Lowden O23 recorded with Sony PCM-D50, honky sound?
 

I'm experimenting with recording fingerstyle guitar in my living room, using the Sony PCM-D50. The goal is to find the best possible situation, considering that I cannot "do" much to the living room (except possibly heavier curtains if wife allows it). Later, I plan to get some better mics and a Sound Devices USB Pre 2. But first, please help troubleshoot this recording.

To me, the guitar sounds "honky" (or maybe boxy... I'm not good at describing sound). I do hear some of that character when I play it, but on the recording it dominates the sound. I would like to get a bit more "open" tone.

The question is, where is the problem? I can think of a few things:

Mic placement (but where to move it).
Room resonance.
Guitar sounds like this up front but I hear it differently since I'm playing.
Eq (how).
Recorder sucks (not that likely, I think it can do better).
The guitar sounds fine, it's all in my head...(?)

Here is a high bitrate mp3: http://www.mateuszherczka.net/guitar/DonalOgg.mp3

Here is the original (only normalized) 24bit, 44100Hz wav file (10 megs): http://www.mateuszherczka.net/guitar...Norm121217.wav

Recorded at 24bit 44100Hz. I can do 96000Hz but that probably won't do much. I think I recorded a touch low (peaks around -12db), next time will shoot for -6db.

I'm sitting in a sofa about 1.5m from the wall on my left. There is a window in the wall with thin curtains. I'm at a sharp diagonal with the back agains the corner, "shooting" along the length of the room (it's a fairly long but not so wide living room). Wooden floor but a lush carpet underneath me. Recorder about 40cm in front of the guitar, mics in X position so one points towards head, the other towards the fat end of the guitar.

My playing, well, I clearly ain't Martin Simpson, still if you hear something I shouldn't do, let me know.

Strings are elixir PB nanos. I normally don't like them, but on this guitar they found a voice. But maybe these are the problem?

Any ideas welcome!

I don't hear boxy or honky, so I'm saying it's either all in your head _or_ in your playback chain/room interaction. The latter is probably more likely. This is the reason that most of our recording budget should _really_ go for monitoring if we're interested in results.

Recording at 96k - if it changes the tonal balance your system is broken.

The PCM-D50 definitely doesn't suck.

This part right here:

I don't quite understand. Digital recording is referenced to 0 dBFS - this is all the bits in your selected word length set to 1 - and no higher level is possible at the a/d. This is why tracking levels are expressed as negative numbers, below 0 dBFS.

Are you judging your levels on the recorder or in a floating point DAW program? It _is_ possible to go "over" 0 dBFS inside a floating point audio program but the signal must be reduced to fixed point levels before the d/a or the output will be clipped.

Fran

Fran, I meant of course - not + (edited post to fix).

The guitar sounds honky (or middy, which in a way makes it sound thin) to me in relation to, say, a Martin Simpson or Pierre Bensusan record (played on the same AKG240 headphones). I guess they have better mics and recording space.

Sounds very good to me.

I think I hear the "honk". It's not a bad sound but I thought I was listening to a small-bodied guitar until I checked up on the model... jumbo!

If you're in close, different parts of the guitar give off slightly different sounds. Try some experiments with different mic positions.

Maybe try recording outside in an open space, if that's possible. That will give you something fairly reflection-free to compare to the room.

Curtains won't help much for sound-absorption. There is no substitute for a couple of inches of high-density mineral wool.

Thanks, that pretty much sums up the problem!

Hm. Do you think one of those homemade reflexion panels behind the mic could do something for the sound?

Like these ones in Fran's blog post? Yes, they can make a significant difference and they don't have to be a permanent feature.

In general, diffuse (ie scattered) reflections from far away are good and anything focussed and/or close is bad. Large flat surfaces in small spaces will tend to need quite heavy treatment. A large room with "broken-up" surfaces will scatter sound waves around and give them room to breathe.

I think that's a very nice sound, sounds like a Lowden :-) I really like the piece and playing, too. There are some things that might help, tho. I don't really hear room sound, tho it never hurts to try to improve the acoustics. I do hear pretty substantial noise - listen to the tail - probably a computer fan?. Even if it's not clearly heard, that casts a shadow over things and can have a negative effect. Once I noticed it on the end, it's clearly audible throughout.

