Friday, April 13, 2012

Piano miking demo

(updated on March 30, 2014 to fix a broken link)

For the 255 Recording Engineering and Production class we took a trip to our sister studio located in the back of the new xMPL Theatre. It really is a fraternal twin to the Meyer Studio except it was designed and is run by Professor Chris Dobrian - Head of the Integrated Composition Improvisation and Technology (ICIT) program in the Music Department. We routinely share facilities, equipment and classes between Sound Design and ICIT. This studio is simply called the "Recording Studio." Like the Meyer, it is equipped with a ProTools HD system, Icon/D-Command control surface and Meyer HD-1 monitors. Chris is fortunate in that he doesn't need to house as much crap in his studio as we do downstairs; it's a comfortable and productive space.

The goal of this session was to listen to different methods of miking a grand piano. I consider piano miking to be the bellwether test of recording ability -- if you you can master all of the challenges presented by recording a piano in stereo (and deliver a great sound) you have the chops to mic up just about anything. The Recording Studio has a Yamaha 6' Disklavier piano. For those unfamiliar with the Disklavier, it's a standard piano with "player piano" mechanical control built in. You can send it MIDI, SMF files, or it will record and playback on its own. For this test we used a piece from the Yamaha PianoSoft library of files for the Disklavier: "I've Got The World on a String" by Koehler and Allen, and performed by Shelly Berg. The Disklavier allowed the same performance to occur repeatedly. Without it, several dozen mics would have to have been stuffed into the piano at the same time and there would be no way to use the same mics and preamps over again. This piano is a stellar instrument and records very well.

I chose the mics based on past experience - with a bias towards those that work well. Here is what we used:

(2) Neumann U89 (large diaphragm, multipattern, condenser, flat, no pad)
(2) Neumann M149 (large diaphragm, multipattern, tube condenser, flat, no pad)
(2) Neumann KM184 (small diaphragm, cardioid, condenser, matched pair)
(2) DPA 4006 (small diaphragm, omni condenser, matched pair)
(1) Earthworks PianoMic (dual omni condenser on built in bar)


The recording chain utilized mics plugged into a Grace M802 preamplifier connected to Protools analog inputs running at 96K and 24-bit, internal sync (sorry audiophiles, no external A/D or clocking for this session). The Grace is a lovely sounding preamp - quiet and clean with just a little bit of high frequency graininess. I would use this on a piano again in a heartbeat. However, the M802 has one stupid flaw for a $5K box -- it causes audible clicks in the audio when changing gain live. Grace could learn something from Millennia or Avalon about "grace-fully" quiet knobs!

All recordings were done with the mics hard panned L and R and flat (with the noted exception of one EQ example). I suggest you listen with the ability to collapse to mono as many of these methods suffer from awful phase cancellation. The mics had pads off -- they all fared well but the KM184 did sound a little edgy but I don't think it was overloading. The cardioid patterns all suffered from proximity effect and you should take that into consideration when comparing -- there is some significant bass boost going on. All high pass filters were bypassed (the M149 was set to 20Hz).

The high resolution files are clean and 96KHz/24-bit WAV (but a hefty 1.4GB download). Feel free to download them from my Dropbox:


You can also listen to them in my SoundCloud:

https://soundcloud.com/hookmusic-1/sets/uci-sound-design-piano-miking

Special thanks go to the sound grads for helping set up these examples and to Jeff Polunas for snapping pics with his iPhone. Kudos!

Here we go...



TEST ONE
Mic - U89, cardioid
Method - spaced cardioid
Position - close to the strings
Lid - full stick

The left mic was in the middle of the piano where the low and mid strings intersect. The fundamental is stronger at the middle of the sting, which leads to a rounder sound. The right mic was closer to the hammers (near the dampers) at about 2 octaves above middle-C. Diaphragms were about 6" from the strings. This is a very standard mic position for popular and jazz piano recording - especially where bleed from other instruments is a concern. The two clicks that you hear are not sync problems - they were tiny mid-stream gain adjustments on the Grace preamp (sigh).

VERDICT:
The U89 has always been my favorite piano mic and it performed great! The sound is super full, not too strident in the highs, and has a ridiculous stereo spread (albeit u
seless in mono). Some strong proximity effect going on leading to some overly loud damper thumps.

