Measure your C50 in REW and tell us how your bass sounds!

[MengW]

That's super high. Over 40 dB at 400Hz! How far was the microphone from the speakers when you took this measurement?

There was only 1.3 meter from the speaker to mic. I plan to update the layout to try out the 1.715 meter reference from Carl Tatz design distance later of this year.

In the directivity mode of w371, the crossover is locked at 180Hz only. In the complementary and side reduction mode of w371, the crossover range is 150~300Hz.
So the result of 400Hz should be presented by the 8351B.

And I filled ～90 pieces 40Kg/m³ density 1.2x0.6x0.05m polyester, and used some perforated acoustic panels and wooden strips to cover.
The reverberation time of the mid-to-high frequency may be a little short. So far, I could basically accept it.

 

[ernestcarl]

Another interesting topic, enjoying the ASR forum more and more. In the past few years, I have never paid attention to this C50 value.

Now, there is only one rented small room of 13 square meters, I think that the W371+AVAA should be the best and easiest and expensive solution for me, and then they did give me this magical and amazing result.

Found out these measurement results, fyi.
View attachment 270427
View attachment 270428
View attachment 270432View attachment 270431

Can you show us a default wavelet spectrogram view — in dB SPL and normalized, thanks!

Also, how many AVAA units were used?

 

[MengW]

Sorry, I'm not sure what is the "default wavelet spectrogram view — in dB SPL and normalized". Is this one?

 

[MengW]

Can you show us a default wavelet spectrogram view — in dB SPL and normalized, thanks!

Also, how many AVAA units were used?

There are 4 AVAAs now. However, the fourth avaa position is not the best and most effective position now.
I bought them about 2 years ago, at that time, I used them in 26.5m² living room of another apartment.
For the current small room of 13m² , I think that 2 or 3 AVAAs and polyester are enough to decay the sub bass energy, reaching the T30 or Topt of 50Hz under 400ms.

 

[ernestcarl]

Sorry, I'm not sure what is the "default wavelet spectrogram view — in dB SPL and normalized". Is this one?

Not quite...

Something like this:


By "default" I simply mean the "apply default settings" and dB SPL being the amplitude setting.

 

[MengW]

Not quite...

Something like this:
View attachment 270591 View attachment 270592

By "default" I simply mean the "apply default settings" and dB SPL being the amplitude setting.

Got it.

Directivity mode @180Hz, Complementary mode and bass management @100Hz

 

[ernestcarl]

Got it.

Directivity mode @180Hz, Complementary mode and bass management @100Hz

Which bass mode among the three sounded subjectively better to you?

For the curious, a bit more info I found in Genelec's manual:

The operating modes are:

Complementary mode where the two woofers operate independently to implement a flat frequency response; this is the default operating mode.

Directive mode uses the two woofers simultaneously to create continuous directivity matching with The One three- way coaxial monitoring loudspeaker; this mode enables the system to maintain constant directivity down to very low bass frequencies.

Null-steering modes use the two woofers simultaneously to enable a directive mode where the direction of minimum output (null) is set toward a problematic orientation in the room – enabling this mode reduces back wall, side wall or floor reflections.


Evaluating Modes

To evaluate different W371A directive mode options, go to the Group tab and select Duplicate Group. Then, in the duplicated Group, double-click the microphone icon to re-run AutoCal.

Press Skip AutoCal. After this, on the Woofer Cal page, double- click on the microphone icon to re-run the W371A calibration selecting your desired directivity options.

For example, you may wish to have four Groups for testing: monitor alone, monitor + W371A in the Complementary Mode, monitor + W371A in the Directive Mode, monitor + W371A in one of the Null Steering Modes. When auditioning, notice that it can take a relatively long time (up to a week) before all the qualities of each mode may be fully appreciated.


Selecting Crossover Regions

W371A calibration enables selecting the crossover frequency ranges. Your selections of the ranges are affected by room acoustics, W371A and listening locations. You may want to avoid notches in the monitor response. The default crossover range is 150-250 Hz. You can select a higher range up to 300 Hz and no narrower than 50 Hz wide. The measurement curves and subjective evaluations may guide your preferences as well as your choice of the W371A operating mode. When the W371A is positioned behind a large console obscuring the upper front woofer, a low crossover range below 250 Hz is advisable.

BTW, which operating mode has been enabled in "bass management @100Hz" in your last image?

 

[MengW]

在您看来，三种低音频模式中哪种低音频模式听起来更好？

Think about something, try to talk about it.

There are five modes of W371.

1.Directivity mode
The highest resolution, the most details of playing skills
Bass tromebone, horn port image size is the smallest, clearest and most stable.

