Stereo piano recording techniques

I thought I'd put these up as a kind of quiz.  Three samples of recordings, all made in the same hall, on the same piano, played by the same pianist (my son, Nicolas Hodges); they sound quite different, but have all had good reviews.

The recordings all use different microphones (AKG C414s, AKG C426, Neumann KM140), and in different configurations (MS, ORTF, Blumlein).  The acoustics of the hall were reengineered between these recordings as well.  You can try to match the microphones and techniques with the samples:

BH-sample  This is from a serial composition by the unknown British composer Bill Hopkins, who died young in the 1970s.  Nic gave the first performances, and this is the only recording.

MF-sample  Michael Finnissy is a contemporary British composer.  This is from a CD entirely of arrangements of Gershwin songs.

JC-sample  Justin Connolly has retired from teaching composition in London.  This is from a Sonata written for Nic to perform.

The answers are here.

Paul

I guessed right on one, and it was the one I like best.  The MF sample, IMHO, stands out as best sounding.  Your son is quite the pianist!

What's up with the quiet sections of the JC sample?

Interesting, but only one of these techniques produces a consistent identifiable sound, regardless of how it is set up, and another of the techniques can sound like that identifiable sound if you set it up right!

Ok, maybe I'm only saying that because I got them all wrong!  But I know why I did.

What's up with the quiet sections of the JC sample?

Well, well!  What a gross artefact of the MP3 encoding, which I didn't bother to check...  That sample was encoded using the Fraunhofer encoder at 192kbps, and I have replaced it with one made using the Lame encoder also at 192bps.

I suspect that the problem arises because of the non-coincident mics giving a signal having less channel correlation for the encoder to take advantage of.

Paul

PS:  for people who want to hear what happened, here is the distorted one.  A good example of why MP3 can't take over the world.

Interesting test - especially when the acoustic goalposts have been moved!

How did I score? Well, I got the MF and the JC the wrong way around, but correctly identified the BH. which I preferred. Maybe it's me, but there was something slightly stranger about the imaging with the MF one than there was with the other two - but that could simply be a mic placement issue. And to be quite frank, you could only really detect this in a direct comparison - they are obviously all good recordings!

Maybe it's me, but there was something slightly stranger about the imaging with the MF one

Because I had decided in an arbitrary fit of idealogical purism to use Blumlein that time, but the hall was so much more reverberant than before (and the drapes that are now commonly used there weren't yet available), I ended up going scarily close in (for a distant technique).  This produced an unnaturally wide image which some people find strange; another strange effect could be the combination of somewhat close direct sound with lots of natural reverberation.

Paul

What's up with the quiet sections of the JC sample?

Well, well!  What a gross artefact of the MP3 encoding

I suspected MP3 was to blame, although I've never heard it do that before.  I'll have to listen again (later).

Here's my technique at work - no prizes offered for identification...  from its date I can see that no postprocessing was used.

The JC sample sounds MUCH better now.

Personally, I like a wide sound for a solo instrument - in fact what I noticed in the 3 samples is how narrow the image is on the BH sample.

My latest piano recording is here (as long as we're showing off   )
http://www.cheap-tracks.com/mp3/cheap-tracks_sig_myer_chopin_sample.mp3
A Steinway in a school auditorium.  A pair of Oktava 319s over the strings (the mic on the high strings was quite far forward and aimed back to try and compensate for some voicing problems with that instrument).  There was also a pair of C3s setup MS about 8 ft from the piano toward the audience.

This mix is NOT untouched - I don't recall the reverb I used, but the auditorium was pretty dry.

I noticed in the 3 samples is how narrow the image is on the BH sample.

The miking was much more distant because of the drier acoustic in the hall at that time.  I then had the flutter echo problem that I alluded to, and found myself moving the piano and mic around together to keep the balance but avoid the flutter.

I find I can't really choose between the samples.  The BH is perhaps the most analytical, as befits the complexity of the music, but I think I most enjoy the greater warmth of the ample acoustic in the MF - I tend to go for a wetter balance than many people.  The JC sounds most natural on headphones, as the spaced mics of the ORTF technique have some affinity with a dummy head technique.

Incidentally, the last piece on the Gershwin arrangements disk ends with a chord that is held (in the take used) for 75 seconds.  Although this was only a 16-bit recording, the end of that chord is still clearly audible in the hiss and dither.  To get the CD under 80 minutes, it was edited to 40 seconds (and suitable damper noise was added from elsewhere in the session).

Paul

Dramatic, certainly Karl!  But for me as a matter of personal taste it's the wrong way round - (at this point Oz takes off his headphones & checks...) - the perspective is as from the pianist's point of view rather than the audience (that is, the bass strings are recorded to the left whereas an audience would hear them predominate from the right).  The danger with close recordings is that as you seem to have found,  it is demanding on the quality of the instrument, not to mention the player.  I recall an organist being unhappy about my close recording of a church organ - "you wouldn't look down the front of an old lady's dress, would you?" was his way of putting it.

The MF was nearest to my taste, Paul.  Again, I'd emphasise "my" and "taste" - there's probably more variation in piano recording and less agreement on what is preferred than in respect of most other instruments.  A good one to discuss!

Dramatic, certainly Karl!  But for me as a matter of personal taste it's the wrong way round - (at this point Oz takes off his headphones & checks...) - the perspective is as from the pianist's point of view rather than the audience (that is, the bass strings are recorded to the left whereas an audience would hear them predominate from the right)

I seem to recall some trouble when mixing this recording.  First, I close mic'd because there were other pieces that included drums and electric bass, as well as vocalists.  The MS mics were setup originally to pick up a vocal quartet, but I found they really enhanced the sound of the piano on the solo pieces.

I don't recall the details, but I seem to remember being mystified by the blend of the close and far mics sounding best with the short strings panned right.  Probably due to the fact that the MS mics were not placed to record the piano. (I think my naiveté is showing   )

[edit]

Dramatic, certainly Karl!

I can't help but feel like this is said with the same tone of voice one would use with a 5-year old proudly displaying a large finger painting done all with fluorescent colors.  I take it you feel it's a bit over done?[/edit]

The danger with close recordings is that as you seem to have found,  it is demanding on the quality of the instrument, not to mention the player.  I recall an organist being unhappy about my close recording of a church organ - "you wouldn't look down the front of an old lady's dress, would you?" was his way of putting it.

There are other good reasons for not doing that with an organ as well - but that would be a serious digression - and I think that I've been through some of the reasoning behind this before.

But this whole 'which way around is the paino?' thing, and to a degree the 'instrument width' consideration, which applies to all acoustic recordings is quite subjective, and even to an extent variable with the program. Ultimately it comes down to what exactly it is that you are trying to recreate. If you go to a piano concert and sit on the LHS of the hall, you would in theory hear the bass notes slightly to the left of the higher registers, and if you sit on the other side, the opposite is true. If you sit in the centre, the impression will be rather more determined by the venue's acoustics, and you won't be hearing a very wide instrument anyway, unless you are really close to it.

My personal take on pianos is that if you are going to go for the 'wide' sound, then this would be the one that the player heard - so the bass goes on the left! But primarily it is the venue that determines what you hear and percieve to be going on - try recording a piano out of doors, and you'll realise just how important the room is to the way the sound develops. And it's worth concentrating on this to a significant degree when you make an acoustic recording. That's clearly what Paul has done, but he was fortunate in that the venue is good enough to allow him to do this without too much compromise - other than the ones that have already been noted.

So I'd say that it's all down to soundscape planning. The purer the mic technique, the more care you have to exercise in its execution, I'd say. I don't record pianos very often - in fact the last time I did a serious recording of a piano was way back in the last millenium - but I recognise all of the compromises! What I have found, though, is a pile of organ recordings that I made privately a long time ago (1970's), and when I've restored a bit of one of them, I'll post it (warts and all), because it does illustrate my take on the width/venue issue, FWIW.

Heh, I see what you mean about the use of the word "dramatic" - I guess I was trying to convey my initial reaction which was "wow!" rather than "umm" and at the same time I was feeling that you'd "got away with something", insofar as it sounds very close (for reasons you've indicated - indeed, the recording of live events has to be full of setting one factor against another) but it "works".  But like I say, one hears every variation on piano sound - both you (Karl) and I would doubtless take issue with some I've heard where the instrument sounds like it's been recorded on the next planet.  I like a close, wide sound (a piano is no small thing) set in a somewhat separated halo of supportive reverb - if forced to put it in words.

As for the "which way round" thing - I guess I tend not to sit on the keyboard side of the hall at a recital, more the other side where you don't see the pianist's hands but you do hear the fully developed sound of the instrument, which normally in my estimation happens at an angle of 45 degrees towards the tail.  I tend to place a mic stand more or less opposite the rear foot, but then about nine inches towards the keyboard from there, with the stereo pair pointing towards a point just to the right of the lid hinge nearest the keyboard, six to eight feet from the piano - I'm looking for a point where the sound integrates but the focus remains clear.  Whether that garbled description conveys anything I'm not sure!