-expensive) cables at interconnect or speaker level" vs 2. "a more distributed approach" where one pair of (possibly stupidly-expensive) interconnects goes to speakers, & "all the fun" happens at the speakers, without cables that cost more than the processors/amps.
It's DIY, so we can do whatever we like. But what "architecture" seems to be the way of the future?
It's DIY, so we can do whatever we like. But what "architecture" seems to be the way of the future?
Thanks Mark for the explanation, and thinking about it more, I guess in reality provided the target slope is always lower than what the speaker would naturally be doing at that particular point, in reality what reaches the speaker will not be forcing it to do something it wouldn't naturally. I should have been thinking of it in a subtractive rather than additive way.
Ie the final transfer function rather than thinking of the dips being boosted, the cut is lessened. At least I think that's what is happening
And yes I can see the Robot going rather crazy over the crossover not being in place!!
Tony.
Ie the final transfer function rather than thinking of the dips being boosted, the cut is lessened. At least I think that's what is happening
And yes I can see the Robot going rather crazy over the crossover not being in place!!
Tony.
I guess the acid test is to compare the waveforms that are being fed to the speaker from the two different methods. If they are essentially identical then there should be no possibility for difference. My comments are still being made with a certain level of understanding absent
Would it be possible to get the same acoustic target using different waveforms, (thinking about resonances here, and differing ones being excited or not) probably not, but if it could be done, would they sound different?
Tony.
Would it be possible to get the same acoustic target using different waveforms, (thinking about resonances here, and differing ones being excited or not) probably not, but if it could be done, would they sound different?
Tony.
Thanks Mark for the explanation, and thinking about it more, I guess in reality provided the target slope is always lower than what the speaker would naturally be doing at that particular point, in reality what reaches the speaker will not be forcing it to do something it wouldn't naturally. I should have been thinking of it in a subtractive rather than additive way.
Ie the final transfer function rather than thinking of the dips being boosted, the cut is lessened. At least I think that's what is happening
Tony.
Cool Tony
I've found it very interesting, approaching different ways of achieving a desired acoustic target, and how they all end up with the same electrical transfer function.
That's the acid test ime/imo, that electrical transfer functions are identical.
I test equivalence using a dual channel FFT to compare one filter set as reference to the other filter set as test.
If the resultant transfer function is not flat line zero, mag and phase, they are not the same.
Someone asked the question, which is the least complicated method, the 'nudge method' or the flatten 'out-of-band and add xover method'.
I think it just comes down to available filter count in the dsp. If sufficient EQ's exist, the flatten method is both easier and also gives better results ime.
The better results are probably due to the simple fact EQing a curve into becoming a flat line, is easier than EQ'ing a curve into becoming a different curve.
Vertical mag distance is very easy to judge against a flat line, vs very deceptive to judge curve to curve.
In-band EQs are the same task for both methods of course, but matching out-of-band nudge EQ's is plain hard ime.
It's also worth noting, the nudge method requires acoustical driver measurements at every step along the way, with every trial and error.
The flatten method can be done entirely via electrical measurement, with the finished filter set ready for verification of the driver's acoustic ouput.
Hmmm...after saying that I may have just realized a way to do the nudge method entirely electrically too.
But it would require a dual channel FFT.
If the reference signal were an electrical xover output that is the desired acoustic output of the driver, it seems the measured raw driver response could be nudged into a flat line transfer function.
I gotta check this out, cause i think it could be an easy way to solve dsp filter shortages (when trying to use the flatten method). And also be easy because dual channel would allow real time nudging.
Anyway, hope that all makes sense.
But I'll still say (over and over haha
), the flatten method along with steep linear phase xovers makes everything else seem very complicated. (And for me at least, has given superior sonic results.)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.