• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Spinorama for the brand new JBL PRX900 serie

If you have a band, or a DJ, there will be subs.
I think that's an aspirational statement. Plenty of bands get by without subs. Plenty of bands don't always need PA systems.

Even back when Pan Sonic was touring they weren't in a position to demand subs. And that's one of the few bands I can remember when the lack of the bottom octave was my main complaint about the sound reinforcement.

I guess this has changed in the last two decades with the dominance of EDM aesthetics in so much popular music and the PA's market advances in the bottom octave. But it changed quantitatively, not to the point of categorical truth.
 
Be very careful with those displays. The bandwidth theorem (iirc) is a bit like Heisenberg uncertainty. If you want to measure frequency accurately you need a lot of time. So the windows and sampling used in the FFTs for real-time displays that update several times a second smear the lower octaves enormously, hiding the truth in a big hump that's really just an artifact of the math involved.

This is where FFT windowing, size, and sample rate becomes a rather touchy art. On the one hand we want the overall FFT time slice to be short enough so that we can see how it corresponds to what we can hear (e.g. bass guitar, drums, etc) but if it is too short then we simply cannot see the peaks. And a really low peak, say at 41 Hz looks, the lowest a 4-string bass in standard tuning can make, looks like a

When I was using FFTs to look at the resonances in my acoustic guitars together the bass modes of my music room, I generally used maximum samples and reduced the sample rate in whatever app it was, and I wasn't looking below 75 Hz.

I made a video for you to show how dramatic this in the lower audible octaves.
Ah, that's interesting. I downloaded some sine tones, played them in Neutron Player and you're right that you can't tell the difference between say a 80Hz sine tone and a 30Hz sine tone - all of the bands in the bass become activated in the spectrum analyser! How do you properly analyse the frequency content in a track then?
 
I see so many PA speakers on your site recently @pierre !

Comparing the PRX912 to the Yamaha DXR 12 mkII, I would take the Yamaha's over the JBL though. They have very similar spins, although I do find the Yamaha's flatter through the crossover region and the midbass 100-300 hz. But for their intended use, there are other requirements that are probably more relevant than the spins only ofcourse.

Where did you find these measurements of the Yamaha's? :)
Edit: Ah, are these the files?
 
Last edited:
I think that's an aspirational statement. Plenty of bands get by without subs. Plenty of bands don't always need PA systems.

Even back when Pan Sonic was touring they weren't in a position to demand subs. And that's one of the few bands I can remember when the lack of the bottom octave was my main complaint about the sound reinforcement.

I guess this has changed in the last two decades with the dominance of EDM aesthetics in so much popular music and the PA's market advances in the bottom octave. But it changed quantitatively, not to the point of categorical truth.
According to this, most rock bands with a PA are using subs. https://www.talkbass.com/threads/does-your-band-use-a-sub-woofer-as-part-of-your-pa.1426793/
 
I see so many PA speakers on your site recently @pierre !

Comparing the PRX912 to the Yamaha DXR 12 mkII, I would take the Yamaha's over the JBL though. They have very similar spins, although I do find the Yamaha's flatter through the crossover region and the midbass 100-300 hz. But for their intended use, there are other requirements that are probably more relevant than the spins only ofcourse.

I agree. Yamaha is usually doing a good job at audio stuff.

Where did you find these measurements of the Yamaha's? :)
Edit: Ah, are these the files?

no. that’s the CFL files I use the GLL ones. That’s 2 differents formats (both proprietary) but GLL is better designed with extension in mind. I like it. It would be a lot better if it was open sources but it’s not. The company being it is https://www.afmg.eu. They are behind the EASE software that looks to be used in many many places.

The GLL viewer is free to use.
 
Last edited:
Thank you @pierre for a high quality post.
Thank you @amirm for front page love.
Kudos to JBL for providing full spin data for their customers.
 
How do you properly analyse the frequency content in a track then?
Like most genuinely interesting questions, the answer is: it depends. As I said above, once you are choosing window functions, sample rates, FFT size, averaging and smoothing then you are working in the subjective domain. And I honestly don't know how to remove the subjectivity. So ... maybe we can find some shelter for subjectivity under your adjective "properly". As in, "Team Y scores again with a superb attack despite a proper defense from Team X.

