Home |
Search |
Today's Posts |
#361
|
|||
|
|||
|
#362
|
|||
|
|||
|
#363
|
|||
|
|||
|
#364
|
|||
|
|||
"Buster Mudd" wrote in message
Kalman Rubinson wrote in message . .. In article writes: Before i forget it, does anyone have thoughts about doing vocals while monitoring with headphones? IME, the critical ingredient is the monitoring mix - what the headphones tell you while you are performing. I'm chasing down some anecdotal information that suggests that monitoring with only the right earcup (which ends up on the left side of the brain) results in singers being more on pitch. Before this thread goes any further, let me point out that the basic premise, that the sound from each ear ends up in the other side of the brain, is untrue. The input from each ear is relayed to a number ob sites in the brainstem and, after the first relay, is represented on BOTH sides of the brain. That certainly holds true for the cerebral hemispheres; they are binaural. Excellent point. Maybe that explains why I hate tracking or overdubbing with one headphone earpiece on & the other off; I'm much more comfortable with both earpieces halfway on (raised up 1" or so) if I need to hear acoustic room sound. I think that could be true, but it's something different - basically Buster you've criticized and indicted your monitoring mix. There's something in the room you need that it does not include - perhaps the sound of your voice as *processed* by the room and your HRTF. |
#365
|
|||
|
|||
"Buster Mudd" wrote in message
Kalman Rubinson wrote in message . .. In article writes: Before i forget it, does anyone have thoughts about doing vocals while monitoring with headphones? IME, the critical ingredient is the monitoring mix - what the headphones tell you while you are performing. I'm chasing down some anecdotal information that suggests that monitoring with only the right earcup (which ends up on the left side of the brain) results in singers being more on pitch. Before this thread goes any further, let me point out that the basic premise, that the sound from each ear ends up in the other side of the brain, is untrue. The input from each ear is relayed to a number ob sites in the brainstem and, after the first relay, is represented on BOTH sides of the brain. That certainly holds true for the cerebral hemispheres; they are binaural. Excellent point. Maybe that explains why I hate tracking or overdubbing with one headphone earpiece on & the other off; I'm much more comfortable with both earpieces halfway on (raised up 1" or so) if I need to hear acoustic room sound. I think that could be true, but it's something different - basically Buster you've criticized and indicted your monitoring mix. There's something in the room you need that it does not include - perhaps the sound of your voice as *processed* by the room and your HRTF. |
#366
|
|||
|
|||
On Tue, 07 Sep 2004 17:33:00 -0700, Bob Cain wrote:
philicorda wrote: I think they took impulse responses of the mics in an anechoic chamber. That's what they say bu I have my doubts. It's easy to extract the IR's they use and I've done that for all their mics and the inverses. When I've compared the magnitude response of these to published curves for various mics and they look way more alike than the mics themselves usually do. I remember reading somewhere that they also got impulses to the capsules of the condenser mics by connecting the signal directly to the mic diaphram. I can't see how that would work, other than to get the eq of the mic electronics. Or perhaps they ran the capsule 'backwards' as an electrostatic speaker while recording it's output at the same time? Can't say either why it would have any relevance. I'm pretty sure it's a convolver with mic impulse responses.. any other way would be a very long way around. That's it. I don't think they use FIR's though. I have reason to think they are mapping the response curves, however they obtain them, to minimum phase IIR's. Do you think they use IIRs to keep latency down? My understanding of such things is a little vague. I can't say I've ever seen a graph that came with a mic that contains any detail whatsoever. They tend to have about a +-10db range and a thick black line. And you'll see pretty much the same curve if you extract one of the Antares impulse responses and transform it. Same smoothness, same shape as published but no thick black line. :-) Having a look at a .amm file in a hex editor, the part that deals with one pattern of a mic, seems to be about 200-350 samples long (If they are IRs with 16bit samples, they don't look like it, but the organisation of the file is quite clear). Is that long enough for an mic impulse response? I wonder if the IRs are rather generated 'on the fly' from a splined set of co-ordinates in that file... That would explain the smoothness. Based on actual measurement I can make my Sony ECM MS-907 mic sound so much like a pair of Sound Room MC012 cards in XY that I'd almost be willing to put them up in ABX. Not quite that confident, though, and it is well nigh impossible to set up an ABX