SPL vs Frequency Graphs

avaserfi

avaserfi

Audioholic Ninja
SPL vs Frequency Graphs

Well earlier I put up a note in a thread saying if someone needed an explanation on what a SPL vs Frequency graph means and how to read one I got a couple replies so I figured instead of answering a few PMs I would just create a thread. If there are any mistakes please correct me as I am in no way a expert, but I feel like I understand the subject that seems to come up quiet often and that I should be able to express myself well enough for others to understand.

This will probably include some information that most posters will find common knowledge, but my hope is that someone who has never purchased a speaker before will be able to read and understand this.

Here is a picture of a graph I am talking about:



As you can see the X axis has frequency and Y axis has Decibel (dB). The frequency corresponds to the wavelength, which basically tells you the pitch of the note so 20,000hz is a very high pitch tone while 20 is very low (Human hearing is generally 20hz-20khz but lower can be felt). The decibel rating refers to the loudness of the sound and is on a logarithmic scale which means 20dB is not twice as loud as 10dB and a change of approximately 6dB is twice as loud or half as quiet depending the direction.

So what does this all tell us?

Well many reputable speaker manufacturers will give you a frequency response like this one: "Frequency Response: 55 Hz to 20 KHz (± 3 dB)." This lets us know that from the low end of the spectrum of 55Hz all the way up to 20KHz this speaker will play a frequency sweep (a tone that starts at one frequency and plays frequencies in between up to its limit or vica versa) without having a loudness difference of 3 decibels in either the positive or negative direction. Imagine a speaker playing at 90dB with this example on the whole given it will be no be within the range of 87-93dB.

Why does this matter?

This is important because a speaker that is not linear in the range needed could sound bad. Here is an example: Say you are listening to a CD and two people are singing. One is a man who is a bass with a very low voice and the other is a woman with a higher voice. Now say your speakers aren't very linear across the spectrum. This could cause the man's voice could be quieter or louder than the woman’s which just wouldn't sound right! While some smaller differences might not be audible a difference of 6dB could make one of the singers sound like a whisper compared to the other.

Some people are more susceptible to noticing these differences in loudness compared to others, but it has been shown that the smallest audible changes varies with the frequency but at about 1KHz it is 3 dB and at 35Hz is 9dB.

The human vocal range is quiet broad running from about 60Hz to 1600Hz that is coincidentally the most sensitive range of human hearing. Instruments play similar frequencies but are able to go both lower and higher depending on the type.





Acoustics:

Room acoustics will play an important role in how a speaker acts in your specific room. In general manufactures use an anechoic chamber that will act completely different from your room! Proper placement of room acoustics will help you get a better response. Remember this if you are purchasing speakers room acoustics can help you as much as if not more than speakers so budget properly.

Acoustics do affect all frequencies but in different ways. With the high end you are correct the waves will reflect off walls and can cause a "harsh" sound or muddy the sound up, but this is not the same as what happens with lower frequencies. When one hears muddied dialog or a harsh sound this is from the same signal produced by a loud speaker hitting various reflection points then being reflected towards one ears at slightly different time. The brain has trouble separating all the information and it becomes muddied.

Room nodes (specifically how sound waves interact with a given room) play more a role in completely canceling the waves out or making them double up on themselves with the lower frequencies that can cause more nulls and peaks in specific response graphs.

Different frequencies respond differently to rooms due to wavelength size.

Other concerns/Information?

Frequency response isn't the only thing to look at when comparing speakers. Be sure to look at all other specs, but more importantly audition speakers, as many as you can, and if they sound good to you then they are what you want! (Again remember the showroom will sound different than yours)

Also, there are two types of frequency response, on axis and off. Most measurements you will see are on axis responses that are important, but nearly as important are off axis frequency responses. These are taken at a certain angle from the driver, but on the same plane. This is important if you are going to be listening to these speakers from across the room due to that important concept, room acoustics.

