Impulse response = frequency response

[mixsit]

https://www.stereophile.com/content/dcm-time-frame-tf1000-loudspeaker
I came across this today and am curious about what's being stated about impulse and frequency response in their first five paragraphs here.
What's said initially seems plausible, but then they seem to drift off point with the '3-way/listening position conclusions.

I really missed The Audio Critic when they went away. I've never been comfortable with the High End audio scene and Peter was one of the good guys in my opinion :>)

 

[waynel]

didn't read the article but yes one can get the frequency response (both amplitude and phase) from the time domain impulse function via Fourier transform https://en.wikipedia.org/wiki/Fourier_transform. The impulse response in time and frequency response (amplitude and phase) are two representations of the same thing, the transfer function (for a memoryless time invariant system).

 

[Vini darko]

Hi I think I have a big dummys grasp of what they mean. @solderdude does impulse measurments on his reviews at diy audio heaven. Frequency response can be infered from the decay after the initial pulse happens. I still don't have anywhere close enough brains to work it out for myself though.

 

[waynel]

Hi I think I have a big dummys grasp of what they mean. @solderdude does impulse measurments on his reviews at diy audio heaven. Frequency response can be infered from the decay after the initial pulse happens. I still don't have anywhere close enough brains to work it out for myself though.

Ok, I’ll put it simply
Impulse response and frequency response contain the same information.

For details see my earlier post

 

[Vini darko]

Ok, I’ll put it simply
Impulse response and frequency response contain the same information.

For details see my earlier post

Thanks I'll read the article. No promises on understanding it though

 

[waynel]

Thanks I'll read the article. No promises on understanding it though

OK I skimmed it:

" a speaker's impulse response can be translated directly into its frequency response: all the information needed to show how its response varies with frequency is contained in the shape of the impulse it produces. "

True, Caveats: 1) on axis for both 2) anechoic for both 3)so long as the speaker doesn't change with time like getting hot, but close enough so I'll say its true.

"If a loudspeaker produces an output pulse that is an exact replica of an input pulse, then by definition it must have a perfectly flat frequency response."
True
"In practice, as to do so would necessitate the speaker being able to reproduce the DC component of the pulse—something that only a fan can do, DC implying a constant-velocity stream of air—some modification of the pulse shape is inevitable, equalizing the areas above and below the time axis. But in general with a minimum-phase system, the less the pulse shape is changed, the flatter the frequency response. "

True but omits that the high frequencies also have to go on for ever too.

If the impulse response is measured with the mic on axis in an anechoic chamber and is perfectly reproduced (within a reasonable bandwidth limitation) then the ON-AXIS frequency response is flat in amplitude and linear in phase. This is good but doesn't tell you what the frequency response will sound like in a room with both direct and reflected sound. John Atkinson is making a mistake of jumping from a single point anechoic measurement to in room tonal balance. Need a spinorama to model that.

Wayne

 

[mixsit]

OK I skimmed it:

"If a loudspeaker produces an output pulse that is an exact replica of an input pulse, then by definition it must have a perfectly flat frequency response."
True
"In practice, as to do so would necessitate the speaker being able to reproduce the DC component of the pulse—something that only a fan can do, DC implying a constant-velocity stream of air—some modification of the pulse shape is inevitable, equalizing the areas above and below the time axis. But in general with a minimum-phase system, the less the pulse shape is changed, the flatter the frequency response. "

True but omits that the high frequencies also have to go on for ever too.

If the impulse response is measured with the mic on axis in an anechoic chamber and is perfectly reproduced (within a reasonable bandwidth limitation) then the ON-AXIS frequency response is flat in amplitude and linear in phase. This is good but doesn't tell you what the frequency response will sound like in a room with both direct and reflected sound. John Atkinson is making a mistake of jumping from a single point anechoic measurement to in room tonal balance. Need a spinorama to model that.

Wayne

Thank you I much appreciate it.
'mixsit is Wayne here as well :>)

 

[mansr]

"In practice, as to do so would necessitate the speaker being able to reproduce the DC component of the pulse—something that only a fan can do, DC implying a constant-velocity stream of air—some modification of the pulse shape is inevitable, equalizing the areas above and below the time axis. But in general with a minimum-phase system, the less the pulse shape is changed, the flatter the frequency response. "

True but omits that the high frequencies also have to go on for ever.

Not true. A speaker produces (primarily) pressure waves, not airflow. The air velocity at a given point is proportional to the derivative of the pressure. A positive DC signal corresponds to a constant elevated pressure. Its derivative is zero, so there is no motion. A very large non-ported speaker in a sealed room could create this. This is obviously not practical, and any real speaker will have a lower limit to its useful frequency response.

Back to the thread title, the frequency response is simply the Fourier transform of the impulse response. As someone already said, either one fully describes a linear time-invariant (LTI) system. Sometimes one representation is more convenient than the other. Both contain exactly the same information. However, frequency response plots often show only the amplitude. To be complete, the phase is also required.

 

[waynel]

Not true. A speaker produces (primarily) pressure waves, not airflow. The air velocity at a given point is proportional to the derivative of the pressure. A positive DC signal corresponds to a constant elevated pressure. Its derivative is zero, so there is no motion. A very large non-ported speaker in a sealed room could create this. This is obviously not practical, and any real speaker will have a lower limit to its useful frequency response.