Just for example, say you decide on a way to produce a single spectrum graph of an audio file, all of it, not an animation. Now you could run that on the whole audio file of the track or you could split the track into segments. Right there, that choice alone is outside the domain of objectivity. So, give that ASR has technicians with the expertise of experience,
I would answer idk and defer to them.

I am a musician and know quite a lot about music and something about physics and acoustics. I like to think about what's on the track (letting others measure it). What could possibly be in there? Humans are generally said to be able to hear 20 to 20k Hz but that's absolutely no reason to believe that there's any signal there.

It reminds me of a time at the Deadverse studio with Oktopus mixing our second album and one of us asked for more deep rumble in such-and-such a part. After a bit of to-and-fro Oktopus demonstrated that boosting that frequency band did nothing because there's nothing there to boost. We would have to track something new to get the effect. The foolish request for more deep rumble was withdrawn.
 
Last edited:
Like most genuinely interesting questions, the answer is: it depends. As I said above, once you are choosing window functions, sample rates, FFT size, averaging and smoothing then you are working in the subjective domain. And I honestly don't know how to remove the subjectivity. So ... maybe we can find some shelter for subjectivity under your adjective "properly". As in, "Team Y scores again with a superb attack despite a proper defense from Team X.

Just for example, say you decide on a way to produce the spectrum graph of an audio file, all of it, not an animation. Now you could run that on the whole audio file of the track or you could split the track into segments. Right there, that choice alone is outside the domain of objectivity. So, give that ASR has technicians with the expertise of experience,
I would answer idk and defer to them.


I am a musician and know quite a lot about music and something about physics and acoustics. I like to think about what's on the track (letting others measure it). What could possibly be in there? Humans are generally said to be able to hear 20 to 20k Hz but that's absolutely no reason to believe that there's any signal there.

It reminds me of a time at the Deadverse studio with Oktopus mixing our second album and one of us asked for more deep rumble in such-and-such a part. After a bit of to-and-fro Oktopus demonstrated that boosting that frequency band did nothing because there's nothing there to boost. We would have to track something new to get the requested effect. The foolish request for more deep rumble was withdrawn.
Yeah, the bit I bolded in your post - so if you do this then you remove all the sample rate, FFT size issues? What software can you use for that, I'd be quite interested to put some of my tracks through it. I think you'd do it on the whole track and also on segments that you know have the most deep bass, that way you can see what you're asking your playback equipment to do. (Afterall, our little back & forth started because we were debating on the applicablity of using subs or not with these PA speakers)
 
Yeah, the bit I bolded in your post - so if you do this then you remove all the sample rate, FFT size issues? What software can you use for that, I'd be quite interested to put some of my tracks through it. I think you'd do it on the whole track and also on segments that you know have the most deep bass, that way you can see what you're asking your playback equipment to do. (Afterall, our little back & forth started because we were debating on the applicablity of using subs or not with these PA speakers)
I have only recently used:
  • Reaper which comes with a spectrum analyzer as I demonstrated in the video, but this doesn't seem suitable to producing a single hi-res characteristic spectrum for long segment (e.g. 10 seconds of music)
  • Visual Analyser https://www.sillanumsoft.org/ idk if it can do it. it is certainly very flexible
  • REW has a decent analyzer but again idk what options it has for combining a sequence of hi-res FFTs
Tbh, idk if it makes much sense to attempt this. I mentioned the idea only to illustrate that frequency analysis can be a tricky business and finding a good position among the inherent trade-offs of the various choices involves subjectivity.
 
