Microphone for Grand Piano Recording

Discussion in 'Acoustic Keyboards' started by bf2008, Apr 12, 2009.

  1. OE1FEU

    OE1FEU Active Member

    Joined:
    Aug 27, 2018
    Location:
    Vienna
    As to the first part, can you be a little more specific as to how you would actually go about recording pink noise? From what source etc., please?

    I feel that the OKM, due to its tiny membrane already comes into saturation. While I usually use the OKMs for concert recordings and have them as boundary mics worn over the lapels of a jacket, in the recording presented here I wore them in-ear, as intended, so very close to the piano itself. So let's try to focus on the Rødes because I don't expect to use the OKMs for anything but recital bootlegs.

    As to the piano: No offense taken, quite on the contrary. As you can hear, I am not a 'real' pianist, but I have worked with quite a number of professional and renowned pianists as artist manager and had the privilege of speaking often and intensely about piano playing. Also, in the past couple of months I took a really deep dive into understanding piano technology, inspired by the movie "Pianomania", whose protagonist is considered the guru of all piano technician gurus. He was the one to appraise the piano and together with him I made a plan how to get the best out of the piano. In short: I have had the action completely reworked with new repetition levers, back checks. Also the piano has new hammers, original Steinway ones and they have been minimally adapted to fit the older parts of the action. Also, the strike line for the upper registers has completely been reworked, i.e. the hammer heads were removed and their position adapted to the best sound possible on the hammer shanks. The sound board has seen significant repairs over the years, but by now it is in a good equilibrium and I can honestly say that this piano has more power and sustain than any modern B I have played. The wood of the soundboard is still the original Appalachian fir, which is by many considered vastly superior to today's Alaskan Sitka wood.

    Altogether the piano has been regulated and tuned by real experts and I am confident that the dynamic range is really huge, both in ultra-pianissimo and fortissimi. And while it is a really loud beast, I can't get it into saturation, the limit being my own playing. I believe that a real Russian pianist or a marvel like Frederic Rzewski could squeeze another 3 dB out of it before actually hitting saturation.

    By now you may understand why I am so frustrated at the limited dynamic range of the recordings.

    For more details on the piano I have documented its arrival, appraisal and repair in a German piano forum:
    https://www.clavio.de/threads/steinway-b-seriennummer-60103-1887.23873/
     
  2. paulears

    paulears Well-Known Member

    Joined:
    Feb 7, 2014
    Location:
    Lowestoft - UK
    Home Page:
    Things like the usual editors - SoundForge, audition etc have a generator built in - but in the absence - record the hiss between FM radio stations - it will do for this. Whatever you use to edit/record on should let you create the test recording. I think that with the specialist work on the instrument you can rule that out then - I was thinking the opposite, maybe a worn out but nice sounding beast!

    If it can produce this range it should be perfectly possible to record it. The files you have are flacs, capable of really good quality - but what format did you record to? wav? Was there a file conversion that has gone wrong somehow? If one of them has more bit depth, and has been reduced to fit the other? To be honest, that's a total guess because my understanding is that with floating point, this doesn't happen - but maybe there is a setting that is doing the conversion - badly? Those OKMs I'm not really sure of. My experiments years before with binaural were not the best. I had a dummy head which was OK, but really needed headphone to make the stereo sound field work. On speakers it was pretty odd. Another test would be to try to record a continuous tone that started inaudible and increased at a steady rate - with one microphone left and another right, and see on a stereo side by side meter the two levels rising - they should track together. If one is performing badly against the other it should be obvious.