The poetry about a fan was John Atkinson's not mine. I read that as a method of increasing the air pressure in a leaky room but yeah its imprecise. Anyway, I think he was trying to make the point that in order to pass a perfect impulse function the speaker bandwidth would have to extend down to DC (and up to infinite frequency as well which he omitted and I mentioned above) therefore I rate the statement as overall true.

As someone already said, either one fully describes a linear time-invariant (LTI) system.

Me

 

[scott wurcer]

Anyway, I think he was trying to make the point that in order to pass a perfect impulse function the speaker bandwidth would have to extend down to DC

Yes the voice coil would have to be superconducting and the amplifier would need exactly 0 output impedance as well as having a leak proof room.. It is worth noting some speakers are in the strictest sense not time invariant or minimum phase.

 

[waynel]

Yes the voice coil would have to be superconducting and the amplifier would need exactly 0 output impedance as well as having a leak proof room..

A fan could be used to increase pressure in a leaky room; I see this everytime I take my kids to a bouncy house. From the hours I spent there, I could tell that the fan-based pressure increase inside the bouncy house was at a very low frequency. I did not arrive early enough nor stay long enough to see if it was DC

 

[mansr]

The poetry about a fan was John Atkinson's not mine. I read that as a method of increasing the air pressure in a leaky room but yeah its imprecise. Anyway, I think he was trying to make the point that in order to pass a perfect impulse function the speaker bandwidth would have to extend down to DC (and up to infinite frequency as well which he omitted and I mentioned above) therefore I rate the statement as overall true.

I get that, but the flawed analogy with the fan is indicative of John Atkinson once again having fundamentally misunderstood the things he is talking about.

 

[Vini darko]

All this talk of fans makes me think of rotary subwoofers

 

[Blumlein 88]

In case it isn't clear, you can get the frequency response from impulses being measured. Some measurement software does this. Also you can get the impulse response from a frequency sweep.

You often read of audiophiles talking transient or impulse response as if it were separate from frequency response.

Here is a post by Ray Dunzl where he calculated the impulse response from a sweep and measured the response to a 1 bit Dirac impulse. They are the same.
https://www.audiosciencereview.com/forum/index.php?threads/impulse-response.1765/#post-44352

And another variation using sweep and step response to a a 10 hz square wave.
https://www.audiosciencereview.com/forum/index.php?threads/impulse-response.1765/post-44440

So Ray shows by actually measuring his speakers that all this math mumbo jumbo really works.

 

[Dmitrij_S]

Not true. A speaker produces (primarily) pressure waves, not airflow. The air velocity at a given point is proportional to the derivative of the pressure. A positive DC signal corresponds to a constant elevated pressure. Its derivative is zero, so there is no motion. A very large non-ported speaker in a sealed room could create this. This is obviously not practical, and any real speaker will have a lower limit to its useful frequency response.

A small clarification. Under free field radiation condition the diaphragm of the speaker creates air flow (i.e. displaces air particles), which is converted to air wave pressure trough acoustic impedance of the medium. But not reverse as follows from your post (you stated that the diaphragm creates pressure but not airflow which is probably valid for closed volumes but actually wrong for free field radiation conditions). So, Atkinsons analogy is correct when we consider possibility to create DC pressure in the free field, not in the closed volume.

 

[mansr]

A small clarification. Under free field radiation condition the diaphragm of the speaker creates air flow (i.e. displaces air particles), which is converted to air wave pressure trough acoustic impedance of the medium. But not reverse as follows from your post (you stated that the diaphragm creates pressure but not airflow which is probably valid for closed volumes but actually wrong for free field radiation conditions). So, Atkinsons analogy is correct when we consider possibility to create DC pressure in the free field, not in the closed volume.

Close to the diaphragm, things are obviously going to be a bit different. Likewise for an antenna radiating electromagnetic waves.

A fan in an open space creates a pressure differential between the front and rear. The overall pressure in a larger surrounding volume remains unchanged. A fan blowing into a leaky container will increase the pressure to until the in and out airflows are equal. While in-wall speakers do exist, most of them are free-standing ported designs, so the analogy still doesn't work.

The flawed thinking by JA is made apparent by the phrase "DC implying a constant-velocity stream of air." Sound is pressure, not motion.

 

[Ville]

This fan discussion is really nice read to avoid working. And a reminder why it is so easy to get to single digits in car with a subwoofer.

Here is a post by Ray Dunzl where he calculated the impulse response from a sweep and measured the response to a 1 bit Dirac impulse. They are the same.

As a pre-emptive measure before someone trips on the "Dirac impulse", it has nothing to do with the room-correction software company:
https://en.wikipedia.org/wiki/Dirac_delta_function

At work, I'd be wise to eat a fruit, work on magnetic field measurements, and perhaps grab another fruit.
But instead, I play with my Apple, and dream about a Tesla and Orange guitar amps.

 

[RayDunzl]

"Impulse response = frequency response"

Mathematically true, I suppose.

Not obvious to the All Powerful Ear of the Beholder.

---

REW tests using a swept sine.

It's intuitive that you look at the level of that one after being passed through the tested gear at each frequency and voila - basic frequency response.

But you click another button and that same data - what REW heard with the swept tone - can be displayed as Impulse Response via some fancy mathematical transformation.

As can Step Response, Phase, and other goodies.

 

[RayDunzl]

Not obvious to the All Powerful Ear of the Beholder

I had to prove it to myself:

https://www.audiosciencereview.com/forum/index.php?threads/impulse-response.1765/#post-44352

https://www.audiosciencereview.com/forum/index.php?threads/impulse-response.1765/#post-44440