LISTEN:
U89_spaced_cardioid_close_fullstick_24_96.wav


TEST TWO
Mic - U89, cardioid
Method - XY
Position - just inside the piano
Lid - full stick

The XY array was inside the piano curve by about 1' with the left pointed at the low strings and the right pointed at the hammers

VERDICT:
XY miking relies on the timbre differences between left and right. Typically XY delivers an OK stereo image but only in the mids and highs with no bass separation at all. This is clearly audible in the recording. It's stereo but slightly pinched sounding. However, it's superbly mono compatible (and boring).

LISTEN:
U89_XY_inside_fullstick_24_96.wav


TEST THREE

Mic - U89, cardioid
Method - ORTF
Position - just inside the piano
Lid - full stick

The ORTF array was inside the piano curve by about 1' (same place as the XY).

VERDICT:
ORTF miking is like an XY but has the capsules oriented at 110 degrees and about 7" apart. It adds something really exciting to the stereo imaging and that excitement is TIME! The spacing of the ORTF capsules makes for a hybrid of XY and spaced pairs - using timbre, level and arrival differences to create stereo. It can be a little smeary above 10K in mono and sometimes leaves a hole in the middle, but in the case of this piano I think you'll agree that it is a much more pleasing sound than the XY.

LISTEN:
U89_ORTF_inside_fullstick_24_96.wav


TEST FOUR
Mic - U89, cardioid
Method - XY
Position - just outside the piano
Lid - full stick

The XY array was outside the piano curve by about 1' and height was about half way from the frame to the lid. The capsules were stacked on top of one another in time-coherent fashion. The left mic faced the back of the piano and the right faced the top octaves of hammers.

VERDICT
It's not as "roomy" as everyone expected but you can definitely hear the midrange smear from the lid. It's a very unobtrusive sound that might sound good in a classical application or backing up a soloist. My past experiences the the U89 have been that it never sounded good at a distance and really shines up close.

LISTEN:
U89_XY_outside_fullstick_24_96.wav


TEST FIVE
Mic - U89, cardioid
Method - spaced cardioid
Position - at the hammer
Lid - off/fully open

The left mic was just above and slightly forward of the dampers with the hammers underneath. The harmonics are stronger at the ends of the strings and this leads to increased articulation of the low notes. The right mic was also close to the hammers (near the dampers) at about 2 octaves above middle-C. Diaphragms were about 6" from the strings. Miking the hammers is another popular style for music requiring a very bright and present sound. It is even more striking sounding on a piano prepared with hardened hammers or even a tack piano.

VERDICT:
This sound is super clean. A lot of that has to do with the absence of resonance from the lid since it has been propped up out of the way. It is also aided by the brighter sound at the end of the strings. The higher frequency noise made by the felt dampers is more evident here.

LISTEN:
U89_spaced_cardioid_hammers_lid off_24_96.wav


TEST SIX
Mic - U89, omni
Method - spaced omni
Position - above the frame edge
Lid - full stick

The mics were placed about 15" apart right at the curve and directly above the frame with a slight downward tilt (yes, I know they are "omni").

VERDICT:
Just like its older, nastier and larger U87 cousin, the U89 is not a good omni mic. Something about the way the two capsules sum together just doesn't work right and has a colored, slightly harsh sound. If you need an omni I suggest using an acoustically-derived single-diaphragm omni and not a electronically-derived dual diaphragm disaster. However, for the scope of this test the omni mode on the same mic used as cardioid isn't necessarily unfair.

The sound has pronounced room in it but seems to be quite a bit less bass-y. It's got a midrange honk to it and a somewhat vintage character. In case you were wondering, the room is very dead - with 2" fiberglass panels on every vertical wall, carpet and ceiling tiles - and about 15' x 20'. The stereo is very natural and subtle. Spaced omni pairs derive imaging almost solely from timing differences from left to right. 15" spacing is about right to capture stereo without a hole in the middle but not enough to present low frequency imaging. This recording is completely mono compatible if you just mute either of the channels.

LISTEN:
U89_spaced omni_outside_fullstick_24_96.wav


TEST SEVEN
Mic - U89, omni
Method - spaced omni
Position - underneath
Lid - who cares?

Sorry no picture here. The mics were placed about 32" apart underneath the piano. Placement was difficult because there is a crapload of electronic boxes and wiring underneath this piano (and a pair of speakers!). There was also a piano dolly frame.

VERDICT:
Discussions among engineers about miking pianos from underneath have been the cause of many bar fights throughout late audio history. I take the side of it being a stupid thing to do unless you want something dull and boxy sounding. However, many FOH engineers swear by it for maximum gain in a noisy environment. Don't pick a fight with me over this - you will lose!