    Going back to the OKMs - be aware that the lapels of a jacket are nowhere near going to work as a boundary, as in an infinite plane that has the microphone as part of it. If the mics are more than a few mm away from the surface, they operate in a non-boundary mode, and of course, fabric is a great absorber of high frequencies. You also cannot move because you rotate the stereo field. I strongly recommend that you fix the mic position. The actual technique should reflect the sound you want. A semi-distant position with either Blumlein fig-8 or XY generally works best in nice sounding rooms, but in a troublesome or less, er, pleasant sounding room, a cheat with close mics, positioned by experimentation with the piano in question and then balanced and blended with some good quality reverb plug-ins may well be the best. My colleague's C3 records best close miked simply because like yours, his room has hard surfaces - his with carpet, but still not nice sounding. We record his piano close in - and use a rather quaint digital Yamaha reverb unit from the 80s - their very first attempt at acoustical modelling, and in this case, we use the one excellent programme - Munich Cathedral, which blended in low sounds great.

    If you oversaturate a microphone, overload it, if you will - the resultant sound is just rough. There is non-linearity when the levels go up to close to maximum, but this can easily be tested as the culprit by recording from an increased distance - in verse square law works for you so a doubling of the distance has a much bigger impact on record levels. Of course the room content increases, but it will let you test dynamic range.
     
  3. OE1FEU

    OE1FEU Active Member

    Joined:
    Aug 27, 2018
    Location:
    Vienna
    Her is something additional that might actually help:

    Internals.jpg

    So, this is a recording I've made with the internal microphones of the Tascam DR44-WL. Hopefully you can deduce some more information from this recording. Instead of finishing just with the chord, I've left it dangling for quite a while, which also made me realize that more than a minute of sustain is quite something for an instrument of that age.

    https://drive.google.com/open?id=1mBfLncvsKraWwX6608Jj_ndsBjm2D66C

    The recordings with the DR44-WL are all 16Bit at 44.1kHz. Conversion to flac via Linux command line with highest possible compression ratio; the original .wav is 21MB, the resulting .flac is 4.5MB.

    The OKMs have a very specific use case for me and as such they have been of incredible service for more than 30(!) years. Taping classical piano recitals is something you don't want to do with visible gear, so back in the 80s together with a WM-D3 this was precisely what I needed. The resulting recordings by now have kind of a historical dimension; I taped pianists like Horowitz, Richter, Arrau, Michelangeli and in some cases the recordings are the only existing ones, such as this one from 1987:



    When placed carefully, they reproduce a really superb piano sound, such as in this case with Russian pianist Igor Zhukov and a piano prepared by one of the best technicians in the world: https://drive.google.com/open?id=1heKuLniowM21wIMoNGXCSpOY81OhqTDU

    Recording is from 1996 with the OKMs and a Sony DAT TCD-D8. Please enjoy, it's gorgeous.
     
  4. paulears

    paulears Well-Known Member

    Joined:
    Feb 7, 2014
    Location:
    Lowestoft - UK
    Home Page:
    I've taken the tracks to my studio and while I'm not sure what the result actually mean, there are a few things that stand out that might be indicative of a problem your end.

    The recording of the recital with the applause at the end first. Your old standard to perhaps use as a guide?

    I tried listening in both Sound Forge and Cubase, to get access to different tools. The recital recording has a very solid stereo image that follows the pianist quite closely, and it is balanced to the centre, which I always think is the aim - centre keyboard = centre of the image. The pianist's perspective, which is of course wrecked for the audience, who don't get this with the usual piano placement. This recording has a swing about the centre, and the field when viewed looks quite typical. I'm never sure what is wrong with some I see, but usually something in the sound leads me to the stereoscope meters. The dynamics are also quite pronounced, but in a realistic manner. The unwanted mechanical noises are absent to the best of my ability to detect them.

    In your OKM recording, there are what sound like 'cuff-link' noises, fabric noises (I think - a sort of swish) and the stereo image is all over the place. This doesn't make itself very obvious in the recording, but I close my eyes and can detect shifts in the stereo image, so I look on the meter and it's visible - and in this case, there is a favouring of the lower part of the keyboard. I still like this recording technique the least.

    The Omni more closely reflects the type of sound in the recital recording - which interestingly I did not detect the audience until they clapped!