There are a variety of factors that cause differing plots in speakers from crossover design, driver selection, to cabinet design/bracing and much more. Pretty much anything and everything from the speaker or room can affect how linear its response is.

One last note on speaker linearity: This primer was written on a speakers frequency response with the idea that a linear response is more desirable that not. Speaker linearity is another term that refers to a speaker’s ability to increasingly powerful signals by generating louder tones proportionally to signal strength. Speakers differ in their ability to play linearly both on the upper and lower ends. A large linear response is desirable in this case as it makes for a more real performance as it will create the illusion of a musician being in your room rather than in the speaker itself.

Please let me know what you guys think and feel free to add to this or correct me! I am sure I left something out/made some mistakes.

Hope this helps and if you need more explanation let me know by posting or through PM and I will get back to you ASAP.

Lastly, thanks to Swerd for his corrections/additions.
 
Last edited by a moderator:
AcuDefTechGuy

AcuDefTechGuy

Audioholic Jedi
I noticed many companies say something like, "20Hz-20kHz +/-3dB", but if another speaker also has the spec. "20Hz-20kHz +/-3dB", does it mean they sound pretty much the same?

I noticed they don't say, "20Hz-20kHz @ 90dB +/-3dB". But is it possible that one speaker may be @ 100dB +/-3dB, and another speaker may be @ 80dB +/-3dB?
 
avaserfi

avaserfi

Audioholic Ninja
I noticed many companies say something like, "20Hz-20kHz +/-3dB", but if another speaker also has the spec. "20Hz-20kHz +/-3dB", does it mean they sound pretty much the same?
Wouldn't it be nice if it was that simplistic? Then we wouldn't have to audition speakers, but remember there are many other types of specs and in the end they will all interact differently with the room they are put in.

That just means they will act similarly over the given frequency range not sound the same.


I noticed they don't say, "20Hz-20kHz @ 90dB +/-3dB". But is it possible that one speaker may be @ 100dB +/-3dB, and another speaker may be @ 80dB +/-3dB?
Yes, some speakers will be able to have a higher output while staying linear than others. That is why having a dedicated sub is a plus ;).
 
M

MDS

Audioholic Spartan
I noticed they don't say, "20Hz-20kHz @ 90dB +/-3dB". But is it possible that one speaker may be @ 100dB +/-3dB, and another speaker may be @ 80dB +/-3dB?
No speaker plays at X dB. The output SPL is dependent on many factors including the sensitivity of the speaker in combination with the amount of power you feed it and as always room acoustics.

The +/- 3 dB spec tells you the variation of the input from the output over the stated range. Music is a combination of many waveforms and thus many frequencies and each individual frequency was given a specific level relative to the others when it was mixed and mastered. Of course the level of any particular frequency changes too - it's not like every instance of 300 Hz will be at the same level all the time. Your voice has a fundamental frequency but you can talk louder or softer.

All that to say: the +/- 3 dB spec means that the output level of any frequency over the stated range will vary no more than +/- 3 dB from whatever level it was at in the mix.
 
avaserfi

avaserfi

Audioholic Ninja
No speaker plays at X dB. The output SPL is dependent on many factors including the sensitivity of the speaker in combination with the amount of power you feed it and as always room acoustics.

The +/- 3 dB spec tells you the variation of the input from the output over the stated range. Music is a combination of many waveforms and thus many frequencies and each individual frequency was given a specific level relative to the others when it was mixed and mastered. Of course the level of any particular frequency changes too - it's not like every instance of 300 Hz will be at the same level all the time. Your voice has a fundamental frequency but you can talk louder or softer.

All that to say: the +/- 3 dB spec means that the output level of any frequency over the stated range will vary no more than +/- 3 dB from whatever level it was at in the mix.
Thanks for going into more detail than I originally did in my reply. You are also a bit more eloquent with your words, I guess I need a little more practice :).

But in practice, it can be said that some speakers have the ability to play more loudly with less distortion than others thus staying in their stated linearity range, correct?
 