2.Complementary mode
Most comfortable, most balanced

3.Back wall / side wall / floor Reduction mode
Side wall reduction mode is best of 3 reduction modes for my room, the feeling is between directivity and complementary mode.

4.Bass management 8351B+7360
Sub bass sound is the richest and warmest. Should be better to play movie and some pop music.

For example，Hal Robinson The Quad vol. 3 Mendelssohn Symphony 4 mvt 4

The height of the image position is obviously different. The sound axis of W371 is 0.868m high, but the 7360 subwoofer is only about 0.3-0.4m.

W371 is closer to what I heard one real musical instrument playing.

If play it with subwoofer, the feeling of box resonance is a little more, which is warmer and pleasant. But it will feel a little unnatural, like as putting the double bass on the wooden podium to play, there is an extra resonance box below it. The third and fourth lower strings playing are more obvious.

I'm not professional, just fyi.

 

[MengW]

Can you show us a default wavelet spectrogram view — in dB SPL and normalized, thanks!

Also, how many AVAA units were used?

The wavelet spectrogram view is new to me. And I think it‘s interesting and try to understand it, but it's not easy to get the key points.
Could you share any skill or lessons learned with us to understand it? What is the ideal or better result view? Or which one is better than another one at one frequency range?
If the purpose is to analyze the bass clarify, is this red band narrower in the vertical direction more clear?
Do you think that a narrower red band means a shorter time from launch to stop of woofer? Or from launch to the room decay?
Is there any reference materials of the Wavelet Spectrogram?
Thanks.
Meng

 

[MengW]

The wavelet spectrogram view is new to me. And I think it‘s interesting and try to understand it, but it's not easy to get the key points.
Could you share any skill or lessons learned with us to understand it? What is the ideal or better result view? Or which one is better than another one at one frequency range?
If the purpose is to analyze the bass clarify, is this red band narrower in the vertical direction more clear?
Do you think that a narrower red band means a shorter time from launch to stop of woofer? Or from launch to the room decay?
Is there any reference materials of the Wavelet Spectrogram?
Thanks.
Meng

I found out that the best guide is the REW help page.

Spectrogram Graph

And there are some wavelet introduction in these two videos.

 

[Toltek]

That's an astoundingly good result in the bass and great all the way across the spectrum. Can you post the RT-60 measurements? I'm guessing they've got to be really low.

Thanks Tim! I've made some changes since then, only slightly moving things around and playing with DSP

High RT time at 50 and below seems to be due to the fact that for whatever reasons, REW used a later portion of the decay after 50 ms, where it already dropped -20 db from peak

Here is a full RT60 decay in 1/3 octave, 60 db range from peak on SPL axis. You can see how RT60 is calculated for lower frequency. I know, it is not applicable, anyways.

C50 is calculated to be above 20 db in 50 to 100 range, accordingly, D50 is 99%+

Bass FR, no smoothing

Pink noise MMM, 1/12 octave bands, no smoothing

Spectrogram at 40 db range, wavelet 1/6

Same but in percentage and shorter time scale

Distortion at my "loud" levels is below the noise floor above 60 Hz

 

[ernestcarl]

The wavelet spectrogram view is new to me. And I think it‘s interesting and try to understand it, but it's not easy to get the key points.
Could you share any skill or lessons learned with us to understand it? What is the ideal or better result view? Or which one is better than another one at one frequency range?
If the purpose is to analyze the bass clarify, is this red band narrower in the vertical direction more clear?
Do you think that a narrower red band means a shorter time from launch to stop of woofer? Or from launch to the room decay?
Is there any reference materials of the Wavelet Spectrogram?
Thanks.
Meng

REW's "wavelet" mode shows us a particular filtered wavelet transform graphical view which makes it easier to see certain details/aspects of the measured IR normally not obvious by just looking at the steady state curves. Unlike the Fourier mode, the time spread appears to us more intuitively/naturally because it already applies a form of frequency-dependent windowing -- where higher frequencies develop in a much shorter amount of time and lower frequencies for longer amounts of time.

https://drc-fir.sourceforge.net/doc/drc.html#sec242

Wavelets have the advantage of being easier to map to a logarithmic frequency scale. To further help the correct interpretation of the graphs the time scale is also stretched, depending on the frequency ... The end result is a graph that provides a tiling of the time-frequency plane which is visually quite close to the kind of time-frequency analysis performed by our auditory system.

Using a higher resolution setting in the spectrogram increases the magnitude contribution/bias while lower resolutions decrease the magnitude bias in favor of time. The default can be okay as your starting point, but I would change or cycle between different settings (including time scales e.g. 50ms vs 200ms) along the way. If I were mainly interested about the peak energy time, for example, I would use "Linear (% peak)" from the drop-down amplitude settings.