    Your new recording with the sustained end note shows that there is dynamic range capability. It also reveals some technical issues I think. The sustained note would normally be specially static. The instrument just decays. Obviously, the length of the strings determines that some will decay more quickly - but the image should stay fixed, perhaps with some tiny room anomalies in a space with large RT60 value, but you have a small space, so I expected stability. The recording shows the left and right decay varies - the meters fluttering and the stereoscope moving, which is odd. As if the two channels are not tracking together?

    The one thing you may have not noticed is that there is virtually nothing above 10K in ANY of the recordings. In fact, the energy is mostly below 4-5K, and tails off rapidly above this. A gentle slope downwards from just over 4K, with it almost gone by 6-7K, and missing above that. I changed the scale to check, and it's below the level my meters can show. The fact that the old recording and the new ones both have this suggests something other than your limited dynamics. The recorder and it's internal microphones often are commented on for being a bit bright - in your case, there is no HF response? Something is adrift here. Can you check this? Do you have any other recording device to use as comparison? Even an iPhone maybe? Something you can use to analyse what is going on? Could you record that piano decay on two devices - recorder and phone, sync them up and route to L and R and then see how the decay and frequency response differs. The sustain recording shows that the piano has a greater dynamic range than you thought perhaps, and you successfully captured it, even though the noise floor was evident. This didn't seem to decay 'weirdly', but was quite natural apart from the left/right flutter. Why was the original recording lacking dynamics that were recorded later? I'm left wondering if you just overplayed the quieter passages and ran out of steam? Seems unlikely, but the decay recording rather removes the piano and recording device from any blame of not being able to cope with dynamics, which only leaves the player. Could this have anything to do with the loss of the HF end? As you play louder, I assume the piano produces more overtones, so if the recorder/microphone system does not capture it, maybe this is what makes you hear it as lack of dynamics? Maybe?
     
  5. John Willett

    John Willett Active Member

    Joined:
    Apr 2, 2014
    Location:
    Oxfordshire, UK
    Home Page:
    I don't know why I haven't noticed this thread before as recording classical piano is my speciality......

    I always use omnis for recording a grand piano as a directional mic. rolls off too early and you miss the bottom end.

    I normally use the Gefell M221 and have also used Sennheiser MKH 20 or 8020 and also have some omnis in the Neumann KM-D series.

    As the OP was talking about B!!!!!!!r mics, I think the budget is very low - if this is the case, then the only mic. I would look at is the Line Audio OM1 as this is an excellent omni at a very small price that has a quality far above its price range.
     
  6. bouldersound

    bouldersound Real guitars are for old people. Well-Known Member

    Joined:
    Jan 23, 2010
    Location:
    Boulder, Colorado
    Home Page:
    I don't suppose the limiter in the recorder is engaged, is it?
     
  7. OE1FEU

    OE1FEU Active Member

    Joined:
    Aug 27, 2018
    Location:
    Vienna
    Sorry for taking so long to reply; lots of things going on in my life right now.

    I have pondered your excellent remarks and done some re-listening and further experimentation. By now I am confident that the culprit is the internal microphone preamps of the Tascam DR44-WL. While the recorder itself may have a huge dynamic range, I believe that the miniaturization of analog components inside the recorder simply makes it impossible to deal with the dynmic range of a piano and semi-professional microphones.

    As a first step to verify my opinion I bough a cheap Steinberg UR22-MK2 preamplifier with an inbuilt ADC that outputs to USB, feeding a laptop computer running Linux and Audacity as recording software. I also played around with microphone positioning and found the ORTF (200mm apart) set up a really interesting thing. I stumbled upon it with this recording that I really enjoyed, both piano-wise and in terms of mixture between direct sound and ambience:



    So, here are three different versions of the piece that by now all of you will have come to love-hate:

    Røde NT5 with cardioid capsules about 1.2m from the piano
    https://drive.google.com/open?id=17kHdfELLmRBQiSGdOaRczbuXAjioLovX

    Røde NT5 with NT45 capsules about 2m from the piano in ORTF set up:
    https://drive.google.com/open?id=1d6R6O-svxsCjEJgIFGNtNOq9HX97Z1HF