M

MDS

Audioholic Spartan
But in practice, it can be said that some speakers have the ability to play more loudly with less distortion than others thus staying in their stated linearity range, correct?
Correct. You did a good job with the post, I just tried to clarify the point a bit.
 
mtrycrafts

mtrycrafts

Seriously, I have no life.
No speaker plays at X dB. The output SPL is dependent on many factors including the sensitivity of the speaker in combination with the amount of power you feed it and as always room acoustics.

The +/- 3 dB spec tells you the variation of the input from the output over the stated range. Music is a combination of many waveforms and thus many frequencies and each individual frequency was given a specific level relative to the others when it was mixed and mastered. Of course the level of any particular frequency changes too - it's not like every instance of 300 Hz will be at the same level all the time. Your voice has a fundamental frequency but you can talk louder or softer.

All that to say: the +/- 3 dB spec means that the output level of any frequency over the stated range will vary no more than +/- 3 dB from whatever level it was at in the mix.

When manufacturers rate their speakers, it is a spec in an anechoic chamber response at a constant signal level input over the frequency range. What is shows it is response head on. So, in such a published data, room is out of the equation..
When it is reviewed properly, it is measured as specified, 1m or 2m, and, gated so take the room response out of the equation. If not, it would be only good for the test room.
 
Tomorrow

Tomorrow

Audioholic Ninja
This was a very good primer on the matter. While it does not go into great detail and explanation, I'd like to see this, and similar posts "stickified" as an introduction into audio (and/or video) elements. Those of us that post a lot and have been here since dirt was invented throw around terms and ideas that I'm certain lose noobs attention and interest rapidly. This is valuable for that kind of casual reader.

You have a nice touch with the explanation, avaserfi. I think you ought to do them all! :D (Cables, subs, room treatments, etc. etc.) A little humor is always a good grabber to get the reader's attention, too. But there is meat on the bones of your post and I hope it gets read by the hundreds of people who surf in here looking for basic information, but never pop in a question.

Good show.
 
M

MDS

Audioholic Spartan
When manufacturers rate their speakers, it is a spec in an anechoic chamber response at a constant signal level input over the frequency range. What is shows it is response head on. So, in such a published data, room is out of the equation..
When it is reviewed properly, it is measured as specified, 1m or 2m, and, gated so take the room response out of the equation. If not, it would be only good for the test room.
Yes, all true but it misses my point. The point I was trying to make is that it doesn't have anything to do with absolute SPL from a speaker as that varies with many different factors.

When testing the frequency response the SPL vs frequency is plotted as in avaserfi's original post. To simplify it, let's say we test 100 Hz, 1000 Hz, and 10,000 Hz. We feed it each frequency one at a time at a standard level. If the 100 Hz tone yields 100 dB, the 1000 Hz tone yields 97 dB, and the 10,000 Hz tone yields 102 dB, then the range is -3/+2 dB. The output varies by at most -3 dB on the low side and +2 dB on the high side.

Now the same speaker in a normal room may not produce 102 dB as a max - it may only max out at 60 dB - but the relationship between the frequencies will be the same.
 
emorphien

emorphien

Audioholic General
Yeah the 20Hz to 20KHz +/- 3db has confused some people I know, but you could look at it and say one speaker may dip everywhere less than 10Khz but still be within the +/- 3db rating and another speaker may dip above 10Khz but again still be +/-3 db.

So it's really an incomplete picture of how the speaker will sound, you really need the FR plot at a minimum.
 
M

MDS

Audioholic Spartan
Yeah the 20Hz to 20KHz +/- 3db has confused some people I know, but you could look at it and say one speaker may dip everywhere less than 10Khz but still be within the +/- 3db rating and another speaker may dip above 10Khz but again still be +/-3 db.
Exactly. It only tells you the maximum variation from flat (no change). It does not tell you where in the frequency range the dip or boost occurs nor how many of the frequencies in the range are up 3 dB or down 3 dB.

This is why two speakers which appear to be roughly identical and have the same stated 20-20 kHz +/- 3 dB variation can and do sound different - the peaks and valleys occur at different points.
 
3db

3db

Audioholic Slumlord
although my room isn't an anechoic chamber

When manufacturers rate their speakers, it is a spec in an anechoic chamber response at a constant signal level input over the frequency range. What is shows it is response head on. So, in such a published data, room is out of the equation..
When it is reviewed properly, it is measured as specified, 1m or 2m, and, gated so take the room response out of the equation. If not, it would be only good for the test room.
I still find that most useful because it really demonstrates the linearity of a speaker and it etsbalishes a baseline from which to work with in subtly adjusting room acosutics to match the speakers output. That is if the manufacturer actually gave a plot instead of the "20Hz to 20 Khz one liner". I wish more speaker manufacturers would publish anechoic data
 
Swerd

Swerd

Audioholic Warlord
avaserfi

Thanks for writing that. It will be a really good introduction if it gets posted as a sticky somewhere.

I want to suggest several corrections or rewordings.

1) In your title, you used the phrase SPL vs. Frequency Graphs (Speaker Linearity). The linear response of a speaker is different than a flat SPL vs. frequency graph. A flat graph, as you explained rather well, is a speaker's ability to make evenly balanced sound across the audio spectrum. Linearity refers to a speaker responding to increasingly more powerful signals by generating louder tones in proportion to the original signals. Every speaker has an upper and lower limit to its linear response range. Speakers with a large linear response range have a greater ability to create the illusion of musicians sitting in your room instead of little tiny musicians sitting inside the speaker cabinets.

2) Just below the graph where you describe the X and Y axes, you say "The decibel rating refers to the loudness of the sound and is on a logarithmic scale." The decibel scale, although it appears to be linear on the graph, really isn’t because decibels are a logarithmic value, 20 dB is not twice as loud as 10 dB. You might want to add, that the smallest audibly detectable change in SPL is about 1 to 3 dB, and a change of +6 dB is about twice as loud, with -6 dB half as loud. You might also comment on how loud 70 or 80 dB is compared to normal speaking levels.

You could also comment some on frequencies. Human voices range from about 60 to 1,600 Hz, and not surprisingly, human hearing is most sensitive in this same range. The lowest musical notes made by any instrument (other than pipe organ) are in the mid to upper 30 Hz range. The far right key on the piano is just under 5,000 Hz. Above 5,000 Hz is all harmonic overtones. PSB has a good chart, showing the frequency ranges of various musical instruments. http://www.psbspeakers.com/audioTopics.php?fpId=8&page_num=2&start=8



3) Where you mention Room Acoustics, you might want to mention that room acoustics, in general, have the greatest effect on the bass frequencies below roughly 150 Hz. The walls, floor and ceiling of a room will generate reflections out of phase with the same tones coming directly from the speaker. This can create a pattern of amplifications and cancellations of these bass tones depending on the listener’s location, and will appear as peaks and valleys in a SPL vs. frequency curve. This is caused by room acoustics and not by the speaker.
 
Last edited:
avaserfi

avaserfi

Audioholic Ninja
Swerd, great response and thank you very much for the feedback! I wish I had actually thought more before making the original post, but it was more of a spur of the moment thing.

That chart is very useful, I was looking for something similar when I was writing up the introduction.

Now if only I could edit the original post to add more information :(.

Again thanks for your clarifications and additions!
 
3db

3db

Audioholic Slumlord
A question or two for you Swerd

avaserfi
3) Where you mention Room Acoustics, you might want to mention that room acoustics, in general, have the greatest effect on the bass frequencies below roughly 150 Hz. The walls, floor and ceiling of a room will reflect these tones, generating reflections that are out of phase with the same tones coming directly from the speaker. This can create a pattern of amplifications and cancellations of these bass tones depending on the listener’s location.
First off, I now know what linearity means thanks to you. :D

About room acoustics, my thinking is that it affects all frequencies, not primarily bass. I think this for several reasons.

