• 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!

An Enticing Marketing Story, Theory Without Measurement?

Absolute

Major Contributor
Forum Donor
Joined
Feb 5, 2017
Messages
1,084
Likes
2,125
Look at the impulse tesponse.
I'll have to measure both to ensure it's comparable. The measurements I have on hand doesn't look comparable. How should I measure this, and would we expect a few spikes lower in level within the first milliseconds without the time alignment?
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden

mitchco

Addicted to Fun and Learning
Audio Company
Joined
May 24, 2016
Messages
640
Likes
2,397
Re: impulse display. Note that the impulse response display is heavily weighted towards high frequencies. The step response is the right display. See Stereophile article: https://www.stereophile.com/features/100/index.html

There is no agreement on what "room-compensated" means. So you need to be more specific. Most commercial algorthms these days appear to improve the low frequencies - for a single listener at least. That is about 30% of the factor weighting in sound quality ratings. Above the transition frequency it is the "Wild West". Some algorithms use broadband "tone control" adjustments (good) while others pride themselves on flattening even small irregularities (probably bad). Masses of people, including many who should know better, think we hear waveforms. Wrong. Humans are greatly insensitive to phase, meaning that focusing on time domain errors is a "fools errand" to quote an earlier poster., especially if it compromises the amplitude response.

Thanks for pointing this out Dr. Toole. The top DSP packages offer quite a bit of user interaction, so sometimes it is hard for folks to figure out that these packages do offer broadband "tone controls" above the rooms transition frequency, while the heavy lifting is done below Schroeder. It is the number one mistake I see (and have done myself) when using some of these powerful DSP software programs (i.e. full range correction).
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden

JoachimStrobel

Addicted to Fun and Learning
Forum Donor
Joined
Jul 27, 2019
Messages
518
Likes
303
Location
Germany
Timbre is the amplitude of the components of the harmonic series in relation to the fundamental tone.

As you "speak" you are constantly changing the amplitude of the components of the harmonics of the fundamentals in your vocal sound, as well as adding some transients - generally consonants, to the vowel sounds. You also add "noise" - the letters C, F, S, H, X, and SH combination usually make noise (no defined frequency).

A clarinet, a trumpet, a flute, a stringed instrument all produce different harmonic series even if playing the same "note". That's what I call "timbre" in the basic case.

Bass Guitar (handy home-grown example), one note played:

The fundamental frequency is marked "1", the harmonics measured are "2" though "9". There are more, but the software (REW) chooses not to count any higher.

The harmonic series is multiples of the fundamental frequency.

View attachment 35325

Timbre may include more information, the noise a violin bow makes scraping the string, the (often) enharmonic series of drums, cymbals...

Timbre is the characteristic "sound" something makes, defined by the amplitudes of the frequencies (or noise) that accompanies the fundamental (if there even is one).

A pure tone - sine wave - is rarely encountered. That's a Fundamental with no harmonics. Its timbre is identified by the lack of harmonics. The closest I've come to producing one occurred when blowing across the top of a beer bottle:

View attachment 35324

If the reproduction system does not reproduce the recorded harmonic series correctly, the timbre of the sound will be changed.

I'd say the reproduction of Timbre is pretty robust - I can't remember mistaking a flute for a trumpet, on any system.
Are these graphs on-axes or off-axes recordings? Could you show the off-axes decay of these instruments? Would be interesting if they are smooth or have holes - natural ones of course.
 

mitchco

Addicted to Fun and Learning
Audio Company
Joined
May 24, 2016
Messages
640
Likes
2,397
There is no right or wrong. Depends what you want to show.

If you want to show a speaker's timing response, the step response is the right measure. The issue with the impulse response to show a speakers timing is that the woofer amplitude will be 1/100th the amplitude of the tweeter and the mid range (assuming a 3 way) will be 1/10th the amplitude of the tweeter. Therefore the woofers and midrange response is buried underneath the tweeters amplitude response when viewing a loudspeaker systems timing response when showing an impulse response. Therefore showing nothing of interest. What's interesting is that you linked to the same Stereophile article I did :)
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden
If you want to show a speaker's timing response, the step response is the right measure. The issue with the impulse response to show a speakers timing is that the woofer amplitude will be 1/100th the amplitude of the tweeter and the mid range (assuming a 3 way) will be 1/10th the amplitude of the tweeter. Therefore the woofers and midrange response is buried underneath the tweeters amplitude response when viewing a loudspeaker systems timing response when showing an impulse response. Therefore showing nothing of interest. What's interesting is that you linked to the same Stereophile article I did :)

Sure but still there is no right or wrong. You need to interpret each measure based on what you are measuring.
 

Absolute

Major Contributor
Forum Donor
Joined
Feb 5, 2017
Messages
1,084
Likes
2,125
Well, if impulse response is useless for looking at the effects of timing, why look for it there?
 

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,717
Likes
2,897
Location
Finland
Just one q. What do you mean that drivers are several ms from each other?

Your question above!

The answer is in step response! You can see each drivers's peak in step response and thus see the timing difference. Typical 3-way with symmetric crossovers mid follows tweeter in 0.2-0.4ms and woofer +2ms. Timing difference comes from xo frequencies used and if slopes are LR2 or LR4. Another way to look at this is group delay.

Hearing is not very sensitive to this kind of timing difference if amplitude response is smooth, which means that timings are matched properly! But still there is some anecdotal legend about what order set at which frequency sounds best! With loudspeakers you shouldn't look at only one measurement, on-axis and off-axis spl, distortion profile, step response, phase curve etc. all should be in harmony!
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden
Your question above!

