Grand microphone test for guitar recording

[Malcolm Kavalsky]

Malcolm, did you listen to the second set (omnis)?

Sorry, missed them, my take is:

A2 - Lots of room reflections, unbalanced sound
B2 - As before but with a wider spectrum of frequencies
C2 - Similar to A2, more bassy, loss of higher frequencies
D2 - Sounds like you added reverb to A2, still too much room interference

In general they all seem to suffer from the room ambience, which produces a much less focused sound. Not my cup of tea ...

[Denian Arcoleo]

I think, up til now Delcamp Forum admits only 128kBs MP3 or less, so we wouldn't gain anything here.

Delcamp allows mp3's of up to 16mb in size. You can encode your mp3 at any quality you wish, up to 320kbs - the only limiting factor is the size of the file.

[rojarosguitar]

Hello Denian, thanks for cerrection. That doesn't make the things much better, soundwise ... it's just far away from the full resolution of the CD quality. Mostly it's the fine aspects that suffer from that, especially the ambience and sound coloration ... (remember even with 320 kB we have a data loss of 75%) Maybe it doesn't make the mp3 sound 75% worse, but there is a significant loss of information, and one has to be half deaf not to hear that, at least on any halfways decent reproduction system.

That is what makes it a bit frustrating to share this kind of test - because the fine points really get lost a bit. BTW This is the tragedy of many great sounding studios: they invest hunderds of thousands if not millions of $$ in their superb equipment, and in the end ... people squash it into their iPods and hear compressed music, because it's quantity and not quality that counts.

But anyway, one can still perceive differences; my point here is to show you how different mics can sound, even if they all can be called good or even very good mics. Which one one would choose for recording is a matter of taste.

best wishes
Robert

[rojarosguitar]

...

In general they all seem to suffer from the room ambience, which produces a much less focused sound. Not my cup of tea ...

Malcolm, here you miss the point a bit: it's not microphones that suffer from room ambience, it's the recording technique (if you like to call it 'suffering', because it's a matter of taste.). If I wanted less ambience I'd have to position the mics closer to the guitar. The point I'm showing you with the omnis is just the opposite: even in a distance where you would have a significant loss of low end frequencies on cardioids the omnis are still having it. Of corse one could bring the omnis closer to the guitar for a more direct sound (actually the should be closer than cardioids, but then I'd have to make the base width also smaller, which I didn't wanted, for other reasons). So please don't forget that in this test I do not compare different recording techniques but different mics in the same configuration.

The main point is to watch out for the differences in the sound texture under same circumstances!

Best wishes
Robert

[Denian Arcoleo]

That is what makes it a bit frustrating to share this kind of test - because the fine points really get lost a bit.

That's really the point. Determining the finer points of high fidelity audio using mp3's or youtube is pointless. The only way to conduct this kind of experiment is by using at least CD quality audio, and don't forget of course that audio quality can be much much higher than CD standard. But again the problem is file size; a one minute sample of audio at truly high quality, say 96 kHz/24 bit, would be around 100mb.

[Malcolm Kavalsky]

...
In general they all seem to suffer from the room ambience, which produces a much less focused sound. Not my cup of tea ...

Malcolm, here you miss the point a bit: it's not microphones that suffer from room ambience, it's the recording technique (if you like to call it 'suffering', because it's a matter of taste.).

I meant that the recordings suffered from room ambience, obviously the microphones never suffer

If I wanted less ambience I'd have to position the mics closer to the guitar. The point I'm showing you with the omnis is just the opposite: even in a distance where you would have a significant loss of low end frequencies on cardioids the omnis are still having it. Of corse one could bring the omnis closer to the guitar for a more direct sound (actually the should be closer than cardioids, but then I'd have to make the base width also smaller, which I didn't wanted, for other reasons). So please don't forget that in this test I do not compare different recording techniques but different mics in the same configuration.

The main point is to watch out for the differences in the sound texture under same circumstances!

Best wishes
Robert

So, which microphones were used for each recording ?

[Jouni Stenroos]

One question: Since you've recorded already with 16bit/44.1kHz, why do you apply dithering? It should be used only when you record with higher bit depths, before converting to the lower depth.

