Monaural Speaker Recordings - Blind Testing Six Speakers for Fun

[dasdoing]

sorry for so many posts. always new things coming into my mind:
if they sound garbage "because we hear it diferently",
that would mean that a recording of real event in a room would also sound garbage like that, wouldn't it?
No, they sound garbage because small rooms are acoustical garbage, and than you mix that garbage with the allready embuted room in the recording

 

[GXAlan]

When you use a Umik for measurements the reason you use a sweep is so you can window it in time. Basically if the software is only listening in the narrow range of frequencies and rejecting what comes with a delay from reflections it can mostly not be corrupted by those reflections.

Makes sense. People have talked a lot about the UMIK-2's "increased directionality" at high frequencies due to the larger capsule. I have compared head-to-head against my UMIK-1 and there's a slight difference in the midrange actually but the high frequencies aren't that different in practice. I have read that the UMIK-2 is better calibrated though? I bought the UMIK-2 primarily for ultra-near field measurement of speaker drivers -- but not sure if I need to keep both or if the UMIK-2's worse polar plot is theoretical rather than practical.

were those speakers in-room corrected?

5/6 were not room corrected. One was corrected with a 30-band EQ to pink noise.

I will disclose that one of two of the non-corrected speakers which were recorded include
1) The Bose 901 Series VI which in the room measured looks like this* (except I powered it with a solid state amp for the recording).

2) The Onkyo D-TK10 which is in a different room from the Bose 901, and measures like this:

 

[GXAlan]

sorry for so many posts. always new things coming into my mind:
if they sound garbage "because we hear it diferently",
that would mean that a recording of real event in a room would also sound garbage like that, wouldn't it?
No, they sound garbage because small rooms are acoustical garbage, and than you mix that garbage with the allready embuted room in the recording

Well, I think it's because real recordings of a real event in a room are done with different types of microphones.

That is, when I record have recorded events with something like this:

Sony PCM-M10 Review

The result is subjectively great. It sounds way better than anything you get from the UMIK-2

I don't know how accurate this is:

PCM-D1 Digital Linear PCM Recorder - Sony Pro

Discover the Sony PCM-D1 Digital Linear PCM Recorder. Built-in Electret Condenser Microphones. Find out more.

pro.sony

but recordings of events in small/large rooms always sounded great.

But since there are no defined FR calibrations for these, they may be adding some sort of coloration.

Even something like a Rode NT-SF1 has a datasheet like this, but no "average" correction you can apply. It's also 1/3 octave smoothed.

 

[krabapple]

The listening position varied between rooms

Add more variables, why don't you?

 

[GXAlan]

Add more variables, why don't you?

Trust me, it will make sense when all is revealed. What I really wanted to do is to give people a virtual listening experience in mono, but it turns out to be a real challenge!

Hopefully I get 50 voters this time. But anyone who has voted and promises to keep it a secret, I can reveal it to you via PM as long as you don’t change your vote or leave comments that could sway people.

 

[Blumlein 88]

Trust me, it will make sense when all is revealed. What I really wanted to do is to give people a virtual listening experience in mono, but it turns out to be a real challenge!

Hopefully I get 50 voters this time. But anyone who has voted and promises to keep it a secret, I can reveal it to you via PM as long as you don’t change your vote or leave comments that could sway people.

Send me the info please. I'll keep it to myself.

 

[Chrispy]

What are "speaker recordings" and why would they be meaningful?

 

[Blumlein 88]

Here is a post about youtube videos of speakers playing music from me three years ago.

How much can you judge a speaker from a YouTube video?

Someone might have mentioned this before, but it bother me nonetheless: One never knows whether the person(s) making these videos are post-processing the audio, in order to advance an agenda.

www.audiosciencereview.com

 

[GXAlan]

What are "speaker recordings" and why would they be meaningful?

Definitely NOT meaningful, but experiential.

1) Would you prefer a good speaker in a great room or a great speaker in a good room?

2) if you wanted to share what you heard/experienced with your home system, how would you do it? Flowery language? Measurements? What does a narrow Q dip really mean?

My thought was to place a mic at the listening position. Option A is a binaural or ambisonic speaker. Unfortunately I don’t see any calibrated microphone. Option B which took time and effort was to do a monaural recording at the main listening position with a UMIK-2.

Turns out that the UMIK-2 isn’t a great solution but once there is a reasonable sample size, I can share the identities and we can see what the 0 votes represented and what the popular setups represented, if there is in fact a preference that is identified.

3) Dr. Toole’s original paper relied on 28 participants only (page 14). We can do real science over time given our numbers!
https://www.pearl-hifi.com/06_Lit_Archive/15_Mfrs_Publications/Harman_Int'l/AES-Other_Publications/LS_Measurements_Listener_Prefs.pdf

 

[Blumlein 88]

There is nothing really wrong with the Umik-2 for recording. It would help if you set recording levels higher to get away from the low level noise.

Here is an alternate idea for your purpose. Using a moving microphone measurement with REW and pink noise, create an EQ curve for each speaker at the listening position. Then apply that EQ to the original source rather than record speakers. See if that result sounds somewhat different upon listening in ways similar to how the different speakers in different rooms sounded. I believe it would be a better, cleaner way to find out if the sound of a great speaker in an average room or average speaker in a great room is preferred. You can even do this for a stereo pair of speakers by doing one channel at a time, and EQing the result of each channel's measurements.

 

[holdingpants01]

did you use one speaker or stereo for the recording?

 

[GXAlan]

did you use one speaker or stereo for the recording?

Stereo.

 

[holdingpants01]

Stereo.

Right, so there are cancellations from the room and mismatched speakers when folding to mono, that would be another reason it doesn't sound anything close to reality. There's a good reason the measurements are done with one speaker at a time

 

[Sokel]

Solely by the first files there are huge differences,specially between A and D and the rest.
Specially D is on it's own level,maybe tonality is not ideal (violin is better in others) but it peaks much-much higher,very different dynamic behavior.

Voted for C even if not so dynamic,sounded more natural (that's not me usualy! )

 

[Hayabusa]

my order of preference:

Spoiler: .............

F E D B A C

 

[GXAlan]

Right, so there are cancellations from the room and mismatched speakers when folding to mono, that would be another reason it doesn't sound anything close to reality. There's a good reason the measurements are done with one speaker at a time

Agree. It seems like a binaural recording would be ideal, but I don’t know if a low cost binaural calibrated microphone that is suitable for this. Maybe the MiniDSP Ears?

Which did you like the most ?

 

[holdingpants01]

Agree. It seems like a binaural recording would be ideal, but I don’t know if a low cost binaural calibrated microphone that is suitable for this. Maybe the MiniDSP Ears?

Which did you like the most ?

Honestly I wouldn't worry about calibration at all, for comparison purposes it's enough if you just use the same measurement device and method for every recording, I don't think you can really calibrate anything other than omnidirectional microphone, the moment you put it inside ears or any baffle it just goes away. Much more useful would be to have some sort of standard to compare, like a reference and baseline that most people know and like, like for example KEF LS50
As for my preference I posted it blurred out on the first page, can't wait for the results

 

[JakeK]

What's interesting is how much all of those setups degraded the sound compared to the original. Some making the instruments sound harsh and unnatural compared to the original recording. A lot of the tonality may be due to the room acoustics but what's making the violin sound so harsh in a couple of them? At least some of the speakers actually managed to keep the sound pleasant for me.

Edit: On reflection the harshness I heard is most likely an amp driven into clipping and not a problem with a speaker. How much care did you take to ensure none of the equipment was driven into clipping? It's not a fair test of the speakers if the amp is over driven and providing a distorted signal.

 

[GXAlan]

Edit: On reflection the harshness I heard is most likely an amp driven into clipping and not a problem with a speaker. How much care did you take to ensure none of the equipment was driven into clipping? It's not a fair test of the speakers if the amp is over driven and providing a distorted signal.

No way it’s the amps, but it could be the mic. The UMIK-2 didn’t clip electrically, and MiniDSP claims UMIK-2 handles “125 dB” but maybe for something that isn’t a frequency sweep, it starts to distort.

Depending on the setup,
1) Bridged NC500, 625W at 1% into 8 ohms
2) Active speaker with 250W class D x 2 (biamplified), no THD rating
3) Class AB amp, 177W and 250W into 8 and 4 ohms at 0.1%, CEA burst of 225 and 396W (1%)
4) Lifestyle product, at risk for clipping (but unlikely once I tell you). Partially measured since it is a Class D with switching voltage Class G/H style. 65 W into main drivers at 0.30% at 4 ohms, and 130W at 0.08% into 2 ohms. Woofers are 2 ohms and power is advertised as 160W and peak 320W there.

 

[goat76]

No way it’s the amps, but it could be the mic. The UMIK-2 didn’t clip electrically, and MiniDSP claims UMIK-2 handles “125 dB” but maybe for something that isn’t a frequency sweep, it starts to distort.

Depending on the setup,
1) Bridged NC500, 625W at 1% into 8 ohms
2) Active speaker with 250W class D x 2 (biamplified), no THD rating
3) Class AB amp, 177W and 250W into 8 and 4 ohms at 0.1%, CEA burst of 225 and 396W (1%)
4) Lifestyle product, at risk for clipping (but unlikely once I tell you). Partially measured since it is a Class D with switching voltage Class G/H style. 65 W into main drivers at 0.30% at 4 ohms, and 130W at 0.08% into 2 ohms. Woofers are 2 ohms and power is advertised as 160W and peak 320W there.

I’m not sure, but maybe JakeK meant that the recorded signal may have been driven above digital zero and caused clipping, not that the amp used for the speakers or the microphone was clipping.

To make sure no clipping occurs while recording you need to leave enough headroom, the peaks should never reach higher than say -3/-6 dB. You must turn the gain knob down on the audio interface until you reach that level of headroom.