First of, consider rooms that at the extremes of sounding dead or alive. I was of the impression that a live room had lots of reflections primarily in the mids to highs which muddied up the sound.

2ndly, I built a pair of acoustic panels to help absorb the reflection off the side walls relative from the speaker to the seating position. These panels were simply 2x4 ceiling tiles held in a frame, wrapped in cotton batten and teh whole thing covered in semi transparent cloth to make it pretty. According to the plans that I followed, it mentioned that absorption of frequency was between a 100Hz to 10KHz.

Thats been my experience with room acoustics. Maybe I'm way off in my thinking. :confused:
 
Swerd

Swerd

Audioholic Warlord
Now if only I could edit the original post to add more information :(.
Instead of editing the original post, you could post it again with all the changes and additions you want.
 
avaserfi

avaserfi

Audioholic Ninja
First off, I now know what linearity means thanks to you. :D

About room acoustics, my thinking is that it affects all frequencies, not primarily bass. I think this for several reasons.

First of, consider rooms that at the extremes of sounding dead or alive. I was of the impression that a live room had lots of reflections primarily in the mids to highs which muddied up the sound.
Acoustics do affect all frequencies but in different ways. With the high end you are correct the waves will reflect off walls and can cause a "harsh" sound or muddy the sound up, but this is not the same as what happens with lower frequencies. When one hears muddied dialog or a harsh sound this is from the same signal produced by a loud speaker hitting various reflection points then being reflected towards one ears at slightly different time. The brain has trouble separating all the information and it becomes muddied.

Swed mentions the low end where room nodes (specifically how sound waves interact with a given room) play more a role in completely canceling the waves out or making them double up on themselves.

The reason different frequencies respond differently to rooms is caused by wavelength.

Again if I made any errors someone please make me aware. I have been reading up much more on the technical side to this hobby but might have misunderstood something in the journals/books I have picked up.

Instead of editing the original post, you could post it again with all the changes and additions you want.
Depending how everything goes I might, but there is no sense in making a new thread or hiding the real point of this thread midway through. I probably will end up doing it either way though. Maybe I'll just ask a mod to move it once I have made the changes :).

Edit: I have sent a PM with a revised version to a Mod asking them to replace my original post. Thanks Swerd. Hopefully the new version is more informative and useful. I also asked this to be stickied if they think it useful enough as there has been a few mentions of that as well.
 
Last edited:
Swerd

Swerd

Audioholic Warlord
About room acoustics, my thinking is that it affects all frequencies, not primarily bass. I think this for several reasons.

First of, consider rooms that at the extremes of sounding dead or alive. I was of the impression that a live room had lots of reflections primarily in the mids to highs which muddied up the sound.

2ndly, I built a pair of acoustic panels to help absorb the reflection off the side walls relative from the speaker to the seating position. These panels were simply 2x4 ceiling tiles held in a frame, wrapped in cotton batten and teh whole thing covered in semi transparent cloth to make it pretty. According to the plans that I followed, it mentioned that absorption of frequency was between a 100Hz to 10KHz.

Thats been my experience with room acoustics. Maybe I'm way off in my thinking. :confused:
Everything you say is true, but for reasons that I don't understand the bass reflections and cancellations show up in a SPL vs. frequency curve as HUGE PEAKS and VALLEYS. These other room effects that you describe don't seem to have as large an impact on these graphs, even though we can easily hear them.

That's why speaker makers often don't show these curves. They don't have an anechoic room, and these curves made in real rooms often look terrible below 150 Hz. A poorman's substitute to an expensive anechoic room is to generate the curve outside.
 
newsletter

  • RBHsound.com
  • BlueJeansCable.com
  • SVS Sound Subwoofers
  • Experience the Martin Logan Montis
Top