    Røde NT5 with NT45 capsules about 1.2m from the piano in ORTF set up:
    https://drive.google.com/open?id=1Kem9OvXjly0un81VDj0IkKGRtfYqgVuF

    I feel it's a distinct improvement over the previous attempts. I loved the total simplicity of setting up the Steinberg UR22; it was really just plugging USB into the laptop, switching to the right inputs on the pulseaudio-manager and hit record in audacity. Now, the Steinberg is a really cheap piece of equipment, clocking in at roughly 140 USD. I'd like to know whether there is something significantly better in terms of preamp dynamics/resolution and quality of the AD-converters with the same ease of use in terms of drivers for Linux. If you say there isn't, I'll probably just keep it as it is, but TBQH I can't really see me using the DR44-WL anymore for my homerecordings.

    I look forward to hearing your opinions; you have all been extremely helpful!
     
  8. paulears

    paulears Well-Known Member

    Joined:
    Feb 7, 2014
    Location:
    Lowestoft - UK
    Home Page:
    The size of components has no impact on the performance of a circuit that can be quantified. In fact, it's pretty standard practice for the prototypes of new circuit designs use different types of components to the production runs - you cannot experiment with circuit mounted devices - for the experimenting stages you need discreet solderable components so you can try substitutions and do the measurements. Once the design. is finalised, the transition to the final PCB design, with the tiny space-saving components is transparent.

    Its only in RF designs that the physical layout can have an impact on things like tuned circuits, with the tiny devices meaning closer parallel tracks which have impact on how the tuned circuits match.

    I'm sure there is a difference in the sound, but it will almost certainly be the design of the circuit, not the components that are playing the part in this.

    In the video clip image - I'm intrigued as to why the piano is not central - one of the microphones is entirely pointing at nothing?
     
  9. paulears

    paulears Well-Known Member

    Joined:
    Feb 7, 2014
    Location:
    Lowestoft - UK
    Home Page:
    This should go down as one of the trickiest listening tests ever. I found it IMPOSSIBLE to judge based on listening to them sequentially, so what I have done is download these files.
    Op11-1-Cardioid-ORTF1
    Op11-1-cardioid
    Op11-1-internal
    Op11-1-Omni-ORTF-1
    Op11-1-Omni-ORTF-2
    Op11-1-omni

    I have placed them one above the other in Cubase 9.5 Pro and been able to switch between them as it plays. However, I found the one I preferred in the beginning section to NOT be the one I preferred in the louder section, so cannot come up with an ideal, only best compromise. Looping just a short section of the quiet and loud sections allowed the simplest and I hope most critical listening.

    Quiet Section impressions
    Op11-1-Cardioid-ORTF-1 A bit thin, emphasis on right hand, not too much room sound
    Op11-1-cardioid Warmer, but left hand more prominent, and right hand a little metallic
    Op11-1-internal quite dry and clinical, but balanced - no noise I could detect
    Op11-1-Omni-ORTF-1 warmer, a little boomy at the bottom - maybe the room?
    Op11-1-Omni-ORTF-2 blurred - the room intrudes - a small rustling just before 7 secs??
    Op11-1-omni warm but right hand defined better

    Louder Section
    Op11-1-Cardioid-ORTF1 image moved left a bit? weird?
    Op11-1-cardioid warm but the repeated low note really intrudes
    Op11-1-internal sounded pretty unpleasant on the really loud notes over emphasised left right shifts
    Op11-1-Omni-ORTF-1 warm but manged the right hand better - the left hand just a bit too much?
    Op11-1-Omni-ORTF-2 the highest notes seem to cut through a little better
    Op11-1-omni cleaner right hand, some resonances in left hand.

    Which ones did I prefer? For realism, as if I had been in the room, I think I'd go for Op11-1-Cardioid-ORTF-1, the old recording. I liked Op11-1-Omni-ORTF-2 and switching between those two - the room just got bigger. Worst for me was the Op11-1-internal which just sounded 'wrong' - tonally weakest, and if I had to say why, I'd probably go with the mics - I don't believe it's the electronics just what they're fed with, or maybe it was just an unlucky placement.

