Impact of Reflections: In-Room Speaker Recordings

[BenB]

I have posted before about a small nested line-array speaker that I designed and built, which I call the Skylarks. These speakers are typically listened to vertically, where they have wide horizontal dispersion, but limited vertical dispersion. However, they are sufficiently small that they can be turned on their side as well. In the sideways orientation, they produce wide vertical dispersion and limited horizontal dispersion. I have done sighted listening tests where I have rotated the speakers back and forth. This isolates for a lot of variables since it's the same speakers in the same place in the same room, with the same frequency response, the same crossover parts, the same dynamic range and distortion profiles. Everything is the same except the dispersion pattern. I have repeated this test for a very limited number of people.

I have also been working on a microphone array using 5 cardioid mics arranged equally around a circle. It's my intention to couple that design with significant post-processing to allow fairly narrow spatial filtering. As a proof-of-concept, I wanted to see if the mic array could capture the difference in sound of the skylarks in their normal (vertical) orientation as compared to in their sideways (horizontal) orientation. I recorded the same song with the same array in the same spot, played a the same level, and only changed the speaker orientation. Although I'm still working out some issues that are preventing me from finalizing the post-processing, I determined that simply mixing the forward three mics into a stereo recording is fairly demonstrative of the impact.

I have attached 2 diagrams showing the speaker orientation change. I have also linked 2 recordings (flac). I have time aligned them sufficiently so that an abx plug-in can be used to switch back and forth between the two in real-time. I roughly level matched them by reducing the level of the recording from the horizontal orientation by 2dB. The reason for this is that the floor and ceiling are closer than the sidewalls, and thus the vertical reflections add more reinforcement than the horizontal ones. Feel free to do a better level matching if you find it necessary.

If you're interested, take a listen and post your impressions comparing and contrasting the recordings. What I'm most interested in are the implications for how speakers ought to radiate sound into a room. Your listening should be done with headphones since there has already been one set of speakers and a whole room added to the chain.

Skylarks_Normal_3mics_mix_245.flac | Powered by Box

app.box.com

Skylarks_Sideways_3mics_mix_245.flac | Powered by Box

app.box.com

 

[tmuikku]

Could you provide rough sketch about the size of room, the setup (stereo triangle) size and location? thanks!

I think the recording has captured effect of the reflections quite nicely. At least, magnitude of the change seems to be realistic. With the normal upright orientation it sounds wider, quite reasonable, but also seems to have more clarity to it. I suspect the "listening triangle" is quite big, and speculate that making it smaller could bring the two closer together in sound. Still I think the normal orientation would sound better assuming the sound translates through the recording method.

Thanks for posting it, fun experiment!

 

[amirm]

Very well done. I am promoting to home page to get more people to listen.

 

[poxymoron]

Well done, this looks interesting. I'll know later on if it sounds interesting.

 

[maty]

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR12 -5.33 dB -21.20 dB 2:45 ?-Skylarks_Normal_3mics_mix_245
DR12 -4.77 dB -20.33 dB 2:45 ?-Skylarks_Sideways_3mics_mix_245
--------------------------------------------------------------------------------

Number of tracks: 2
Official DR value: DR12

With my modded KEF Q100 coaxials, nearfield. The difference is very obvious. I like the sound of the second audio better, more "realistic".

Maybe is a better idea with other orchestral recording, with at least DR15.

 

[Geert]

My impression is that the first recording has a wider stereo image. The second does not only sound narrower, I also hear the negative effect of something that sounds like comb filtering or phase issues. The first recording also seems to have a more powerful low end, but difficult to conclude with this track and the headphones that I'm using (HD600 but not corrected for the loss in low end). Not my style of music for judging sound quality.

 

[holdingpants01]

I much prefer the normal position, sideways they are way less natural and seem almost like a mono recording

 

[DavidK442]

Thought for sure I would hear a difference in frequency response through my TruthEar Zero IEM’s, but no.
The only difference I heard was a wider sound stage in sample 1.

 

[uwotm8]

First one is more balanced and "right sounding". Second one sounds more chaotic and muddier.
- AKG K361 -

 

[Drengur]

Bose QC 35 (active). First one is as you would expect from a recording, wide soundstage, good separation, all that boring stuff. The second one is fun. More depth to the soundstage and more "realistic" in terms of an orchestral performance. I would prefer the second one for this type of music, at least I would try it out.

 

[STC]

I have also been working on a microphone array using 5 cardioid mics arranged equally around a circle

Is there a particular reason why you are using 5 cardioid mics instead of one Omni or binaural? I am wondering if the ceiling/floor reflection fully captured by your mics.

 

[Thomas_A]

My usual question; shouldn't the sound be captured binaurally and reproduced with binaural-adapted headphones?

 

[AdamG]

If you elevated the speakers from the floor in the horizontal position that would help explain the changes in reduced low end performance of the second recording. Careful with boundary gain. Just my suspicion from listening to the two recordings. Can you elaborate on the physical speaker proximity to boundaries please?

 

[Geert]

That's 2 times I read 'more realistic' for the 2nd recording. Question is why it sound more realistic? That recording almost sounds mono. Someone believes a real large orchestra sounds mono?

 

[Cote Dazur]

That's 2 times I read 'more realistic' for the 2nd recording

Make it a 3rd time with me. the first recording sounds less real than the second. The second let the music trough, feels more like a real venue.

 

[Drengur]

That's 2 times I read 'more realistic' for the 2nd recording. Question is why it sound more realistic? That recording almost sounds mono. Someone believes a real large orchestra sounds mono?

It could be that those who found it more real did not find it to almost sound mono. It could also be that realism is judged by other factors than stereo vs. mono. In my case I am not sure what caused my brain to feel as if I was within the room with the orchestra as opposed to the first recording where I felt I was in the room with the speakers.

 

[Geert]

It could be that those who found it more real did not find it to almost sound mono.

I'm pretty sure a phase correlation graph would show the first recording having a wider stereo image. What could be of importance is the type of headphones. So I had a second listen with my Beyerdynamic DT 770 Pro's, and the difference in stereo width is now even more outspoken.

It could also be that realism is judged by other factors than stereo vs. mono

What else is there besides frequency response, which as confirmed by others doesn't differ that much? Hence why I'm interested to see people describe why it sounds more real to them.

In my case I am not sure what caused my brain to feel as if I was within the room with the orchestra as opposed to the first recording where I felt I was in the room with the speakers.

Than it maybe depends on at what distance to the orchestra you usually sit when you attend a concert. At a larger distance it will sound more like a monolithic wall of sound, but at close distance I'm used to more or less being able to locate the musicians.

It's funny I regularly see people asking for headphones with a more spatial sound (they hate my HD600s), never the other way around, and now we suddenly see people prefering a narrower image.

 

[Thomas_A]

Once I listened to a pair of Megatrends placed quite close together in a small room.

The sound was like listening into a tunnel or mono. There room was not there.I guess the effect is similar here when speakers are vertical. Less vertical reflections.

Either you are trying to "go to the venue" or "letting your room fuse with the venue". The first "mono experience" should have added ambience from other angles (multichannel) to "be there" while the second one should have reflections from your room. Center speaker and side-wall speakers could however simulate the venue better if the correct processing is made. In the end, it will be both personal preference and music/recoding dependent.

 

[Geert]

Later on it would be interesting to hear the original recording, so we could assess to what extend room reflections increased the width of the stereo image, or the horizontal speakers narrowed it down.

 

[fricc]

On my headphones (AirPod Pro II) the first configuration sounds more spacious and more tonally balanced.