I have only recently used:
  • Reaper which comes with a spectrum analyzer as I demonstrated in the video, but this doesn't seem suitable to producing a single hi-res characteristic spectrum for long segment (e.g. 10 seconds of music)
  • Visual Analyser https://www.sillanumsoft.org/ idk if it can do it. it is certainly very flexible
  • REW has a decent analyzer but again idk what options it has for combining a sequence of hi-res FFTs
Tbh, idk if it makes much sense to attempt this. I mentioned the idea only to illustrate that frequency analysis can be a tricky business and finding a good position among the inherent trade-offs of the various choices involves subjectivity.
I think I've seen @RayDunzl create an "average frequency response graph" for a track, but I don't know how he did it. I'm assuming this takes away the drawbacks you mentioned of real time frequency analysers?, the common spectrum analyser a lot of us probably know and have used.
 
I think I've seen @RayDunzl create an "average frequency response graph" for a track, but I don't know how he did it. I'm assuming this takes away the drawbacks you mentioned of real time frequency analysers?, the common spectrum analyser a lot of us probably know and have used.
Not on its own. You still need to choose the FFT parameters given the frequency-domain resolution you want. Then you could start averaging a sequence of spectra.

But take care! What if the one musically interesting thing that happens in the sub 50 Hz space is only apparent in one or two of them and you average thousands?
 
I think I've seen @RayDunzl create an "average frequency response graph" for a track, but I don't know how he did it.

Drag a track into the RTA window of REW,

Select the desired smoothing parameters before the drag.

View in the RTA window or view the saved trace

1667961331098.png


1667961371339.png
 
Many options to analyze tracks, here is some. They will show exactly what is there.

Audacity:

Spectrogram using SOX:

Good realtime analyzer for foobar2000 users:
 
I appreciate the relatively flat response for "What they are", but one issue I have with almost all P.A. speakers of this type and size..............

They can MEASURE relatively flat and have huge sound output, but for whatever reasons, tend to never actually sound like a normal Hi-Fi home speaker in some key regards.

At the huge church auditorium I work for, it has GREAT sound for What it is, and they spared no expense for sure, but I always get the feeling it is Great PA speaker sound and never truly great Actual speaker sound that is like a great home speaker.

Not bashing on P.A or sound reinforcement speakers, but I am never either to remove those colorations and things that sorta irk me about the sound, that appear to just be part of Very efficient high output speakers.

With our live music is is really good sounding.
With anything pre-recorded commercial music, it just somehow always sounds fairly good, but always like a P.A. speaker, for lack of a better term.
 
Excellent, that's very cool! Do you have any thoughts on some of the pitfalls that @Multicore has been mentioning?

Well, I suppose there are different ways to collect and look at data.

For me, the point was to see how playback through speakers related to the source.

I figured if the peaks match up, then what goes on at lower levels likely matches too.

I came close enough for my purposes, I think.

Below, Left and Right of the CD source (no EQ), and the combined result coming from the speakers measured at the listening position, at 10 feet (with EQ)...

Top trace is the "in-room", with considerable DSP to make the in-room frequency response follow the source. The "room curve" applied via DSP is flat - no curve or slope.

The speakers are playing the stereo source, so what is picked up by the microphone is the sum of the two channels.

The most inaccurate area is the bass, below maybe 80Hz, background noise intrudes on the silence of the source below 25Hz, and the rest of the bass response is a bit ragged.

Such is life.

index.php
 
Last edited:
Well, I suppose there are different ways to collect and look at data.

For me, the point was to see how playback through speakers related to the source.

I figured if the peaks match up, then what goes on at lower levels likely matches too.

I came close enough for my purposes, I think.

Below, Left and Right of the CD source (no EQ), and the combined result coming from the speakers measured at the listening position, at 10 feet (with EQ)... Top trace is the "in-room".

index.php
Now that's a really interesting approach.

You said CD so I guess 44.1k sample rate was the basis for the FFT. What FFT size did you use and what frequency domain resolution does that provide? EDIT: is that visibe in the screengrabs? 128 KiPoint FFT?

If I were to do this then I would prepare fairly short test files (between 5 to 30 seconds long?) by selecting segments of music that are specially relatively static. I don't mean it has to be a drone, not that static, but also not a whole symphony. In other words, I want to be able to relate what I can hear in the music to the peaks. Looking at a spectrogram as an intermediary step and to select segments would probably be useful. I would use Reaper for that, because it's my preferred DAW.
 
Last edited:
Back
Top Bottom