|
Post by jazznoise on Aug 25, 2015 2:21:07 GMT -6
The old fashioned way, I thought, was a phase comparator. If the signal consistently has more signal on one side of the rail to the other the allpass filters are shifted up or down until the voltage difference between the top and bottom is made as small as possible.
|
|
|
Post by Pueblo Audio on Aug 25, 2015 11:29:47 GMT -6
On reflection, I believe my interpretation of that square wave test is not totally correct. I ran another test with a sine wave + dc offset and an added spike. Here is the pic of that after +45deg shift of the right channel (left untouched for reference). As I observed with the square wave, there does not seem to be a time shift (notice the spike, while distorted, remained in place). With this sine wave, it is easy to observe the apparent 45deg shift, which was not evident with the square wave. Unexpected! Additionally, the DC offset has been affected this time - indeed, making the sine wave more symmetrical about zero crossing. Interesting. I am not familiar with this type of circuit so I can not provide further insight as to what the devil is going on. If memory serves, this process has it origins in analog radio broadcast. Announcer's voices could be asymmetrical about ground thereby over-modulating before all available headroom was utilized. The phase rotator could process that signal so more signal swing could be utilized. Even though the process generates artifacts/distortion, having the hotter signal for reaching a wider audience was probably a greater priority. Besides, the distortion would probably not have been noticed much due to the limited fidelity of radio itself.
|
|
|
Post by Pueblo Audio on Aug 25, 2015 11:41:50 GMT -6
So if i see it right, it is more like a kind of proprietary phase distortion algorithm to probably fix potential phase distortion problems of a source track? I don't think so, smallbutfine, see above. Now the question; is how useful is this process in modern music production? Popmann has done some experimenting already and seems to have found a beneficial application for it (and thanks for sharing your findings!) I would still be cautious. Don't look at this as a tool for correcting phase problems (because it can't). Rather, I feel it is closer to a sound design tool. Something to get a particular effect, not as a standard mixing process.
|
|
|
Post by jazznoise on Aug 25, 2015 13:17:05 GMT -6
Yeah, the concept of the Phase Rotator was for maximum modulation. Compressors were crude beasts and could not move fast enough to knock the louder cycles down to the same size as the others without essentially just being a clipper. So this method + some compression + a diode limiter was the best way to get maximum modulation in the days of yore, when your main worry was getting the station loud without going into other people's bands and getting shut down & fined.
Pueblo this is very interesting. Square Waves literally drive phase comparators nuts afaik because if the windowing is supposed to be reacting fast enough to deal with single transients, it could end up seeing the waveform itself as 2 DC states rather than an oscillation. See what it does to a saw or something very asymmetrical like a Trumpet or Sax.
|
|
|
Post by Pueblo Audio on Aug 25, 2015 15:02:05 GMT -6
I have a direct-to-2track jazz record I engineered on hand. Flugelhorn front man, acoustic bass, piano, drums. The flugelhorn is upfront and its asymmetry big in this very natural sounding mix. I applied the "adaptive phase" setting of the phase rotator which did make the signal symmetrical. But, oh my stars and garters, the transients are smeared, bass is hollow, presentation is boxy. More importantly, the artist's personality and the ensemble's "swing" is lost in translation.
|
|
|
Post by popmann on Aug 25, 2015 17:22:25 GMT -6
Oh, never allow that adaptive thing. Talk about weird. If you want THAT sound, go grab that Pi Mixer plug in--I think it only inverts, but it dynamically inverts "as needed" on every channel in the mixer so one bass note might have the polarity one way and the next 180 out.... Maybe I should've mentioned that. Nothing good comes from that, IME--I actually wonder why they have it there. Like what situation would dynamically even flipping the invert switch (which is basically what the Pi Mixer does) help? So, if anyone is interested in experimenting....lord, don't use that dynamic/adaptive algo....unless what you want is total audio weirdness unlike you've ever heard. In which case--go for it. It has the same adaptive option on the azimuth, but even if dropouts are such that I need to apply a different setting to some section than another--it still needs to be static for the sections of tape....sometimes I think maybe they just include options like that because they can....
|
|
|
Post by NoFilterChuck on Aug 26, 2015 6:25:23 GMT -6
when your main worry was getting the station loud without going into other people's bands and getting shut down & fined. See what it does to a saw or something very asymmetrical like a Trumpet or Sax. I'm curious. how would amplifying a signal so much cause it's transmission frequency to change? i don't know much about how radio transmits signals, so feel free to be verbose. Sax doesn't produce asymmetrical waves (at least when I record myself on a LDC, i don't have that problem). Brass definitely do.
|
|
|
Post by jazznoise on Aug 26, 2015 17:25:51 GMT -6
when your main worry was getting the station loud without going into other people's bands and getting shut down & fined. See what it does to a saw or something very asymmetrical like a Trumpet or Sax. I'm curious. how would amplifying a signal so much cause it's transmission frequency to change? i don't know much about how radio transmits signals, so feel free to be verbose. Sax doesn't produce asymmetrical waves (at least when I record myself on a LDC, i don't have that problem). Brass definitely do. My experience is that most wind instruments have some offset. But maybe it depends how you mic it, I've only mic'd sax a handful of times. To answer properly we need to consider what is radios used method of transmission. The oldest of the old is AM and the second to that is the naive earlier implementation of FM, we're a long way in some senses but not many in others. In AM you basically modulate the signals amplitude thousands of times per second in order to make a high frequency enough wave to actually transmit. Full modulation comes from effectively modulating from 100% down to 0%. Your main worry is overmodulation, which will cause weird phase distortion. I'm not sure it could interfere with other stations though unless you modulated high bandwidth white noise. However FM or frequency modulation is way messier in that you literally wiggle the radio frequency by the musical frequency. The carrier signal sends the data via a spectrum generated by this process, and the bandwidth of this wiggling mess is basically described by a bessel function, which describes the generation of sidebands surrounding the carrier frequency. Essentially the modulated signal will look something like this: So as you modulate via the audio, you get more of these peaks above and below at higher amplitudes. Great! But you have a 200Khz bandwidth or so before your little peaks touch the other stations little peaks. This will fuck up everyone's day and a mad-as-hell broadcast engineer will be in to yell you you've turned 600Khz of radio bandwidth into useless white noise.
|
|
|
Post by din on Dec 6, 2021 15:52:25 GMT -6
Stumbled upon this thread while researching whether or not I should get a Little Labs IBP for dual mic guitar setups. It didn't help me decide, but this thread blew my mind. I've been at this a long time and probably picked up the "high pass what you don't need" sort of ethos early on and never gave it another thought. In the last couple years I came to the conclusion that I would no longer do that guitars, which I'm really picky about, unless there was something problematic because it did something I did not like; though, I couldn't put my finger on what it was. To me, it just seemed like it made them sound slightly more digital or hollow, for lack of a better description. I wasn't even doing it aggressively either. I only used to high pass at 75Hz or 35Hz w/ -24dB slope, but still, I did not feel it was worth the trade off. Finally, I have an explanation for why it sounded off to me in frequencies beyond where I rolled off! I will much better about forgoing high pass filters, unless necessary, from now on. So thank you! I also feel weirdly bummed out because now I'm going to be overthinking whether or not I should employ all these nice analog EQ's I've spend a fortune on.
While I've got this thread necro'd, I wondered if I might posit a few more questions to the experts, if they're still around. How does dynamic EQ play into this? Say I have a dynamic EQ high pass filter set to just take care of rumbles like from when a singer steps on the mic stand or whatever, would the dynamic EQ only cause phase shift for the brief period when it's working? What about multiband compression/expansion? Does that cause phase shift?
|
|
|
Post by mcirish on Dec 6, 2021 16:14:15 GMT -6
I use faTimeAlign to make adjustments for phase. The way I do it on drums is this: 1) Set the scale to samples (instead of bars and beats) 2) Record some drums 3) Use the overheads as the guide and measure the time between a snare hit on a snare mic channel and the overheads (with the range tool) 4) Use that number of samples in faTimeAlign to get the snare mic and the overheads to be perfectly in phase It's worked well for me, though sometimes I wonder if just flipping a phase switch might be quicker with similar results. I do something similar to get the inside and outside kick mics to line up. It does make a huge difference in how much low end and punch you get from snare and kick
Forward Audio
As far as EQ and phase shift, it's going to be frequency dependent. Often I will group all the kicks to a sub and then EQ that sub, to avoid the individual and different phase shifts I would get from having EQ on both channels. Not a deal breaker at all. I've got great results both ways but I go back and forth on it. Alsays trying to improve.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 7, 2021 7:17:03 GMT -6
While I've got this thread necro'd, I wondered if I might posit a few more questions to the experts, if they're still around. How does dynamic EQ play into this? Say I have a dynamic EQ high pass filter set to just take care of rumbles like from when a singer steps on the mic stand or whatever, would the dynamic EQ only cause phase shift for the brief period when it's working? What about multiband compression/expansion? Does that cause phase shift? It's as simple as this, polarity operates on the basis of positive and negative (like sound pressure or voltage). Phase is just a function of time.. Where many seem to get confused is if you delay a signal 0.5 ms you're essentially flipping it 180 degrees. But the signal is still out of time, that's a phase shift not a polarity inversion. Normal EQ's work by shifting phase which is again time based, as long as it's on the EQ will affect the signal. There is such a thing in the digital domain as linear phase EQ but generally they will induce latency and Multi-band compressors usually cause phase shift at their crossover point. The question I pose to you is how much does it matter? There were some damn good recordings made on a console in the late 70's / 80's (when noise became less of an issue etc.), they didn't avoid EQ because of some phase shift and back then linear EQ's didn't even exist.. I mean if a track doesn't require EQ or a HPF then cool, leave it.. Going into the subjective department here I'm quite the advocate for less is more on occasion.