But... I have to admit that this piano actually doesn't sound half bad underneath. That also means it also sounds half good. You can hear the awful resonance of the sound board but I think some of it could be EQ'd out. It has zero attack, though. I think it sounds like a towel was tossed over the mic.

Unfortunately, our C-Ducer contact tape mics are in a show picking up tap dancing on a platform. A contact mic on the soundboard is something that I do like a lot. It doesn't sound natural and actually makes the best grand sound like a Yamaha CP-80 electric piano (it's a piano with 2' strings that packs up into a road case -- early Keane songs use one a lot, so does "That's All" by Genesis) but that sound is really retro and unique. Maybe we'll do another recording with the C-Ducer and compare at a later date.


LISTEN:
U89_spaced omni_underneath_24_96.wav


TEST EIGHT
Mic - U89, omni
Method - spaced omni
Position - outside the piano
Lid - short stick

The mics were placed about 15" apart well outside the piano and at the height of the gap between frame and lid.

VERDICT:
Short stick is often the result of sightlines in the pit or on stage. However sometimes it is called for by choice in some music circumstances. A good short stick recording either requires a wood floor to pick up the soundboard or it requires that the mics be placed inside the piano and really close to the strings. For this test I chose to put the mics outside so that we could hear what the lid does. It sounds similar on the inside because you are hearing the same higher resonance (boxiness) caused by the cavity created by the lid. This recording is devoid of any excitement.

LISTEN:
U89_spaced omni_outside_shortstick_24_96.wav


TEST NINE
Mic - Earthworks PianoMic
Method - pre-spaced, random-incidence omni mics on a bar
Position - inside and very close to the strings
Lid - full stick

This mic was placed 180 degrees (reversed) from what Earthworks recommends. We did it to get the elements closer to the hammers.

VERDICT:
The PianoMic is the greatest recent mic development that I can think of. It is a serious problem solver because it does a lot of things that have been challenging until now:
1. Allows mics to be placed with the lid fully closed with little resonant issues
2. Allows about 9 dB more gain-before-feedback on stage than closely placed cardioids.
3. No worries of mics falling into the strings or scratching the underside of the lid.
4. Doesn't present any precedence effect bass buildup.
5. It sounds great flat!

The PianoMic is befuddling because (when you look at it) your engineering brain says it can't possibly sound good with really tiny diaphragms, omni elements, etc. It's probably best for you to listen for yourself. It has a slightly overly-present character (slightly edgy high end) that is balanced by a general openness and really great imaging. It has detail that the U89 can't even come close to resolving - probably due to the perfect impulse response of those tiny diaphragms and their very small mass. It does struggle to deliver low end and sounds slightly EQ'd in this regard but that is compensated by its very punchy sound. If I had the time, I would love to try combining this mic with a set of nice room mics.

For those wondering, these are random-incidence (RI) omnis. RI mics are essentially omnis with a special high boost curve that compensates for sounds that are diffuse-field - like out in an auditorium. Most RI mics use an acoustic equalizer created by the shape of the grid that covers the capsule.

RI or not, all this comes crashing to a halt when you find out that it costs over $3K!

LISTEN:
Earthworks PianoMic_hammers_fullstick_24_96.wav


TEST TEN
Mic - M149, cardioid
Method - spaced cardioid
Position - close to the strings
Lid - full stick

These mics were nearly in the same position as the U89s were in test #1 - but up another few inches to try and avoid some of the proximity effect. They are an unwieldy size to stuff into a piano.

I made a huge goof with this recording by not engaging record in time. The first note is lopped off! Sorry.

VERDICT:
The M149 is a modern reinvention of a series of classic tube mics from Neumann (Neumann copying Neumann). They actually did manage to capture a lot of the old tube mic character - even-harmonic saturation (warmth), high-mid presence boost, huge proximity effect bump (lows) - and none of the tube mic problems like slow warmup, instability, capsule drift, and microphonic distortion (ringing from excess LF SPL). It is a caricature of an older mic and has an incredible coloration that can't be achieved any other way with processing. It's sounds fat and bigger than life.

Just as expected, the sound was warm and huge - almost ridiculously so - but it would be imminently EQ-able to fit into most mixes. The proximity effect is just too large and the highs are pronounced - the curve would look like a huge smile. The stereo imaging is wide with this placement but not as striking as the U89. Damper thump was very pronounced - as was some of the higher damper noise. But that warm tube gooeyness makes you forgive everything - just like a cute puppy that just piddled on the floor.

