jmansion":g2mlotlz said:
sploo":g2mlotlz said:
Definitely! However what is slightly odd is the use of ecasound for XO and sox for dither, when it would seem that sox could probably do all of it:
http://sox.10957.n7.nabble.com/impl...ossovers-using-Sox-LADSPA-plugins-td5464.html.
So little time - so much to learn - I wish I could retire right away. :-(
Potentially the sound quality could be awesome (modulo the rank amateur designing the crossover). Consider the Pink Faun card driving 3-off I2S DACs. If you can get the resolution expanded to 32 bits on the way into the crossover system and then to the DACs after scaling, and if the system clock really is any good, then maybe the 'do the volume control after the crossover' stuff becomes unnecessary. I confess I eBay'd an Audiolab 8200AP largely to do that role, also hoping that HDMI might be better than OK with it. When I bought the 8000Ps they really were based in Huntingdon which is quite close to me, but I think now its a warehouse and IAG is largely run from China.
(And in defense of their recent products, I bought a new Q-DAC for 200 quid and I love it as a USB audio device, also to drive headphones)
Anyway - do we get the record for going off-topic on a woodworking board? Can't see why anyone would buy a router except to make round holes for drivers, personally! ;-)
[On going off topic, better to be hung for a sheep, etc...]
The cynic in me suggests that you won't have any audio sources good enough to make any of the above worth the candle. It's one of the differences between the pro- and amateur approaches: is it "good enough" to serve its purpose? If so, use it!
It's also instructive to look at the entire recording chain, from microphone to listener's ears, and see where the most egregious distortions of the original sound occur.
The fact you can make nth order filters doesn't mean they are nice in use, or that they don't introduce issues of their own. My dad, who was an excellent hi-fi designer in his own right, created some very interesting three-way active speakers, but found he had to significantly reduce the slope of the crossovers because the effect of too-steep ones was just nasty. IIRC, he settled on 12dB/octave.
Many of the fundamentals* don't change in a digital world: for speakers, how resonant or otherwise is the enclosure, and where on the spectrum? Likewise the drivers - how peaky are they, and how do they tail off out of band, and most importantly, how well damped is everything?
Back EMF has been mentioned - if the amp's source Z is correct (i.e. low enough) it won't be significant. There's also been mention of diffraction at HF - actually irrelevant unless you're designing something with slot apertures for a specific purpose (some BBC designs used this at LF, but not HF). HF resonances do make an audible difference, as does driver phase alignment at MF and HF (but you have to work out what part of the driver you are phase-aligning to!), but it's all marginal and difficult to measure objectively.
Talking of which, do you have suitable test equipment? Sir will be needing a decent condenser mic, with a known (meaured) frequency response curve, and a test acoustic in which to use it, ideally anechoic (within band), and something to calibrate against. Sir will also be needing a spectrum analyser that sir can trust (just because it can doesn't mean the A2Ds are accurate and so on...)
In the industry, 24/96 isn't used because it delivers better quality as such, it's used (a) because it's a sales gimmic, and (b) amongst the cognoiscenti because you can put such a signal through a lot of DSP with less degradation (essentially rounding errors) before you finally "downgrade" it to 16/44.1.
And I'll take a lot of persuading that anything you do will compensate for the horrible noise that is most MP3...
Have fun by all means, but don't waste your time pursuing a mirage.
E.
*see what I did there?