Resonances may appear more clearly as vertical "streaks" and reflections as horizontal streaks. Our rooms are rather complicated, though, so the graph usually appears even more "busy" or splotchy than some of the examples shown in the REW manual. Additionally, one can also visualize pre-ringing and/or pre-echo artifacts e.g. small ripples and/or long curved streaks appearing right before t=0 if using FIR "correction" wrongly/excessively -- for this, I use 1/2 or 1/3 resolution.

Depending on what filtering is used, a correlation can be seen more clearly between the GD curve and wavelet's peak energy time as well.


The ideal, perfect "dirac delta" impulse response would be like this (REW wavelet mode):

example from REW manual


or even like this (step response):

example from DRC documentation

You can get something closer to this "perfect ideal" if you listen outdoors well-away from boundaries or move into an anechoic chamber.

From your graphs, we can see that the peak energy time flows very evenly, however, not quite "transient perfect" as would be the case when everything's aligned exactly at t=0. This isn't normally an issue as long as there isn't any excessive discontinuity, smearing/streaking, and/or splotchiness (from delayed-reflected higher magnitude SPL energy) seen over time.

The more you compare between REW's different views, it should eventually become easier to see how the clarity curves -- C50, C80, and D50 -- and other graphical views such as spectral decay and RTx correlate with each other.

https://www.fulcrum-acoustic.com/wp-content/uploads/2019/10/spectrogram-loudspeaker-transient-response-2005.pdf

 

[ernestcarl]

And there are some wavelet introduction in these two videos.

Wavelets (the real thing, not the software) intro: exceptionally good tutorial.

An amazing video imho. Captures the essence of those beings, fundamentally and intuitively, in a very accessible manner!

www.audiosciencereview.com

 

[MengW]

REW's "wavelet" mode shows us a particular filtered wavelet transform graphical view which makes it easier to see certain details/aspects of the measured IR normally not obvious by just looking at the steady state curves. Unlike the Fourier mode, the time spread appears to us more intuitively/naturally because it already applies a form of frequency-dependent windowing -- where higher frequencies develop in a much shorter amount of time and lower frequencies for longer amounts of time.

https://drc-fir.sourceforge.net/doc/drc.html#sec242


在频谱图中使用更高分辨率的设置会增加幅度贡献/偏差，而较低的分辨率会降低幅度偏差，有利于时间。默认值可以作为您的起点，但我会在此过程中更改或在不同设置（包括时间尺度，例如 50 毫秒与 200 毫秒）之间循环。例如，如果我主要对峰值能量时间感兴趣，我会使用下拉幅度设置中的“线性（% 峰值）”。

共振可能更清晰地显示为垂直“条纹”，而反射则显示为水平条纹。不过，我们的房间相当复杂，因此该图通常比 REW 手册中显示的某些示例显得更加“繁忙”或更加斑点。此外，如果错误/过度使用 FIR“校正”，还可以看到预振铃和/或预回声伪影，例如在 t=0 之前出现的小波纹和/或长弯曲条纹——为此，我使用 1/2或 1/3 分辨率。

根据所使用的滤波，可以更清楚地看到 GD 曲线和小波的峰值能量时间之间的相关性。


理想的、完美的“dirac delta”脉冲响应应该是这样的（REW 小波模式）：

REW 手册中的示例


甚至像这样（阶跃响应）：

View attachment 270723
DRC 文档中的示例

如果您在远离边界的户外聆听或搬进消声室，您可以获得更接近这个“完美理想”的东西。

从你的图表中，我们可以看到峰值能量时间流动非常均匀，但是，并不像所有东西都在 t=0 时完全对齐时那样“瞬态完美”。这通常不是问题，只要随着时间的推移没有看到任何过度的不连续、拖尾/条纹和/或斑点（来自延迟反射的更高幅度的 SPL 能量）。

您在 REW 的不同视图之间比较得越多，最终就会越容易看出清晰度曲线（C50、C80 和 D50）以及其他图形视图（例如光谱衰减和 RTx）如何相互关联。

https://www.fulcrum-acoustic.com/wp-content/uploads/2019/10/spectrogram-loudspeaker-transient-response-2005.pdf

Got it! And I realized that I needed to go back and review some REW tutorials to discover treasures I had missed. Thanks.

 

[ozzy9832001]

This could be coincidence, but I've noticed that C50 values over 20 from 120hz and below cause significant ringing. I think that's just for me though because 120 is on of my room modes. It also doesn't seem to be EQ dependent per se, so I'm not sure how it actually derives the numbers. To me, the best indication of tight bass is the decay times.