Here is IMO a very good explanation of dithering:
http://www.users.qwest.net/~volt42/cadenzarecording/DitherExplained.pdf

Regarding SoundCloud, they do store the original file you upload, and you can choose to allow downloading it. However, the free account has limits on number of downloads.

I asked for SoundCloud support here in Delcamp a while ago. It's a big disappointment that instead of having better support for it (an option to embed the SoundKloud audio player), it was banned altogether...

-Jouni

The microphones were on top of each other, separated by 36cm basis width and a good arm length away from the guitar. They went through identical channels of a very high quality preamp and were recorded in 16 bit/44.1kHz format onto a Zoom R24. This made it possible to record four pairs at the same time, so you'll here exactly the same take through the different pairs.

The wav files were imported into Nuendo, the snippets cut out, normalized and dithered with the Apogee UV22Hr algorithm. Other than that there are no effects whatsoever (and I mean it, absolutely nothing).

A called the pairs of this part of the test A1, B1, C1 and D1, so that you can listen to it and be completely free from considerations of brand, price or what else.

I'd love to hear many opinions first, before I'll divulge the secret which is which ...

best wishes
Robert ... still waiting for the first upload to be completed ...

BTW the guitar used here is a Sebastian Stenzel CD/EIR, built 2000. Strings are: D'Addario Pro Arte NT for bass, SoftSteel fluorocarbon 0.9mm for g, Augustine Imperial for b and e.

PS I'm not so sure whether sound cloud preserves the quality of the soundfiles, because I see a message 'transcoding completed', so maybe the use also some kind of compression?

[rojarosguitar]

Hello Jouni, thanks for the nice source to understand dithering.

Here's the reason why I always apply dithering as final step in mastering:

the 44.1kHz/16bit file is imported into Nuendo as 44.1/24bit session. Then different processes are applied, some of which involve 32bits calculations, some floating point calculations, that are saved as 24 bits data. To get down to CD Audio format again you have to reduce it back to 16 bit (even if there are no other processes like normalizing or other plug ins involved). For this dithering is IMHO a must. I use an excellent dithering algorithm by Apogee called UV22HR for this last step.

Please dont hesitate to present your arguments if you think I made an error in my assumptions or reasoning.

best wishes
Robert

[Jouni Stenroos]

Robert,

Thanks for the clarification. I based my comments on your original statement where you said you recorded at 44.1/16 and did no processing.

I'm not sure if the dithering is necessary if the audio temporary goes through higher bit depths, but I might as well be wrong.
Also, I'm not sure how the dither affects the following conversions to lossy formats like mp3. Bouncing the tracks in Logic Studio
allows you to dither only if you bounce to wav of aiff, with mp3 or aac it's not possible. I've read somewhere that those encoders
have their own dithering systems. If this is true, then it would be optimal to make the mp3 (or aac) directly from the DAW, not
exporting a WAV first (with dithering) and then converting that. But again, this goes to the guessing category

-Jouni

Hello Jouni, thanks for the nice source to understand dithering.

Here's the reason why I always apply dithering as final step in mastering:

the 44.1kHz/16bit file is imported into Nuendo as 44.1/24bit session. Then different processes are applied, some of which involve 32bits calculations, some floating point calculations, that are saved as 24 bits data. To get down to CD Audio format again you have to reduce it back to 16 bit (even if there are no other processes like normalizing or other plug ins involved). For this dithering is IMHO a must. I use an excellent dithering algorithm by Apogee called UV22HR for this last step.

Please dont hesitate to present your arguments if you think I made an error in my assumptions or reasoning.

best wishes
Robert

[rojarosguitar]

Hello Jouni,

That was a part of my original post:

...
The wav files were imported into Nuendo, the snippets cut out, normalized and dithered with the Apogee UV22Hr algorithm. Other than that there are no effects whatsoever (and I mean it, absolutely nothing).

...