    If it was me - I'd pick one of the two I liked, then experiment with placement to maximise the good characteristics and reduce the slight annoyances.

    I have also concluded that this is all down to personal preference and NOT anything technical at all. I liked some, didn't like others. While typing this, I've been listening to it all over and over - and if I have to have a winner, its Op11-1-Omni-ORTF-2, based on the entire piece.
    Paul
     
  10. Boswell

    Boswell Moderator Distinguished Member

    Joined:
    Apr 19, 2006
    Location:
    UK
    Home Page:
    The type of components and the layout are part of the design. I've related before in these forums how in one of my contract designs I carefully specified that the output feedback resistor had to be three discrete (not "discreet"!) axial-leaded resistors in series rather than a single surface-mount (SMT) resistor. This was because the resistance-voltage curve of the SMT parts was sufficient to introduce distortion at higher output levels. Sure enough, when the finished products came back from the factory in the Far East, the three discrete resistors had been lumped into a single SMT part, and the performance dropped from very good to average.

    BTW, designers do work at the surface-mount level, even for prototypes. During a development phase of a project I can spend a fair proportion of my days working down a microscope. Soldering devices with half-millimeter pin spacing takes a little skill and practice.
     
  11. paulears

    paulears Well-Known Member

    Joined:
    Feb 7, 2014
    Location:
    Lowestoft - UK
    Home Page:
    Interesting - I have never been organised enough to keep the separation required to swap and change - ending up with unidentifiable parts I can't then categorically say is the right bit!

    I suspect age and cutting-edginess dulls my ability to comprehend. I shall have to do some reading up to comprehend the electrical differences between an SMT component and 3 separate series connected individual resistors. I guess I'm looking for differences in resistance as voltage changes? Measurable differences I presume rather than theoretical? I take my hat off to you for having the ears for this. I'm positive that I could never hear the differences in combinations of resistors. As the designer, you deserve your designs being followed to the letter. I'm just having trouble imagining what I'd actually be listening for, and how my knowledge that passive components like resistors are not at audio frequencies anything other than resistive? Probably best as a new topic, rather than derail this one, but I'm just having real trouble understanding this one.
     
  12. Boswell

    Boswell Moderator Distinguished Member

    Joined:
    Apr 19, 2006
    Location:
    UK
    Home Page:
    Agreed - maybe we'll come back to that in a more appropriate thread.

    Regarding the piano recordings: what was not said about the ORTF samples is whether the capsules were the NT45-C or the NT45-O. The C would be correct for ORTF, but (as John Willett also mentioned), omni (-O) capsules usually give a much better overall feel to a grand piano recording, largely due to the extra octave on the bass. If using one of the standard named configurations with omnis for this type of recording, spaced A-B works better than a near-coincident mounting such as ORTF.

    In my experience of piano recording, you don't necessarily get the best sound from an instrument just by putting up a standard configuration in a pre-selected position. You need to experiment, either with moving a standard configuration around to find a position that captures a good sound from both the instrument and the stage acoustics, or else by going non-standard with the configuration and positioning microphones to get the best from the instrument while reducing any unpleasant reflections or other sounds from the venue acoustics.

    Regarding the Atsuko Kinoshita video, I'm sorry to say I did not personally find it particularly convincing. This was at both the performance level and the quality of the recorded sound, even taking into account the caption implying that she was playing an 1853 Bechstein.
     
  13. John Willett

    John Willett Active Member

    Joined:
    Apr 2, 2014
    Location:
    Oxfordshire, UK
    Home Page:
    I agree with this.

    Although my standard starting positioning is normally a 20cm spaced pair of omnis (Gefell M221 or Sennheiswr MKH 20 are normally my first choices) at about 2m high and 2m from the piano - I will listen first and vary the positioning.

    The best position will differe as to the piano, the room, the pianist and teh work being played.
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice