Perlisten S7t Speaker Review and Measurements (Audioholics)

[hardisj]

thank you for clearing this up! it was already clear that both speakers are very well designed.

Check my edit. I made a math error. I blame lack of sleep.

 

[abdo123]

Check my edit. I made a math error. I blame lack of sleep.

then this might be one of the best speakers in the world then!

 

[Matthew J Poes]

Hi @Matthew J Poes! Glad to see that you're on this forum as well!

for everyone else that didn't bother to read the full review, this is the distortion at 95 dB, which is basically below 1% for all frequencies above 100Hz.

View attachment 130884

One thing to keep in mind as well is how accurate this approach is to capturing the distortion of a speaker. There are a lot of unknowns. I talked with one of Harmans speaker engineers about measuring speaker distortion and what he told me was they always do it in an anechoic chamber with a 1” B&K capsule. It’s a very specific mic with extremely low noise and distortion. The mics own self noise and distortion ideally needs to be at least 10dB better than the DUT. The room should also be. While there are some tricks that let you stretch these margins, his comment to me was he doesn’t trust them.

Dan uses a G.R.A.S. Mic for his measurments so it would be interesting to know if they found the same results or if any of this is actually mic issues.

I had suggested James consider two of the tricks at his disposal but he was concerned it would make the measurements non-comparable. One method uses much longer FFT lengths (2M for example) and multiples (say 5-10). REW struggles to do this without a loopback but James can manage that.

the other option is to use the stepped sine procedure in REW. This is again a better method when it’s low distortion and a noisy environment.

neither overcomes mic limitations. I have distortion plots for 1/2” measurement mics and know what those often look like. The 1/4” polymer capsule used in EW mics is an unknown to me. Normally these are not the kind of mics used to measure distortion. They work fine with the distortion is fairly high. As soon as things get below 1% to .5% at higher output levels you can start to have questionable issues.

matt

 

[Descartes]

One thing to keep in mind as well is how accurate this approach is to capturing the distortion of a speaker. There are a lot of unknowns. I talked with one of Harmans speaker engineers about measuring speaker distortion and what he told me was they always do it in an anechoic chamber with a 1” B&K capsule. It’s a very specific mic with extremely low noise and distortion. The mics own self noise and distortion ideally needs to be at least 10dB better than the DUT. The room should also be. While there are some tricks that let you stretch these margins, his comment to me was he doesn’t trust them.

Dan uses a G.R.A.S. Mic for his measurments so it would be interesting to know if they found the same results or if any of this is actually mic issues.

I had suggested James consider two of the tricks at his disposal but he was concerned it would make the measurements non-comparable. One method uses much longer FFT lengths (2M for example) and multiples (say 5-10). REW struggles to do this without a loopback but James can manage that.

the other option is to use the stepped sine procedure in REW. This is again a better method when it’s low distortion and a noisy environment.

neither overcomes mic limitations. I have distortion plots for 1/2” measurement mics and know what those often look like. The 1/4” polymer capsule used in EW mics is an unknown to me. Normally these are not the kind of mics used to measure distortion. They work fine with the distortion is fairly high. As soon as things get below 1% to .5% at higher output levels you can start to have questionable issues.

matt

So simple question how do they sound compared to Revel Performa 328Be

 

[Matthew J Poes]

So simple question how do they sound compared to Revel Performa 328Be

Im not the one to ask. I hope to have that comparison available soon. Gene has the 328Be in house and the S7T is on its way to his house next. He is joking he is getting them to demo to my prospective clients. In any case, we will have a large selection of similarly priced speakers to compare against.

obviously I am going to have a bias if someone asks. I think the Perlisten is better. I can say more assuredly the build quality of the Perlisten is on par with the Ultima series. Not the Performa series. But I have found the sound of the Performa stuff as good or superior to the Ultima Salon2 I heard previously. So we shall see. I think I know what Gene will say but I rather let him say it and not ruin it. We had some issues getting the Revels to sound good in his room but haven’t had the same problems with the other speakers. Bass is weak and the natural room curve we are getting is overly flat. We tried applying ARC but it didn’t really address the issue. However we are reconstructing the room and so I expect the acoustics to change a lot.

If we do listening tests. I might try to find a way to ensure that we compare apples to apples here. I don’t want bass to dictate preference. That is very personal and very room dependent.

 

[amper42]

So simple question how do they sound compared to Revel Performa 328Be

If the FR response James shows in the S7T review is correct and starts to roll off at 80Hz then the Bass response of the Revel F328Be will be stronger. My F328Be is flat to 40Hz and only down -2dB at 29Hz in my REW measurements. It would be interesting how that FR difference sounds in a blind test with Bass Candy tracks.

As far as price, it's not hard to find new Revel F328Be speakers available in the $12K area from resellers that work with clients. I don't know if discounts on the S7T will be available or not?

 

[Matthew J Poes]

If the FR response James shows in the S7T review is correct and starts to roll off at 80Hz then the Bass response of the Revel F328Be will be stronger. My F328Be is flat to 40Hz and only down -2dB at 29Hz in my REW measurements. It would be interesting how that FR difference sounds in a blind test with Bass Candy tracks.

As far as price, it's not hard to find new Revel F328Be speakers available in the $12K area from resellers that work with clients. I don't know if discounts on the S7T will be available or not?

The anechoic -10dB point in ported setup is 22hz. Keep in mind these have way more output than the Revel. A good 10-13dB at least. My guess is the Perlisten actually has more bass.

as for Discounts, just ask one of us resellers.

 

[amper42]

Revel F328Be Sensitivity = (2.83V/1m) - 91 dB
Perlisten S7t Sensitivity = (2.83V/1m) - 92 dB

If these specs are correct the Perlisten S7t shouldn't be much louder than the F328Be at my 75 to 80dB listening levels. I would need to get into "ear protection" territory to notice the additional output.

 

[Matthew J Poes]

Revel F328Be Sensitivity = (2.83V/1m) - 91 dB
Perlisten S7t Sensitivity = (2.83V/1m) - 92 dB

If these specs are correct the Perlisten S7t shouldn't be much louder than the F328Be at my 75 to 80dB listening levels. I would need to get into "ear protection" territory to notice the additional output.

that’s misunderstanding my point. Listening at an average of 80dB requires a maximum clean and uncompressed output of about 100dB at your listening position. That means the speaker needs to be still clean in output at 110dB at 1 meter for a typical 3 meter listening distance. It’s already exceeding 10% and 3dB of compression. The Perlisten has another 10dB before it would hit the same point. You may not hear or notice anything. But then why would you if you haven’t heard it without that compression and distortion.

you can laugh all you want. For the same money that isn’t as capable of a speaker. Doesn’t make it a bad speaker. It’s a great speaker. The Perlisten is an arguably better speaker and that difference in output is more valuable than most realize. It’s why I’ve never loved direct radiator speakers. None of them could play realistically loud levels without onset of compression and distortion. This speaker happens to be an exception.

 

[richard12511]

Sensitivity and power handling alone won’t tell you a lot about maximum output. It should, but it doesn’t. I will be shocked if any of those come even close in terms of maximum output with low distortion. My guess is all of those will be compressing anywhere from 10-15dB sooner. I would also be shocked if anyone had spins nearly as good. There simply isn’t a lot of speakers for any amount of money that measure that well.

Would you take these over the Revel F328Be? They're similarly priced, and the Revel's spins look a little better. I guess the Perlisten's play a lot louder?

 

[richard12511]

that’s misunderstanding my point. Listening at an average of 80dB requires a maximum clean and uncompressed output of about 100dB at your listening position. That means the speaker needs to be still clean in output at 110dB at 1 meter for a typical 3 meter listening distance. It’s already exceeding 10% and 3dB of compression. The Perlisten has another 10dB before it would hit the same point. You may not hear or notice anything. But then why would you if you haven’t heard it without that compression and distortion.

you can laugh all you want. For the same money that isn’t as capable of a speaker. Doesn’t make it a bad speaker. It’s a great speaker. The Perlisten is an arguably better speaker and that difference in output is more valuable than most realize. It’s why I’ve never loved direct radiator speakers. None of them could play realistically loud levels without onset of compression and distortion. This speaker happens to be an exception.

That's crazy! Better than F328Be and Salon2? Perlisten should send one of these to @amirm.

What do you make of the directivity measurements? It's kinda hard to see with your non segmented polars, but at first glance the beamwidth looks pretty inconsistent compared to other SOTA speakers. The DI looks great, though, which is weird, as they must have great directivity.

*Edit: I think I'm just struggling with the lack of color separation and demarcation on those graphs. The general trend is a somewhat smooth triangle.

 

[phoenixdogfan]

Would be interesting to see a smaller floorstander. Something with two 7" woofers instead for four priced at around $7k. In other words their answer to the F226be.

 

[amper42]

The key phrase is, "You may not hear or notice anything."

 

[richard12511]

After reading the full review, these really do have my eyes open. Will be really interesting to hear Gene's comparison of them to his F328Be. Great for us as consumers to have another SOTA speaker to choose from.

A few questions for @Matthew J Poes

1. Are you aware of any good research on preferred vertical beamwidth and/or ceiling and floor reflection preferences. Toole was on @hardisj podcast recently, and he mentioned a study where it was found that people generally prefer to leave the floor reflection in tact, but I honestly don't know much else. I mention this because you note that they purposely aim for a very small vertical beamwidth(which should minimize that reflection) with this speaker. Logically, I think it makes a lot of sense, but I'm just not aware of any research to support that.

2. Where does the incredible dynamics come from? You note that they have ~10-13dB more output than the F328Be, but they're only 1dB more efficient, and seem to have comparable power handling(600 watts)? How do they achieve so much more spl? 12dB would be 8x the power rating, right?

3. How would you compare these to your Abbey's?(in terms of dynamics, tonality, and spaciousness).

 

[abdo123]

After reading the full review, these really do have my eyes open. Will be really interesting to hear Gene's comparison of them to his F328Be. Great for us as consumers to have another SOTA speaker to choose from.

A few questions for @Matthew J Poes

1. Are you aware of any good research on preferred vertical beamwidth and/or ceiling and floor reflection preferences. Toole was on @hardisj podcast recently, and he mentioned a study where it was found that people generally prefer to leave the floor reflection in tact, but I honestly don't know much else. I mention this because you note that they purposely aim for a very small vertical beamwidth(which should minimize that reflection) with this speaker. Logically, I think it makes a lot of sense, but I'm just not aware of any research to support that.

2. Where does the incredible dynamics come from? You note that they have ~10-13dB more output than the F328Be, but they're only 1dB more efficient, and seem to have comparable power handling(600 watts)? How do they achieve so much more spl? 12dB would be 8x the power rating, right?

3. How would you compare these to your Abbey's?(in terms of dynamics, tonality, and spaciousness).

power ratings are bullshit, the only thing that matters is how much distortion there is at playback volume.

and it has roughly 10 dB lower distortion at the same playback volume as the Revels.

 

[ctrl]

The first thing that stands out when comparing the manufacturer's measurements with @Matthew J Poes's are the glaring deviations of early reflections (ER) and sound power (SP) in the frequency range 2-4kHz - something doesn't add up at all.

You have to realize that these are "average values" from many measurements. SP deviations of more than 2.5dB in the range 2-4kHz are differences like day and night.

I consider Matthew's measurements to be more realistic, since the horizontal directivity mismatch in the 3kHz range is actually very clearly visible in the manufacturer's horizontal spectrogram, and should therefore also be reflected in the SP and ER (had this, like others, already addressed here).

Using Mathews measurements, we can take a closer look at the discontinuity of horizontal directivity. In a normal listening room, the lateral reflections are very important and should be as similar as possible to the axial frequency response (or slope evenly to it).

To do this, let's look at the axis frequency response and the 40° and 60° frequency response measurements (as a "proxy" for the lateral reflections). For the simplest possible visualization, we normalize to the axial frequency response (strait red line).

You can easily see that the horizontal directivity is really very good (up to 9kHz) except for the range 2-4kHz. This sudden increase of the horizontal sound pressure level in the range 2-4kHz, together with the hump in the SP (in the same range), should be clearly audible as "harsh sound" at the latest at high sound pressure levels, in a normally reflective listening room.
If this is true, it should be possible to fix this via EQ - at the expense of linear axis frequency response.

The causes for this have already been mentioned several times, so only briefly, the two tweeters above and below the waveguide radiate broadly (no waveguide) and interact strongly with the side edges of the speaker (the slight rounding of the side edges has little influence on this), which leads to a widening of the horizontal radiation in the range 2-4kHz.

Above 9kHz the directivity gets a bit chaotic, other waveguides with 1'' dome manage this often better and achieve a uniform directivity up to 12 - 15kHz.

 

[richard12511]

power ratings are bullshit, the only thing that matters is how much distortion there is at playback volume.

and it has roughly 10 dB lower distortion at the same playback volume as the Revels.

Rereading @hardisj comment above explains some (5dB) of the discrepancy.

I agree that power ratings are BS(Toole even says so himself).

I don't think you can compare the absolute distortion levels like that from different rooms and different mics to conclude "10 dB" lower like you are. We would need to see to both speakers measured in the same situation by the same mic to really see. Also, using the two graphs you posted, even if trying to compare 1 for 1(sorry for the scaling, did my best to get the vertical the same from 12ft away, but the horizontal is different)

Looking at that, I'm not sure I see a full 10dB as you say. More like 3-5ish most of the time, with maybe max of 9db(43 vs 34) at 350Hz? Also, distortion doesn't always increase linearly with SPL, so just because it's 5dB higher at one spl doesn't mean that will hold for another spl. We've seen examples of one speaker being higher at 86dB but then actually being lower at 96dB.

Assuming the distortion does scale the same, and assuming Amir's distortion measurements are directly comparable to Matthew's, along with the 5dB difference that Erin notes, that does go along way towards explaining that 10-13dB discrepancy. I've Matthews are relatively higher(even after accounting for 2m difference), that could explain more of the distance.

Those first two assumptions are almost definitely not true, though, so I'm still curious to hear from Matthew. I think we'd really need to see output testing like @hardisj does to see what the real output difference is. Matthew's also measured the F328Be, so I'm guessing that's what he's basing his 10-13dB figure on, which is kinda why I asked.

Still, I keep coming back to sensitivity similarity. Are we not still looking at 8 times the amplifier power needed to realize those 10-13dB more spls? Can those 7" Perlisten woofers really take ~8x the power of those 8" Revel woofers, and is that reasonable to expect from an amp/socket? With any sort of reasonable power handling estimate for the F328Be(200-1000watts?), we're still looking at ridiculous needed power levels that I'm not sure any home (US)wall socket or speaker can take. You'd almost have to assume the Revel can only take 100-200 watts, which seems silly, or am I missing something? 200 watts seems way too low honestly for an upper estimate for those 3 8" woofers. Given the sensitivity similarity, I think amplification power may be the spl limiting factor here, and not distortion. Maybe the sensitivities are very off?




Still, it really does seem like this speaker really may have an output advantage. I'm not convinced of 10-13dB yet(mainly due to amplifier power limitations), but I do think there is at least some output difference. Also, we're comparing the S7T to the lowest distortion measurement that Amir has found, and it's coming out on top by any reasonable guess, which is very impressive. Compared to any other speaker(say the D&D 8C, for example), the difference would be much larger.

I also trust Matt(one of the few reviewers I really do), and I know he's heard and measured both. I also agree with Matt that you can't just say "well I don't hear any distortion at the levels I listen to with my speakers". I've found that distortion/compression is kinda what determines that limit that we can enjoy and/or tolerate. I can enjoy/tolerate a much higher spl with my JTR 212RTs than I can with my Revel M105s. You really need to hear them side by side, as Matthew points out. I find that "way too loud" is often only way too loud because the speaker can't really handle that spl; switch to another much more dynamic speaker, and suddenly way too loud is "really awesome" and not painful at all.

Can't wait to hear Gene's side by side impressions. These look to be another truly SOTA speaker that I definitely need to consider. Would love to see NFS measurements of these from Amir(mainly to see a more comparable beamwidth measurement).

 

[abdo123]

Rereading @hardisj comment above explains some (5dB) of the discrepancy.

I agree that power ratings are BS(Toole even says so himself).