|
|
|
Post by din on Dec 7, 2021 7:50:09 GMT -6
The question I pose to you is how much does it matter? There were some damn good recordings made on a console in the late 70's / 80's (when noise became less of an issue etc.), they didn't avoid EQ because of some phase shift and back then linear EQ's didn't even exist.. I mean if a track doesn't require EQ or a HPF then cool, leave it.. Going into the subjective department here I'm quite the advocate for less is more on occasion. Good point. I think reading this whole thread in one sitting made me unnecessarily paranoid about using EQ. Haha! Have to remember that, ultimately, if it sounds better, it is better. I think the main lesson here, for me, was that high pass filters aren't 'free'--they come with trade off that I wasn't listening for. In most cases I was using HPF's more as a misguided rule-of-thumb without realizing they were imparting some negative effects. I mean, why would I need anything below 50Hz on vocals? So I engaged the filter on the mic or whatever. I guess now I'll be more deliberative about that sort of thing.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 7, 2021 8:34:34 GMT -6
My guess is that this rotation function was quite a bit simpler. I'm pretty sure there was a variant of this in the old 480L and I remember putting it somewhere in the PCM90/91. It's really a polarity rotator that's quite cheap CPU-wise. Assuming Lt and Rt are the outputs and L and R are the inputs. Let's express the rotation in degrees rather than radians, simply for simplicity. It goes like this:
Lt = [L * cosine(degrees)] + [R * sine(degrees)] Rt = [R * cosine(degrees)] + [L * sine(degrees)]
That's it. Cheap as chips. It's an ancient trick that long ago fell out of any patent protection. A limited variant of this is often used for signal widening. It is most definitely NOT mono-compatible, but the polarity inversions that happen as you change rotation can be pretty spacy. In an old matrix surround setup (like Dolby ProLogic 4) you could move stuff through the surrounds as you diddled it.
|
|
|
Post by Guitar on Dec 7, 2021 13:10:31 GMT -6
din are you sure you resurrected the right thread? There was a mega blow out dozens page phase thread a while back. And this short one in a few pages started by popmann. You can always use an EQ such as SSL X-EQ2 for frequency specific all pass filters! But after reading this particular thread... why would you? Well, "Sometimes it sounds better." Some people call this "phase scrambling." Airwindows has a free plugin for this too. If you use the TDR Slick EQ GE, there's a phase scrambling button for the bass filter. I think this effect is more audible in low frequencies due to the "size" of the waveform, the length of them over time for low frequencies to complete one cycle. Sometimes it hurts, sometimes it helps, mash that button and give a listen. I guess I agree with the idea in this particular thread that if you're chasing all pass filters and phase shifts, you're probably chasing your tail. Maybe you could use a Lissajou meter as a reference? Simple polarity shifts and micro-time delays should be enough for most "fixes" until you go back and re-track or whatever.
|
|
|
Post by din on Dec 7, 2021 19:28:41 GMT -6
din are you sure you resurrected the right thread? There was a mega blow out dozens page phase thread a while back. And this short one in a few pages started by popmann. You can always use an EQ such as SSL X-EQ2 for frequency specific all pass filters! But after reading this particular thread... why would you? Well, "Sometimes it sounds better." Some people call this "phase scrambling." Airwindows has a free plugin for this too. If you use the TDR Slick EQ GE, there's a phase scrambling button for the bass filter. I think this effect is more audible in low frequencies due to the "size" of the waveform, the length of them over time for low frequencies to complete one cycle. Sometimes it hurts, sometimes it helps, mash that button and give a listen. I guess I agree with the idea in this particular thread that if you're chasing all pass filters and phase shifts, you're probably chasing your tail. Maybe you could use a Lissajou meter as a reference? Simple polarity shifts and micro-time delays should be enough for most "fixes" until you go back and re-track or whatever. Oh dammit! I definitely responded to the wrong thread. Not sure how that happened. I meant to post to this one. realgearonline.com/thread/12687/low-hpf-bottom-octave
|
|
|
Post by the other mark williams on Dec 7, 2021 19:40:40 GMT -6
Man, I was confused as hell on this thing. Thanks for clearing it up Guitar! I'm glad you brought this stuff up again, though, din!
|
|