The stereo image is pretty narrow - that's mostly the effect of the X/Y mics on the recorder. Not a bad thing, but I often find narrow mic patterns accentuate a "closed" sound, which may be what you're hearing as honk. Wider mics tends to open the sound up and make it more airy. You're also panned substantially to the right, so getting the sound centered will help make a nicer stereo image.

I did a quick pass on this, centering the image, then widening it a bit, reduced the noise centered at about 100Hz a bit, then EQ'd it to boost the low end (probably undoing the noise reduction, but bringing up the low end of the guitar some), and added some reverb. See what you think of this:

http://www.dougyoungguitar.com/mp3/DonalOggNorm_mix.mp3

(most of the stuff I did - centering and widening the image, reducing noise, etc, would be more effective if done on at record time)

There's still a sort of unique "sound" to the guitar, but I think it fits to the piece. I could easily hear Martin Simpson or Tony McManus getting this sound. Beautiful playing!

Darn! That mix is just amazing, what exactly did you do? Mind if you post the settings? That's the difference between knowhow and nohow...

The noise is just "house noise", partly from the outside but also the chimney... Possibly also the fluorescent lamp above me. Everything else was turned off to minimize noise. The next step will be to make some Fran-panels which might isolate things a bit.

Good tip. The mics are obviously fixed on the recorder, but this makes the case for thinking about a dual mic setup in the future.

Thanks a lot for the kind words! But it's your mix that really brought that out. I have clearly a lot to learn.

In the meantime I experimented a bit with mic positioning and found that the sound changed drastically when the recorder was elevated about 50cm. I tried different positions relative the guitar, but the biggest change seems to come from the additional distance to the floor. Possibly some of the "honk" was created by reflections from the floor?

I changed strings to John Pearse PB and did a new recording (in retrospect it might have been better to keep the old strings on for comparison). The new strings make the guitar more woody (and add some squeak), but the mic position is the main difference in sound. It's much closer to what I hear when I play. The stereo image is a bit off again, I'm just posting the raw recording. The offset comes from the xy mic position, one mic picks mostly up from the neck, the other from the body.

Here is the mp3: http://mateuszherczka.net/guitar/DonalOgg2.mp3

Here is the 24bit wav: http://mateuszherczka.net/guitar/Don...norm121218.wav

After hearing Dougs mix I realize that both sounds are useful, depending on where you want to go with it.

Which do you think is "better" in terms of (1) raw recorded sound for later processing (2) pleasing to the listener? Are these criteria always the same?

I'm beginning to wonder if it might be the Sony's mics. Seems to be recoverable though with EQ. Sometimes the information just isn't there but Doug's tweaks really bring it to life - nicely judged reverb too.

Wait a minute... I just checked and there's a low-cut switch on the unit. $10 says that's the problem.

I didn't do a lot, but my EQ has a 9.5 db boost at 88Hz to bring up the bass (that's a *lot*, so moon's guess about a low cut switch may be correct). I also cut 2 db at 350, and 2 db at 2Khz - those were debatable, but its subtle, and I was trying tame the low mids with the 350 and a bit of the agressiveness at 2K.

I also used Ozone 4, which has an "exciter" tool, and I added a bit of that on the low end 20-300Hz, which adds some warmth. A little gos a long way on this, I just had it set on 1%.

I widened the sound with the Voxengo free MS tool, mids at -02 and sides at +1.2.

Reverb's hard to describe, but this was a "Guitar Room" preset on a Bricasti M7.

The noise reduction, I did with Izoptope Rx. There was a clear low hum around 100Hz, and I was able to isolate fairly well that on the tail. Izotope has a "learn" mode, where you give it a sample of noise, and then it nukes it everywhere. That could have worked better if I had a clean sample of nothing but your room sound without the guitar in there.

Having separate mics provides more options, but even with the fixed mics, there should be a lot of sounds to find, by moving things around, as it sounds like you've discovered.

Over all it sounds good to me – clean, detailed.


Part of that midrange sound and smaller guitar sound is due to the tune and the fifth fret capo.


A little reverb is all I would do. The right one will warm up the sound a bit and widen the soundstage.

How about swinging the PCM-D50 mics outward instead of using them in XY?

Fran

Aha - capo! Yeah that's probably what I heard.