LISTEN:
M149_spaced_cardioid_close_fullstick_24_96.wav


TEST ELEVEN
Mic - M149, omni
Method - spaced omni
Position - just outside the frame
Lid - full stick

The mics were very wide (about 2').

VERDICT:
Like the U89, the M149 makes for a less-than-optimal omni. Uneven response front/back and side/side. However, it's better than the U89 in this application.

The sound was warm but not huge. This is probably from the lack of proximity effect. The stereo imaging has a hole in the middle with the notes clumping left or right.


LISTEN:
M149_spaced omni_inside_fullstick_24_96.wav


TEST TWELVE
Mic - KM184
Method - ORTF
Position - just inside the piano
Lid - full stick

The ORTF array was inside the piano curve by about 1'

VERDICT:
The KM184 is the modern revision of the classic Neumann KM84. I think the KM84 is a nicer mic but the KM184 is still an excellent small-diaphragm condenser. Typical of other mics in this class, it struggles to capture low frequencies and it has a very broad and high presence peak. This is a mic that sounds flat at a distance and hyped up close.

This was a favorite of much of the class. The sound had much better transient response (clarity and articulation) than the U89 and a pleasingly forward and bright sound. The ORTF imaging worked really well in this combination. The image is very stereo but not overblown. I think the piano sounds smaller here but with a modern edge.

LISTEN:
KM184_ORTF_inside_fullstick_24_96.wav


TEST THIRTEEN
Mic - DPA 4006
Method - spaced omni
Position - inside the piano
Lid - full stick

VERDICT:
The DPA 4006 is a stunning mic. It is super flat, very fast, handles high SPL and has one of the most neutral sounds of any mic - almost to the point of being boring to listen to. These happen to be mics from our Meyer SIM3 measurement system and were hand-matched 5-wide.

The piano sounded very different with these mics. Arguably, it sounded the most like the piano did live. There is a very open character but still sounds close-miked. The clicks in the middle of the recording are from the damn Grace mic preamp gain adjust.

LISTEN:
DPA4006_spaced omni_inisde_fullstick_24_96.wav



TEST FOURTEEN
Mic - U89
Method - spaced cardioid
Position - close to the strings
Lid - full stick
(same track as test #1)

This is the same recording as test #1. However, it has been EQ'd with what would be a very typical curve to make it sit well amongst other tracks in a popular style song. This exercise was to demonstrate the problems that can be solved with EQ, if needed. The U89 needed these fixes.

PROCESS:
Starting from low to high, we corrected imbalances and talked about the cause of each...

THUMP - the proximity effect of the U89 led to enhanced thumping sounds from the dampers and some very hyped low fundamentals in the bass. This was countered with a gentle 1st order (-6dB/octave) high pass filter.

RESONANCE - pianos have a resonant character, which isn't very audible from a distance but becomes problematic with mics up close. Typically this resonance falls between 200-400 Hz. It is a very wooden tubby/boxy sound. This was reduced by a -2.2dB cut at 213 Hz about 3/4 octave wide.

LACK OF SHEEN - pianos easily become lost in a mix with the typical solution being to bring them up in level. This can lead to muddiness in the mix as it competes with other instruments or voices. Sometimes a better approach is to brighten up the top end. Instinctively you might grab a parametric band and this can lead to problems where certain harmonics jump out on specific notes or groups of notes. A smoother way is to use a shelving EQ. In this case we rolled it down all the way to 2K and applied a gentle 1.6dB boost. This is similar in sound to adjusting the high crossover output on a reinforcement speaker.

RESULT:
The piano track took on a very present and well-balanced sound. Compare this to track #1 to hear the flat version.

LISTEN - EQ VERSION:
U89_spaced_cardioid_close_fullstick_EQ_24_96.wav

LISTEN - FLAT VERSION:
U89_spaced_cardioid_close_fullstick_24_96.wav


I hope you enjoy this demo, too. Let me know if you would like more of this kind of posting here on our blog.

Cheers....

-Mike Hooker

5 comments:

  1. This is a great post... but I can't get any of the sound files to play, even after trying multiple browsers.

    I can tell you that I am LOVING the Earthworks PianoMic for the tour of Here To Stay: The Gershwin Experience.

    ReplyDelete
  2. This comment has been removed by the author.

    ReplyDelete
  3. Great job!! Thanks so much for sharing. I've just bought a matched pair of RODE NT55 mics for recording my Mason & Hamlin BB, and your post is exactly what I needed to start up supported by some basic knowledge. Thanks again, bro!

    ReplyDelete