Maybe there is a slight difference with mp3 made from dithered and undithered files, but I'm quite sure that it's much smaller than the loss that MP3 does to audio files anyway...

anyway, thanks for sharing your knowledge, after all we're busy enough with music and most of us are not computer scientists or digital audio engineers ...

best wishes
Robert

[Per Lindhof]

Hi Robert.
Great work. And very nice playing

In the cardiod test (1) my bet is:

A1 nice and linear sound - quality mic

B1 a little thin but okay - medium quality mic

C1 This is a LDC and I don't like it - too boomy

D1 Cheap mic or a mic for some other purpose


The omnis (2)

A2 nice sound a little spacious

B2 Nice too

C2 I guess this is a cardiod it's less spacious

D2 A little neutral maybe a little thin hard to judge


The ribbons

Well. Hard to judge but no 3 is a different mic....
Still the positioning can be crucial. I've succeded in making a cheap Octava sound better than a Neumann by changing position (but the Neumann in the right position was better overall).

I'm now very curious to see if my favourites A1 and A2 are cheap chinese condencers and my "loosers" are the same Neumanns as I spend last months salary paying out

Kind regards

Per Lindhof

[fatwarry]

I think, up til now Delcamp Forum admits only 128kBs MP3 or less, so we wouldn't gain anything here.

Delcamp allows mp3's of up to 16mb in size. You can encode your mp3 at any quality you wish, up to 320kbs - the only limiting factor is the size of the file.

Denian, This thread viewtopic.php?f=16&t=51113 says that samples can be uploaded to this subforum and must be in wav format at 44.1khz but limits the length to 15 seconds. Why the inconsistency with the MP3 "rules" . If we could upload 16mb files to this subforum rojarosguitar would be able to upload his unadulterated source files. Actually if I recall it was something to do with copyright. But if the piece is out of copyright or, as in this case, the copyright holder has given permission, that shouldn't be a problem.

Oh, and being a bit of a pedant, have any of you claiming that compressed audio is easily distinguished from uncompressed ever done any double blind tests to check? I think you may be surprised (although MP3 is generally not felt to be as good as more recent codecs)

[sphinx]

Thanks for sharing your tests. In my opinion they confirm that in acoustic recording the room is key.

In a perfect room I'd take the ribbons (preferably A3) or LCDs (presumably C1), as they deliver a fuller and to my ear more natural guitar sound, especially on the nylon strings. Unfortunately, they hear and expose too much of the room artifacts.

In a noisy room (like the one in the test, and just about any untreated living or practice room) I'll have to chose the tighter cardoids, despite their edginess, to eliminate as much of the room artifacts as possible. In that case I'd take A1.

For me the omnis are impossible to judge in a noisy room, but in my experience they tend to sound more natural than cardoids in stellar acoustics.

[Malcolm Kavalsky]

Denian, This thread viewtopic.php?f=16&t=51113 says that samples can be uploaded to this subforum and must be in wav format at 44.1khz but limits the length to 15 seconds. Why the inconsistency with the MP3 "rules" . If we could upload 16mb files to this subforum rojarosguitar would be able to upload his unadulterated source files. Actually if I recall it was something to do with copyright. But if the piece is out of copyright or, as in this case, the copyright holder has given permission, that shouldn't be a problem.

Oh, and being a bit of a pedant, have any of you claiming that compressed audio is easily distinguished from uncompressed ever done any double blind tests to check? I think you may be surprised (although MP3 is generally not felt to be as good as more recent codecs)

Actually, I think for purposes of comparison 15 seconds is probably enough. If it is too long, I tend to forget the previous clip, while listening to the new one. Though if you want to highlight in one clip different sounds, eg bass vs treble vs chords, it may be cutting it a bit fine. Still, it does seem to be the best solution so far.

Thank you Moderators!

[need2Practice]

Robert,

I think Malcolm and I either have very different systems (I'm listening through good headphones) or very different tastes, as my preferred samples are generally the reverse of his. My favorites are:
C1, D1, B1, A1
C2, D2, A2, B2
C3, B3, A3

My overall favorite is the omni C2 as I find it to have a nice balance of reflected and direct sound, and the low end isn't missing like some of my least favorite samples (A1 and B1). I'm very surprised that others like A1 as that one in particular seems to me to have a hard, metallic edge to it. I listened to the samples before reading that they were compressed by sound cloud, but now that I know this it makes sense that I favor the more bass-rich samples as I think I hear compression most in the mid-high range.

Regardless of the compression issue, I was surprised how much I preferred all of the omnis over the cardoids. The omnis seem to present a much smoother, more natural sounding spectrum.

I wanted to like the ribbons better but I found them a bit harsh, especially A3. I'd love to listen to the uncompressed originals using ABX software.