bass reconstruction processors

Yeah, I know what you mean.

gandalf91 wrote: By ratios resulting from the overtone series...it's really not all that fantastic as it seems. It's not assumption.

That assumes a non-integer harmonic - eg, if 60Hz & 90Hz are present, it could be assumed that 30Hz is the fundumental - but that is an assumption... It's a lot of Fourier series matching to determine "matched" harmonics - ie, that 60Hz & 90Hz are from the same source. I'd presume DSPs are used...

oldspark wrote: gandalf91 wrote: By ratios resulting from the overtone series...it's really not all that fantastic as it seems. It's not assumption. That assumes a non-integer harmonic - eg, if 60Hz & 90Hz are present, it could be assumed that 30Hz is the fundumental - but that is an assumption... It's a lot of Fourier series matching to determine "matched" harmonics - ie, that 60Hz & 90Hz are from the same source. I'd presume DSPs are used...

The ratio between any two given harmonics are always the same, regardless of the instrument. For example the first the ratio of the second harmonic and first harmonic is always 3:2. If 60hz was first harmonic and 90hz the second, then yes, 30hz would be the fundamental. Yes, they could be the sixth and ninth harmonic, resulting in a 10hz fundamental, but sixth and ninth harmonics are likely to be easily discernible from first and second, even by terms of relative amplitude attenuation. You bring up a good point about the Fourier series and matching or differentiating sources. I hadn't heard of that series before, but I'll read more in depth about it in the near future. All I'm saying is with all the math we've unveiled behind music down to the study of waves, and as you said yourself through the use of DSPs, I'm sure something like this could be accomplished. However, there is always room for error.

I think you are missing my point. And I am being consistent with your Wiki link (which is merely about basic harmonics and does NOT address the methods & algorithms for this device...). The device samples a low frequency. How does it know there should be a lower harmonic. (Assuming all higher harmonics are integer multiples.) Hence if you sample your domestic 60Hz (else 50Hz) AC, will it synthesize 30Hz? (Assuming no non-linear loads/interference, all AC mains harmonics will be multiples of the fundumental - namely 60Hz (else 50Hz).)     Is that clearer? PS - it has to be DSP but even then there will be assumptions - ie, it is algorithm dependent which IMO must be based on the type of music etc. I can accept recovery (eg, SN ratios <0dB), but the rest is Hollywood stuff (eg, resolving a number plate from a single or few pixels...). Sorry for my poor explanation, but to me DSPs are so last century....

oldspark wrote: I think you are missing my point. And I am being consistent with your Wiki link (which is merely about basic harmonics and does NOT address the methods & algorithms for this device...).

I know, but idealistic this would be the most basic starting point for developing the algorithms.

oldspark wrote: The device samples a low frequency. How does it know there should be a lower harmonic. (Assuming all higher harmonics are integer multiples.)

Because it would not only sample one harmonic frequency itself! If the fundamental was 30hz (we'll use the same example), it should register a 60hz band, 90hz, 120hz, etc. However, if 60hz was indeed the fundamental, a 90hz harmonic for that source would not exist. The first harmonic would be 120hz. The idea is not to just automatically synthesize an arbitrary note for every sound in the spectrum one octave lower.

oldspark wrote: Hence if you sample your domestic 60Hz (else 50Hz) AC, will it synthesize 30Hz? (Assuming no non-linear loads/interference, all AC mains harmonics will be multiples of the fundumental - namely 60Hz (else 50Hz).)     Is that clearer?

I'm not exactly sure what you are saying with this last part. Are you referring to the frequency of alternating current noise interference entering the device? If so, would these even bear any harmonics? (I personally do not know), and if it did not, no lower note would be reproduced. The idea being SETS of harmonics are registered, note a single random frequency. Harmonics can again, easily be distinguished between fundamentals by nature of their ratios between other surrounding harmonics. Now if there are multiple sources, I could see SOME room for error like you mentioned.