for microphones. That's part of what makes it so easy for some to say just about anything they like about mics and classify their quality by how much they cost. Move off axis, or work in anything but an anechoic chamber, and all bets are off.... The patterns won't be the same. Also, put them over a kit as overheads and I bet they sound totally different, if only because they will crap out in different ways. (The pads on the octavas are not very nice). Then try a pair of KM84s instead of the octavas... Since real life recording situation variations are at least as great and usually far greater than the differences in patterns between the microphones, what I am saying is that what I get out of a transformed ECM-MS907 is a _very_ plausible MC012 XY. If you want to loan me a pair of KM84's and a Sony ECM-MS907 for about a week, I'll send them back to you with IR's that will transform MS907 recordings to a plausible facsimile of KM84 recordings that you can judge for yourself. If you decide to take me up on that, I will need a bit of setup time first to get my new hemi-anechoic measurement rig rung out. Pics he http://www.arcanemethods.com/Hemi_Anechoic/ The thing will be set up out of doors far from reflections. Cool. I'd guess that knowing the response of an exact set of source and modelled mics would work rather better than generic examples. Any particular reason for using a ECM-MS907? Also, what's the driver at the bottom of that strange device? And... What convolver(s) do you use afterwards? I have Acoustic Mirror and SIR. Sadly, I can't loan you a pair of KM84s, as I don't own any. I tend to work on a beg and borrow basis to supplement my meagre collection of mics. They do sound nicer than my octavas when I've got them. It's an interesting offer though. While I 'know' it should not be possible to turn an ECM-MS907 into a passable pair of KM84s, I also 'knew' that no softsynth could ever sound fat like a moog, and that all digital eqs sound nasty and harsh. Bob |
#367
|
|||
|
|||
On Tue, 07 Sep 2004 17:33:00 -0700, Bob Cain wrote:
philicorda wrote: I think they took impulse responses of the mics in an anechoic chamber. That's what they say bu I have my doubts. It's easy to extract the IR's they use and I've done that for all their mics and the inverses. When I've compared the magnitude response of these to published curves for various mics and they look way more alike than the mics themselves usually do. I remember reading somewhere that they also got impulses to the capsules of the condenser mics by connecting the signal directly to the mic diaphram. I can't see how that would work, other than to get the eq of the mic electronics. Or perhaps they ran the capsule 'backwards' as an electrostatic speaker while recording it's output at the same time? Can't say either why it would have any relevance. I'm pretty sure it's a convolver with mic impulse responses.. any other way would be a very long way around. That's it. I don't think they use FIR's though. I have reason to think they are mapping the response curves, however they obtain them, to minimum phase IIR's. Do you think they use IIRs to keep latency down? My understanding of such things is a little vague. I can't say I've ever seen a graph that came with a mic that contains any detail whatsoever. They tend to have about a +-10db range and a thick black line. And you'll see pretty much the same curve if you extract one of the Antares impulse responses and transform it. Same smoothness, same shape as published but no thick black line. :-) Having a look at a .amm file in a hex editor, the part that deals with one pattern of a mic, seems to be about 200-350 samples long (If they are IRs with 16bit samples, they don't look like it, but the organisation of the file is quite clear). Is that long enough for an mic impulse response? I wonder if the IRs are rather generated 'on the fly' from a splined set of co-ordinates in that file... That would explain the smoothness. Based on actual measurement I can make my Sony ECM MS-907 mic sound so much like a pair of Sound Room MC012 cards in XY that I'd almost be willing to put them up in ABX. Not quite that confident, though, and it is well nigh impossible to set up an ABX for microphones. That's part of what makes it so easy for some to say just about anything they like about mics and classify their quality by how much they cost. Move off axis, or work in anything but an anechoic chamber, and all bets are off.... The patterns won't be the same. Also, put them over a kit as overheads and I bet they sound totally different, if only because they will crap out in different ways. (The pads on the octavas are not very nice). Then try a pair of KM84s instead of the octavas... Since real life recording situation variations are at least as great and usually far greater than the differences in patterns between the microphones, what I am saying is that what I get out of a transformed ECM-MS907 is a _very_ plausible MC012 XY. If you want to loan me a pair of KM84's and a Sony ECM-MS907 for about a week, I'll send them back