Impact of Reflections: In-Room Speaker Recordings

[goat76]

This also raises the question (again) whether "near-field" speakers should be designed to have wide dispersion to add more reflections (while far-field should have more narrow). Or with other words, should the DR ratio be kept constant with listening distance?

I think I prefer speakers with fairly wide dispersion in combination with listening with a small listening triangle, with that I think I get a high ratio of direct sound and also later arriving reflections from the listening room that adds to the envelopment. That's what I think I hear in my setup and I like that a lot.

 

[goat76]

It's a big orchestra, so the concept of phantom image is a bit vague in this context. You can expect to have different instruments groups being spread from left to right. I clearly hear the flutes somewhere in the middle. In comparison, what's happening at the outer edges is indeed to strong in this recording, but that's also the case also in the original track. For me it was clear this was an issue with the original recording, and the recording with the speakers in vertical position was least detrimental, certainly when you take the frequency spectrum issues into account. Like I said at the start, not I track I would use to judge sound quality.

While it's hard to judge if the phantom image is optimized with classical recordings that don't have a strong center image as a multi-mono recording, the phantom image is still as important to get right for classical music, it's just less obvious to notice.

In the original track, I hear the flute much more distinct as a sound source in the middle of the stage, but in the first recording, it's pretty far to the right with a more hazy kind of sound that I think has to do with too long distance between the loudspeakers. The unbalance of the stereo image is of course caused by the microphone's positions or mismatched levels between them.

So unfortunately this test missed its objective a bit because it became a discussion about recording quality instead of an assessment of what speaker placement does to sound.

I really appreciate listening tests like this as they are fun, but it's almost impossible for any other of us to know if the difference we hear in the recordings is due to the recordings or the positioning of the speakers.

 

[Geert]

I really appreciate listening tests like this as they are fun, but it's almost impossible for any other of us to know if the difference we hear in the recordings is due to the recordings or the positioning of the speakers.

It was clear to me. Clear comb filtering, low mid bump and far to narrow stereo image in the second recording. Things not present in the first recording, and a sound that's very far off from the sound of other recordings. The second recording just sounds deteriorated. With some basic EQ-ing and panning you can make the first recording sound like the second, but it would be very hard to restore the second one to make it sound like the first.

For me the answer to which speaker position sounds best is very clear.

 

[krabapple]

first one sounds like stereo with perhaps slight bias towards right channel.

second one sounds almost mono, everything clustered in the center

CD rip has better imaging and clarity than either

phones are a Logitech sealed wired headset

 

[BenB]

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

I searched my phone and camera to see if I snapped a picture of the setup, but I couldn't find anything. The recording was made in my basement. The geometry was standard: distance between speakers = distance from speaker mid-point to listener. In this case that distance was only about 7.5 feet. When I rotated the speaker from vertical to horizontal, I supported them in such a way that the middle of the tweeter remained in exactly the same location (both in height and separation). The listening triangle was pushed into the front, right corner of the room, with about 2.5 feet behind the speakers and 4 feet to the right wall, but 6 or 7 feet to the left wall. There are also some things on the right wall that could cause even earlier reflections, though there is acoustic treatment on those. The back wall of the room is about 12 feet behind the listening position.

The musical selection was chosen for authentically capturing the breadth of the soundstage of a full orchestra with directional microphones, rather than spaced omnis, but also because that movement brings me joy. The adaptive filtering that I am applying to the mic array requires that I be able to model the phase and magnitude relationships of the sounds recorded by the mics, and spaced omnis or studio work with synthetic reverb would be totally unpredictable. Note, the filtering wasn't used for the recordings I shared, and it is still under development, but I did make some major improvements a day or 2 ago.

 

[Trdat]

I have two unique sound systems which I switch through depending on the recording. Some recordings lacking the reverberation sound better on my wider dispersion 3 way placed near the walls with 3 subs utilizing the room. Live music on the other hand sounds better on the horn narrower directivity with time aligned stereo subs as the track has more reverberation.

Of course the directivity plays a role when it comes to enjoyment and preference however, I wonder if the preference might be somewhere in the amount of reverbertion/reflections in the recording which is either offset or inhanced by the specific sound system.

 

[davidc]

Am I missing something or do you consider the pinna, concha, and the helixes are not part of our hearing mechanism?

You are correct if the point of the recording was to determine things like "soundstage," stereo/mono effects, and stuff like that. Listening with speakers puts our ears into play, but now we have added another inaccuracy (albeit one that we all live with when we listen with speakers to live recorded music) because we aren't considering the interaural crosstalk between the ears. If everyone here listened to a proper binaural recording through headphones or through properly set up Polk SDA speakers or a Carver Sonic Hologram, they would be impressed.
However, the OP is also correct. His method eliminates the interaural crosstalk but introduces the headphone effect.
So, you can still hear the differences between the recordings.
If you really want to do the experiment to determine soundstage and similar terms based on the speaker's orientation, you must do a binaural recording and listen with headphones or listen with your home speakers (with them being Polk SDA's or any speakers with Carver's Sonic Hologram). The problem with this choice is not everyone has Polk SDAs, and the setup of the Sonic Hologram with speakers is critical, and the sweet spot for listening is as well.
If any of ASR's members have either, I would love to hear their thoughts.

 

[davidc]

No contest for me. I prefer the second recording. It is narrower, but the first recording sounds diffuse and unfocused.

However, I’m not sure your mic technique here would correlate that strongly to preference in the actual room. Because reflections are coming from different directions, they undergo a different HRTF, which helps our brain separate them somewhat. When you’re recording with cardioid mics, that information is lost. What you’re capturing here is a recording with delays, not a spatial recording.

Agreed.

 

[davidc]

There was never any doubt that both recordings would lack in width compared to the original sound file as crosstalk is obviously at play with such recordings.

The reason I prefer the second recording doesn't have anything to do with the lack of stereo width compared to the first recording, it's more about faults caused by the recordings themselves and maybe also the positioning of the loudspeakers, and these faults are exaggerated and way more obvious in the first recording. If you compare the original sound file to the first recording you can easily hear that the first recording is much skewed to the right channel, this or a too large distance between the loudspeakers in the upright position causes the phantom-centered sounds to sound very "bland" and "washed-out". The fault with the lacking phantom center is way less obvious with the second recording even though it obviously lacks in width compared to the first recording, if that got to do with the loudspeakers laying down or if they are acoustically wise closer to each other, I don't know?


I notice a lot of faults with both of the recordings including the fault you are talking about, but because of the faults I mentioned above which are way more obvious in the first recording, I still prefer to go with the lesser width but with the better phantom-center of the second recording.


Again, both recordings have faults that deteriorate the sound and it's more about "picking your poison", don't you hear that the first recording is lacking a lot in the phantom center compared to the original sound file?

I think the lack of a phantom center is caused by both the skewed stereo image and also that the speakers being positioned too far apart from each other. So just for fun, I did a few adjustments to the first recording. Some adjustments to get a better balance to the stereo field and an EQ adjustment just to the phantom-centered sound to give it a bit more presence as the original file.
With the adjustments I prefer the first recording, without it I prefer the second recording.

https://www.dropbox.com/scl/fi/1zdp...ted.flac?rlkey=as6fqb5jy1h9zvkvwrcxoo3f9&dl=0

I hope that @BenB has no problems with my little manipulation of his recording.

You are correct about the crosstalk. I proposed a way to solve this in my previous post.