I don't think you can compare the absolute distortion levels like that from different rooms and different mics to conclude "10 dB" lower like you are. We would need to see to both speakers measured in the same situation by the same mic to really see. Also, using the two graphs you posted, even if trying to compare 1 for 1(sorry for the scaling, did my best to get the vertical the same from 12ft away, but the horizontal is different)

View attachment 131313

Looking at that, I'm not sure I see a full 10dB as you say. More like 3-5ish most of the time, with maybe max of 9db(43 vs 34) at 350Hz? Also, distortion doesn't always increase linearly with SPL, so just because it's 5dB higher at one spl doesn't mean that will hold for another spl. We've seen examples of one speaker being higher at 86dB but then actually being lower at 96dB.

Assuming the distortion does scale the same, and assuming Amir's distortion measurements are directly comparable to Matthew's, along with the 5dB difference that Erin notes, that does go along way towards explaining that 10-13dB discrepancy. I've Matthews are relatively higher(even after accounting for 2m difference), that could explain more of the distance.

Those first two assumptions are almost definitely not true, though, so I'm still curious to hear from Matthew. I think we'd really need to see output testing like @hardisj does to see what the real output difference is. Matthew's also measured the F328Be, so I'm guessing that's what he's basing his 10-13dB figure on, which is kinda why I asked.

Still, I keep coming back to sensitivity similarity. Are we not still looking at 8 times the amplifier power needed to realize those 10-13dB more spls? Can those 7" Perlisten woofers really take ~8x the power of those 8" Revel woofers, and is that reasonable to expect from an amp/socket? With any sort of reasonable power handling estimate for the F328Be(200-1000watts?), we're still looking at ridiculous needed power levels that I'm not sure any home (US)wall socket or speaker can take. You'd almost have to assume the Revel can only take 100-200 watts, which seems silly, or am I missing something? 200 watts seems way too low honestly for an upper estimate for those 3 8" woofers. Given the sensitivity similarity, I think amplification power may be the spl limiting factor here, and not distortion. Maybe the sensitivities are very off?




Still, it really does seem like this speaker really may have an output advantage. I'm not convinced of 10-13dB yet(mainly due to amplifier power limitations), but I do think there is at least some output difference. Also, we're comparing the S7T to the lowest distortion measurement that Amir has found, and it's coming out on top by any reasonable guess, which is very impressive. Compared to any other speaker(say the D&D 8C, for example), the difference would be much larger.

I also trust Matt(one of the few reviewers I really do), and I know he's heard and measured both. I also agree with Matt that you can't just say "well I don't hear any distortion at the levels I listen to with my speakers". I've found that distortion/compression is kinda what determines that limit that we can enjoy and/or tolerate. I can enjoy/tolerate a much higher spl with my JTR 212RTs than I can with my Revel M105s. You really need to hear them side by side, as Matthew points out. I find that "way too loud" is often only way too loud because the speaker can't really handle that spl; switch to another much more dynamic speaker, and suddenly way too loud is "really awesome" and not painful at all.

Can't wait to hear Gene's side by side impressions. These look to be another truly SOTA speaker that I definitely need to consider. Would love to see NFS measurements of these from Amir(mainly to see a more comparable beamwidth measurement).

I based my assumption based on the fact that Audioholics' measurments were at 101 dB at 1 meter, while still showing significantly lower distortion than the Revel at 96 dB at 1 meter (Amir's measurments).

Indeed it would be much more clear if both were measured by the Klippel, but we have to work with the data that we have.

 

[amper42]

Until the Revel F328Be and the S7T are measured in the same exact manner and listen to side by side it's hard to know which a listener might prefer. The Revel F328Be sounds absolutely great to me. I can't imagine wanting much more. I can easily use them without sub for music listening. Based on the S7T roll off at 80Hz a sub would most likely be required with it.

As far as distortion, Amir's review sums it up nicely for the F328Be.

 

[hardisj]

To be honest, all this discussion about max SPL is just conjecture until someone provides
1) measurements conducted in the same manner
2) description of how measurements are taken (and hopefully the method is reasonable)

I'd prefer the method I use for compression/enhancement testing, both short and long term. But I don't think I'll get that.

Until then I'm just looking at sensitivity, drivers used and crossovers and figuring both are reasonably in the same ballpark.

There also the matter of radiation pattern and room excitement. But that's uncharted territory.