to you with IR's that will transform MS907 recordings to a plausible facsimile of KM84 recordings that you can judge for yourself. If you decide to take me up on that, I will need a bit of setup time first to get my new hemi-anechoic measurement rig rung out. Pics he http://www.arcanemethods.com/Hemi_Anechoic/ The thing will be set up out of doors far from reflections. Cool. I'd guess that knowing the response of an exact set of source and modelled mics would work rather better than generic examples. Any particular reason for using a ECM-MS907? Also, what's the driver at the bottom of that strange device? And... What convolver(s) do you use afterwards? I have Acoustic Mirror and SIR. Sadly, I can't loan you a pair of KM84s, as I don't own any. I tend to work on a beg and borrow basis to supplement my meagre collection of mics. They do sound nicer than my octavas when I've got them. It's an interesting offer though. While I 'know' it should not be possible to turn an ECM-MS907 into a passable pair of KM84s, I also 'knew' that no softsynth could ever sound fat like a moog, and that all digital eqs sound nasty and harsh. Bob |
#368
|
|||
|
|||
philicorda wrote: Do you think they use IIRs to keep latency down? My understanding of such things is a little vague. After what you said further down I'm not so sure now that they are encoded as IIR's but they most definitely are minimum phase. But latency is the reason for minimum phase. Of all the ways a filter can match a given magnitude curve, the minimum phase filter gives the smallest latency. Having a look at a .amm file in a hex editor, the part that deals with one pattern of a mic, seems to be about 200-350 samples long (If they are IRs with 16bit samples, they don't look like it, but the organisation of the file is quite clear). Is that long enough for an mic impulse response? That's a _whole_ lot of coeficients for an IIR. I don't know of any IIR design algorithms that could calculate that many, but the IR's that come out of it are certainly longer than 200-350 samples. When I run a single sample floating point impulse through it, the resulting IR has much greater than 16 bit resolution. I wonder if the IRs are rather generated 'on the fly' from a splined set of co-ordinates in that file... That would explain the smoothness. Not sure. They may be encoded some way other than IIRs. The size you see does sorta indicate that possibility. If you decide to take me up on that, I will need a bit of setup time first to get my new hemi-anechoic measurement rig rung out. Pics he http://www.arcanemethods.com/Hemi_Anechoic/ The thing will be set up out of doors far from reflections. Cool. I'd guess that knowing the response of an exact set of source and modelled mics would work rather better than generic examples. Always, but I still don't think the ones in MicModeler were done from actual measurement despite the claims and I think it can be done generically from measurement with some averaging over instnces to get much better results. Of course that presents some very serious aquisition issues (another reason I don't think all those MicModeler mics were measured.) Any particular reason for using a ECM-MS907? I like it. It's small, self powered, quieter than you would expect, mid/side, and very inexpensive. It's frequency response (mag/phase) is not at all good at either end so it presents an ideal demonstration source mic. It gives a superb stero image transformed or not. It's all I use now for live recording because I have the SW I mention below and have measured a few pretty good mics that I can transform the recordings to. I can flatten them to the response of the Earthworks TC30K I use as a "flat" reference mic for my measurements and sorta like it the best. You can't doubt the honesty of that transform if you were there when the recording was made. I was moving toward a PC SW product and wanted to show what could be done with the methods to a consumer grade mic when approached correctly but issues with the programmer I hooked up with terminated that collaboration. I have a (mostly working) engineering prototype for the PC (without source code) but the implementation doesn't really belong to me. There is a very nicely polished Mac OS X version (screen shots at my site) done with another, much more civilized and cooperative, collaborator but sundry individual impediments arose along the way and momentum was lost. The problem in the next paragraph is what has me stalled right now. I had considered just using MicModeler's IR's. I extracted them and built a working library for each of the two prototypes but have decided both that the ethics of that are highly questionable and that they aren't of sufficient quality. I was also down to the point where building a virtual mic cabinet for it presented problems I couldn't overcome. There's no way I can afford to buy an instance of all the mics that I'd need to have an attractive product and I'm not well enough known or connected to go borrowing them for a measurement (and turned around in a day.) Also, what's the driver at the bottom of that strange device? Not sure what you mean. And... What convolver(s) do you use afterwards? I have Acoustic Mirror and SIR. It could be done with either and an appropriate editor. A more convenient combination than that would be AudioMulch with SIR. In the product prototypes I have, it is done with internally implemented overlap-add fast convolution employing a library FFT. It's an interesting offer though. While I 'know' it should not be possible to turn an ECM-MS907 into a passable pair of KM84s, I also 'knew' that no softsynth could ever sound fat like a moog, and that all digital eqs sound nasty and harsh. Precisely. :-) Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein |
#369
|
|||
|
|||
philicorda wrote: Do you think they use IIRs to keep latency down? My understanding of such things is a little vague. After what you said further down I'm not so sure now that they are encoded as IIR's but they most definitely are minimum phase. But latency is the reason for minimum phase. Of all the ways a filter can match a given magnitude curve, the minimum phase filter gives the smallest latency. Having a look at a .amm file in a hex editor, the part that deals with one pattern of a mic, seems to be about 200-350 samples long (If they are IRs with 16bit samples, they don't look like it, but the organisation of the file is quite clear). Is that long enough for an mic impulse response? That's a _whole_ lot of coeficients for an IIR. I don't know of any IIR design algorithms that could calculate that many, but the IR's that come out of it are certainly longer than 200-350 samples. When I run a single sample floating point impulse through it, the resulting IR has much greater than 16 bit resolution. I wonder if the IRs are rather generated 'on the fly' from a splined set of co-ordinates in that file... That would explain the smoothness. Not sure. They may be encoded some way other than IIRs. The size you see does sorta indicate that possibility. If you decide to take me up on that, I will need a bit of setup time first to get my new hemi-anechoic measurement rig rung out. Pics he http://www.arcanemethods.com/Hemi_Anechoic/ The thing will be set up out of doors far from reflections. Cool. I'd guess that knowing the response of an exact set of source and modelled mics would work rather better than generic examples. Always, but I still don't think the ones in MicModeler were done from actual measurement despite the claims and I think it can be done generically from measurement with some averaging over instnces to get much better results. Of course that presents some very serious aquisition issues (another reason I don't think all those MicModeler mics were measured.) Any particular reason for using a ECM-MS907? I like it. It's small, self powered, quieter than you would expect, mid/side, and very inexpensive. It's frequency response (mag/phase) is not at all good at either end so it presents an ideal demonstration source mic. It gives a superb stero image transformed or not. It's all I use now for live recording because I have the SW I mention below and have measured a few pretty good mics that I can transform the recordings to. I can flatten them to the response of the Earthworks TC30K I use as a "flat" reference mic for my measurements and sorta like it the best. You can't doubt the honesty of that transform if you were there when the recording was made. I was moving toward a PC SW product and wanted to show what could be done with the methods to a consumer grade mic when approached correctly but issues with the programmer I hooked up with terminated that collaboration. I have a (mostly working) engineering prototype for the PC (without source code) but the implementation doesn't really belong to me. There is a very nicely polished Mac OS X version (screen shots at my site) done with another, much more civilized and cooperative, collaborator but sundry individual impediments arose along the way and momentum was lost. The problem in the next paragraph is what has me stalled right now. I had considered just using MicModeler's IR's. I extracted them and built a working library for each of the two prototypes but have decided both that the ethics of that are highly questionable and that they aren't of sufficient quality. I was also down to the point where building a virtual mic cabinet for it presented problems I couldn't overcome. There's no way I can afford to buy an instance of all the mics that I'd need to have an attractive product and I'm not well enough known or connected to go borrowing them for a measurement (and turned around in a day.) Also, what's the driver at the bottom of that strange device? Not sure what you mean. And... What convolver(s) do you use afterwards? I have Acoustic Mirror and SIR. It could be done with either and an appropriate editor. A more convenient combination than that would be AudioMulch with SIR. In the product prototypes I have, it is done with internally implemented overlap-add fast convolution employing a library FFT. It's an interesting offer though. While I 'know' it should not be possible to turn an ECM-MS907 into a passable pair of KM84s, I also 'knew' that no softsynth could ever sound fat like a moog, and that all digital eqs sound nasty and harsh. Precisely. :-) Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein |
#371
|
|||
|
|||
|
#372
|
|||
|
|||
Bob Cain wrote:
philicorda wrote: Do you think they use IIRs to keep latency down? My understanding of such things is a little vague. After what you said further down I'm not so sure now that they are encoded as IIR's but they most definitely are minimum phase. But latency is the reason for minimum phase. Of all the ways a filter can match a given magnitude curve, the minimum phase filter gives the smallest latency. If they are minimum-phase, they are not FIR filters. I have used some digital systems with FIR filters, and they are very strange to work with. They don't feel like "normal" EQ at all, and using them to try and undo minimum-phase problems doesn't work very well. But they are great for notching and slicing, for example. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#373
|
|||
|
|||
Bob Cain wrote:
philicorda wrote: Do you think they use IIRs to keep latency down? My understanding of such things is a little vague. After what you said further down I'm not so sure now that they are encoded as IIR's but they most definitely are minimum phase. But latency is the reason for minimum phase. Of all the ways a filter can match a given magnitude curve, the minimum phase filter gives the smallest latency. If they are minimum-phase, they are not FIR filters. I have used some digital systems with FIR filters, and they are very strange to work with. They don't feel like "normal" EQ at all, and using them to try and undo minimum-phase problems doesn't work very well. But they are great for notching and slicing, for example. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#374
|
|||
|
|||
|
#376
|
|||
|
|||
Scott Dorsey wrote: Bob Cain wrote: philicorda wrote: Do you think they use IIRs to keep latency down? My understanding of such things is a little vague. After what you said further down I'm not so sure now that they are encoded as IIR's but they most definitely are minimum phase. But latency is the reason for minimum phase. Of all the ways a filter can match a given magnitude curve, the minimum phase filter gives the smallest latency. If they are minimum-phase, they are not FIR filters. Any FIR (all zero filter) can be made minimum phase by reflecting all its z-domain zeros that are outside the unit circle to the inside. The magnitude response will be identical and the phase shift will be as low as it can be at every frequency (for a filter of that length.) If you mean that there will always be a longer filters that can give the same magnitude response with less phase shift and that the minimum minimum phase filter for a given magnitude response must be infinite then I think you could well be right. I've never seen that particular discussion come up but it certainly has a ring of truth to it. In a strict sense, no IIR can be implemented by an FIR, obviously, but real world IIRs nearly always have a tail that is below signifigance within a length easily handled by an FIR. If you just truncate and window, or often just truncate, the difference will usually be negligable with just some fractional DB and group delay differences at the lowest frequencies. Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein |
#377
|
|||
|
|||
Scott Dorsey wrote: Bob Cain wrote: philicorda wrote: Do you think they use IIRs to keep latency down? My understanding of such things is a little vague. After what you said further down I'm not so sure now that they are encoded as IIR's but they most definitely are minimum phase. But latency is the reason for minimum phase. Of all the ways a filter can match a given magnitude curve, the minimum phase filter gives the smallest latency. If they are minimum-phase, they are not FIR filters. Any FIR (all zero filter) can be made minimum phase by reflecting all its z-domain zeros that are outside the unit circle to the inside. The magnitude response will be identical and the phase shift will be as low as it can be at every frequency (for a filter of that length.) If you mean that there will always be a longer filters that can give the same magnitude response with less phase shift and that the minimum minimum phase filter for a given magnitude response must be infinite then I think you could well be right. I've never seen that particular discussion come up but it certainly has a ring of truth to it. In a strict sense, no IIR can be implemented by an FIR, obviously, but real world IIRs nearly always have a tail that is below signifigance within a length easily handled by an FIR. If you just truncate and window, or often just truncate, the difference will usually be negligable with just some fractional DB and group delay differences at the lowest frequencies. Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein |
#378
|
|||
|
|||
"Arny Krueger" wrote in message ...