The answer is in step response! You can see each drivers's peak in step response and thus see the timing difference. Typical 3-way with symmetric crossovers mid follows tweeter in 0.2-0.4ms and woofer +2ms. Timing difference comes from xo frequencies used and if slopes are LR2 or LR4. Another way to look at this is group delay.

Hearing is not very sensitive to this kind of timing difference if amplitude response is smooth, which means that timings are matched properly! But still there is some anecdotal legend about what order set at which frequency sounds best! With loudspeakers you shouldn't look at only one measurement, on-axis and off-axis spl, distortion profile, step response, phase curve etc. all should be in harmony!

So how does the different curves individually look !like at time-aligned step respnse vs a traditional crossover and what is the relatve time difference p-p?
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden
Well, if impulse response is useless for looking at the effects of timing, why look for it there?

Similar to my question above what is the time difference for the start of the tweeter and woofer response in a step response in a traditional crossover (not p-p)?
 

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,717
Likes
2,897
Location
Finland
https://www.audiosciencereview.com/forum/index.php?threads/step-response-does-it-really-matter.1999/

https://www.audiosciencereview.com/forum/index.php?threads/measuring-time-delay.7644/#post-179152

https://en.wikipedia.org/wiki/Step_response

http://audiojudgement.com/loudspeaker-step-response-measurement/

https://bksv.com/media/doc/17-198.pdf

etc. There is no comprehensive and self-explanatory limit or shape for a "good" step response. It all depends... Some people trust in one-knee step and call it time-coherent (full range speaker, 1st order 2-way, manipulated with FIR)
 

UliBru

Active Member
Technical Expert
Joined
Jul 10, 2019
Messages
123
Likes
337
There is no comprehensive and self-explanatory limit or shape for a "good" step response.
There is.
Take a 2nd or 4th order minimumphase highpass at e.g. 20 kHz and calculate its step response. This is the "ideal" shape.
You can play of course with the corner frequency (e.g. a speaker may play above 40 Hz only) but anyway the step response of the highpass shows up the best possible time behaviour.

Whereas typical higher order minimumphase speaker crossovers (passive and active) create in sum an excessphase "error" known as tweeter first, then midrange and finally bass. This can be corrected by FIR filters and it does not need an echo chamber for this. The price to pay is a delay in the playback.

As quite often the crossovers also create deviations in the frequency response a common trick is to change the polarity of a driver to get a better result. The underlying assumption: it's only the FR which counts.
At the end you get the typical step response display and typical comments in speaker reviews like
The tweeter and woofer are connected in positive acoustic polarity, the midrange unit in negative polarity, and the decay of each unit's step smoothly blends with that of the next lower in frequency, suggesting optimal crossover design
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
Hearing is not very sensitive to this kind of timing difference...
But the question that prompted this part of the discussion was saying that if human hearing distinguishes between source and reflections on the basis of delays of the order of >1ms, why is it acceptable for the speaker itself to be breaking a sound into separate steps of that order.

The answer, I would suggest, is that it isn't acceptable. But maybe what saves the situation is that the steps are split into different frequency bands and so don't directly resemble each other, therefore the ear/brain doesn't register them as reflection-like. Maybe the effect varies with content, too, in a very complex way.
 

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,717
Likes
2,897
Location
Finland
^But maybe what saves the situation is that the steps are split into different frequency bands and so don't directly resemble each other, therefore the ear/brain doesn't register them as reflection-like.

Yes, that the common expalanation, but some people (like me) say that they can hear a difference... Harmonics having different onset (transient) than basic note might be the explanation, when xo happens between them. Sensitivity of hearing is highest around 1-4kHz. Reflections have multiple random delay and are perceived differently.
 
Last edited:

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden
Again, are you looking at the p-p difference of woofer-tweeter between a time-aligned vs traditional crossover? You need to compare intertime difference whether you use p-p or start of the response.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
^But maybe what saves the situation is that the steps are split into different frequency bands and so don't directly resemble each other, therefore the ear/brain doesn't register them as reflection-like.

Yes, that the common expalanation, but some people (like me) say that they can hear a difference... Harmonics having different onset (transient) than basic note might be the explanation, when xo happens between them. Sensistivity is highest around 1-4kHz. Reflections have multiple random delay and are perceived differently.
I am prepared to believe that although we may not hear much difference consciously, it's certainly adding to confusion in the brain, fatigue, etc. And maybe if we trained ourselves to listen for it, we could hear it consciously. (I'm pretty sure I wouldn't be able to hear any difference between two recordings of someone speaking an African click language even if occasionally substituting different words! But if I learned the language, I would suddenly be able to. But it wouldn't be my hearing that had changed). So I am a DSP phase and time alignment person even if just to be on the safe side. Why mess with mechanisms that have evolved for a reason?
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden

UliBru

Active Member
Technical Expert
Joined
Jul 10, 2019
Messages
123
Likes
337
The individual woofer and tweeter step response of a time-aligned 2-way system (here 2nd order, both drivers positive polarity) must look like this

2-way-ideal.png


The blue curve is the resulting perfect sum step response.
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden
The individual woofer and tweeter step response of a time-aligned 2-way system (here 2nd order, both drivers positive polarity) must look like this

View attachment 35419

The blue curve is the resulting perfect sum step response.

The scale is a bit difficult too se the intertime difference p-p, is it possible to give an estimate?
 
Top Bottom