REW's "wavelet" mode shows us a particular filtered wavelet transform graphical view which makes it easier to see certain details/aspects of the measured IR normally not obvious by just looking at the steady state curves. Unlike the Fourier mode, the time spread appears to us more intuitively/naturally because it already applies a form of frequency-dependent windowing -- where higher frequencies develop in a much shorter amount of time and lower frequencies for longer amounts of time.

https://drc-fir.sourceforge.net/doc/drc.html#sec242


Using a higher resolution setting in the spectrogram increases the magnitude contribution/bias while lower resolutions decrease the magnitude bias in favor of time. The default can be okay as your starting point, but I would change or cycle between different settings (including time scales e.g. 50ms vs 200ms) along the way. If I were mainly interested about the peak energy time, for example, I would use "Linear (% peak)" from the drop-down amplitude settings.

Resonances may appear more clearly as vertical "streaks" and reflections as horizontal streaks. Our rooms are rather complicated, though, so the graph usually appears even more "busy" or splotchy than some of the examples shown in the REW manual. Additionally, one can also visualize pre-ringing and/or pre-echo artifacts e.g. small ripples and/or long curved streaks appearing right before t=0 if using FIR "correction" wrongly/excessively -- for this, I use 1/2 or 1/3 resolution.

Depending on what filtering is used, a correlation can be seen more clearly between the GD curve and wavelet's peak energy time as well.


The ideal, perfect "dirac delta" impulse response would be like this (REW wavelet mode):


example from REW manual


or even like this (step response):


example from DRC documentation

You can get something closer to this "perfect ideal" if you listen outdoors well-away from boundaries or move into an anechoic chamber.

From your graphs, we can see that the peak energy time flows very evenly, however, not quite "transient perfect" as would be the case when everything's aligned exactly at t=0. This isn't normally an issue as long as there isn't any excessive discontinuity, smearing/streaking, and/or splotchiness (from delayed-reflected higher magnitude SPL energy) seen over time.

The more you compare between REW's different views, it should eventually become easier to see how the clarity curves -- C50, C80, and D50 -- and other graphical views such as spectral decay and RTx correlate with each other.

https://www.fulcrum-acoustic.com/wp-content/uploads/2019/10/spectrogram-loudspeaker-transient-response-2005.pdf

I think this is one of the most useful posts I've ever read in terms of interpreting the graphs. I know this sounds awful, but I didn't even know there was a manual.

 

[ernestcarl]

I've noticed that C50 values over 20 from 120hz and below cause significant ringing. I think that's just for me though because 120 is on of my room modes. It also doesn't seem to be EQ dependent per se, so I'm not sure how it actually derives the numbers. To me, the best indication of tight bass is the decay times.

High C50 values should not cause increased ringing... so something else like speaker and/or room mode resonances (or some other distortion) may be the culprit there instead. Have you tried playing bass "burst" test tones (can be done with REW) while listening and walking around the room to find the apparent source/location of the ringing? There are other test signals one can use as well -- although, you might try a simple (high SPL) slow/lengthy swept sine signal, too.

 

[ernestcarl]

Unlike the Fourier mode, the time spread appears to us more intuitively/naturally because it already applies a form of frequency-dependent windowing -- where higher frequencies develop in a much shorter amount of time and lower frequencies for longer amounts of time.

Or, perhaps, "scaling" may be a better term -- one often finds such plots described as a scalogram:

https://www.audiomatica.com/wp/wp-content/uploads/ALMA_2011.pdf

 

[Flaesh]

measured onboard souncard:

 

[ozzy9832001]

High C50 values should not cause increased ringing... so something else like speaker and/or room mode resonances (or some other distortion) may be the culprit there instead. Have you tried playing bass "burst" test tones (can be done with REW) while listening and walking around the room to find the apparent source/location of the ringing? There are other test signals one can use as well -- although, you might try a simple (high SPL) slow/lengthy swept sine signal, too.

100%. Room modes at 105, 120 and 140. Creates a gigantic mess with the fundamentals and the harmonics. Mid bass was a mess. Takes many, many very thick traps to get it under control and even then, still not even close to where I'd like it.

 

[tmuikku]

Hi,
I've been interested about stereo listening triangle size, and how to extend it, how to get good sound further in room (DIY speakers). This is apartment living room with no acoustic treatment other than furniture. Room is open plan to kitchen, asymmetric positioning on the long side, about 8mx6m or so, 2.5m ceiling.

Something happens in between, around 2.2meters there is quite clear transition from good imaging to slightly blurred one, like stepping in / out to good sound, some threshold gets crossed so that brain changes the impression of it. Difference isn't very big in these clarity graphs so its interesting and thought to post it even if its not bass motivated

Measurements are at 2m and 2.6m distance from left speaker.

Same for the right


I thought ~1-2kHz dip on left was flutter echo as its relatively constant between the two, but not sure what it is. Perhaps problem in speaker as its DIY after all right speaker data looks more plausible.