"Buster Mudd" wrote in message Maybe that explains why I hate tracking or overdubbing with one headphone earpiece on & the other off; I'm much more comfortable with both earpieces halfway on (raised up 1" or so) if I need to hear acoustic room sound. I think that could be true, but it's something different - basically Buster you've criticized and indicted your monitoring mix. There's something in the room you need that it does not include - perhaps the sound of your voice as *processed* by the room and your HRTF. Actually, when I'm singing I rarely take the cans off; it's when I'm playing bass and/or (especially) when I'm playing with a drummer that I want to hear direct sound. Yeah, that's an indictment of headphone monitoring in general, sure. In a perfect world, headphones would have no place in the tracking room. |
#379
|
|||
|
|||
On Sat, 04 Sep 2004 16:51:25 GMT, "Bob Olhsson"
wrote: ... There's just as big of an ego trip available from using the cheapest, "good enough" gear as copping an attitude that only expensive gear is "good enough" for you. Of course, there are people to who your "cheapest" is "the best they could lay their hands onto", or at least "the most expensive they could afford". |
#380
|
|||
|
|||
On Sat, 04 Sep 2004 16:51:25 GMT, "Bob Olhsson"
wrote: ... There's just as big of an ego trip available from using the cheapest, "good enough" gear as copping an attitude that only expensive gear is "good enough" for you. Of course, there are people to who your "cheapest" is "the best they could lay their hands onto", or at least "the most expensive they could afford". |
#381
|
|||
|
|||
|
#382
|
|||
|
|||
|
#383
|
|||
|
|||
We have a global economy now. Which functiones in a way that quality goods always go one way while worthless dollars go another. |
#384
|
|||
|
|||
We have a global economy now. Which functiones in a way that quality goods always go one way while worthless dollars go another. |
#385
|
|||
|
|||
On Wed, 8 Sep 2004 15:08:05 +1000, in rec.audio.pro you wrote:
Adobe Premiere or Final cut pro anyway. Or Vegas. Or plethora of freeware stuff. |
#386
|
|||
|
|||
On Wed, 8 Sep 2004 15:08:05 +1000, in rec.audio.pro you wrote:
Adobe Premiere or Final cut pro anyway. Or Vegas. Or plethora of freeware stuff. |
#387
|
|||
|
|||
On Mon, 06 Sep 2004 22:25:56 -0400, in rec.audio.pro you wrote:
Mike Rivers wrote: I have known professionals who have issued recordings made on cassette with inexpensive mics. They didn't have the low lows or the high highs, but they had what it takes to make a professional recording. What, in your mind, determines whether a recording is professional or not? And I know professionals who have issued recordings on cassette tape without cassette. You by tape in small plastic bag. If you want to listen to it, you have to put the tape into empty cassette shell. |
#388
|
|||
|
|||
On Mon, 06 Sep 2004 22:25:56 -0400, in rec.audio.pro you wrote:
Mike Rivers wrote: I have known professionals who have issued recordings made on cassette with inexpensive mics. They didn't have the low lows or the high highs, but they had what it takes to make a professional recording. What, in your mind, determines whether a recording is professional or not? And I know professionals who have issued recordings on cassette tape without cassette. You by tape in small plastic bag. If you want to listen to it, you have to put the tape into empty cassette shell. |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
F.S. tons of gear for sale, keys, modules, pro audio, etc | Pro Audio | |||
common mode rejection vs. crosstalk | Pro Audio | |||
Topic Police | Pro Audio | |||
Artists cut out the record biz | Pro Audio |