So the device does not introduce any lower frequencies from what you are saying. IE - it won't assume lower "integer" harmonics. I got the impression it did....

I'm not sure about lower harmonics. (For example, I don't know that it will fill in a first and second harmonic if a third and fourth is detected.) If the upper harmonics of a particular note are discernibly there though the point (as I've read and understand) is to fill in missing fundamentals (if they are missing). I believe if it detects the harmonics it will scan for the fundamental, and if it returns under a certain amplitude (dB) or is missing all together the device will fill it in. My concern with using them on this issue would be in cases where the fundamental should be sounding softer than the harmonics and whether or not it would pose a problem distorting the harmonic content too much. [For example, the lowest fundamental note on a bass guitar is ~41hz. However, in most recordings, the 82hz harmonic may be heard more loudly than the fundamental. Think about hearing a low electric bass note on a high end system. Now think of hearing a 41hz sine wave. The latter obviously sounds deeper. I would think a possible downside to the processors is that it could mess up the timbre of certain musical sounds.] I guess I still don't exactly get the last bit of what you're saying. What precisely do you mean "integer" harmonics? I guess I can say, it won't generate "undertones" (harmonics below the fundamental) because those don't naturally exist in music, but I don't know if that means anything along the lines of what you're saying. The math is really there though to make pairs of harmonics very distinguishable from a harmonic and a fundamental. The ratios are all constant no matter what the fundamental note is.

Integer is as per your wiki link - ie, 2nd, 3rd, 4th, .. nth harmonic - ie, "whole number". Not 3/2 etc. And now we are back to the start.... "....is to fill in missing fundamentals (if they are missing)". So how does it KNOW if there is a missing fundamental that has been filtered out? The answer is based on my replies - it requires DSP analysis and then assumptions ie, pre-programed algorithm(s). If it is inserting assumed fundamentals, then that can lead to amazing effects. (I suspect that is why the others are saying stay away, don't do it etc.) Detection of existing but quiet frequencies is merely amplification (albeit with noise reduction). And that has the relative amplitude issues that you describe. (Hence the need for recognition and then the pre-programmed algos - ie, that is an acoustic base 82Hz therefore..., or (partly) a grand piano 82Hz, therefore... And that recognition can only be done via full bandwidth assessment.)      I'm familiar with the type of intelligent programing that is required for that sort of stuff and I reckon I can almost certainly assure you that suitable hard programing does not yet exist. (Face and voice recognition is easier!)

"And that recognition can only be done via full bandwidth assessment" Ahhh, is this perhaps where our misunderstanding has been all along. To "correctly" use them, as I've read, you're supposed to run them as a preamp (even to any amplifiers you may be running for woofers, mids, tweeters, etc.). They only work by analyzing the full spectrum of sounds (even though they only affect the frequencies in the sub-range). Yes, there are still assumptions such as the amount of amplification to apply to missing fundamentals, but altogether missing fundamentals are not wild guesses/assumptions in this regard. "I can almost certainly assure you that suitable hard programing does not yet exist." I believe you. This is basically what I was asking from everyone in my OP. I knew the basics of how they work, just wanted to know how mature the technology was, or if it was still very crude. (The latter seems to be the general consensus). :P

gandalf91 wrote: ... but altogether missing fundamentals are not wild guesses/assumptions in this regard.

So yet again, HOW do they guess - I mean, "know"? (You must have read their method??) Or - do they or do they not insert non-existent harmonics (ie, the first harmonic aka the fundamental). [ You have indicated both.... ] If you have some link that explains it, cool. But your previous link does not - Wiki knows what the fundamental is. [ I was thinking of suppressed fundamentals with odd harmonics (flutes, clarinets, etc) versus odd & even (sax) or even only (square, triangular, etc), but that's too hard to tie in with that simple "single fundamental" Wiki example. ] Or maybe the bottom line, does that device employ a DSP, or is it merely some cute analog filters & mixers?