Timbre matching done with a microphone?

[mustang guy]

The question of which center would be a voice or timbre match to mains comes up a lot. I have been doing some reading on the topic and I feel there may be a way to do this scientifically rather than subjectively, not that opinions are in any way irrelevant. Using Audacity or a similar sound analyzing program, a sound can be broken down into individual frequencies with amplitude on undertones and overtones. This is a BIG question: Are two speakers playing the same exact note and having matching waveform/frequency plot timbre matches for that particular note?

As I understand it, pertinent notes for voice matching tend to come from higher frequency notes. If this is so, then why we plot notes in octave increments starting at 100 Hz or so for a particular speaker using the exact same sound file from the exact same amplifier and the exact axis, distance, amplitude, etc? Then we could compare speakers by overlaying the plots. Note: I feel it is important to have as high a starting point as possible because the center speaker is usually going to be less capable in low frequencies than the mains.

It would be interesting to SEE the timbre of one speaker compared to another.

The plot above is actually the same note played on the same speaker with a varying fundamental to overtone. I am only using this as an example. Each column represents the same note with varying undertone and overtone adjustments. I wanted you all to see an example of what I meant. In my experiment, each column would be a different frequency for a speaker.

I may be oversimplifying this. If timbre is much more complex than waveform/frequency then how can we graph that so we can get that overlay? If the overlay is extremely complex, can we use a computer program to compare the plots instead of our eyes?

I do not want to reinvent a wheel. If there is already a way to measurably compare timbre, I would love to know about it. Empirical evidence that say an Academy is a great center LaScalas would be more influential to me than the subjective opinions of many individuals.

Some of you are going to beat me up a bit on this. Please be kind.

Edited October 31, 2013 by mustang guy

[ZAKO]

HHHHmmm,,,interesting plot,,,fundamentals of most musical instruments are below 5K and their harmonics are above 5K ,,,,,I WISH THAT IMAGE WAS LARGER,,,,,

Edited October 31, 2013 by ZAKO

[derrickdj1]

This sounds similar to voice recongnition. Very interesting.

[Chris A]

...This is a BIG question: Are two speakers playing the same exact note and having matching waveform/frequency plot timbre matches for that particular note?

As I understand it, pertinent notes for voice matching tend to come from higher frequency notes. If this is so, then why we plot notes in octave increments starting at 100 Hz or so for a particular speaker using the exact same sound file from the exact same amplifier and the exact axis, distance, amplitude, etc? Then we could compare speakers by overlaying the plots. Note: I feel it is important to have as high a starting point as possible because the center speaker is usually going to be less capable in low frequencies than the mains.

It would be interesting to SEE the timbre of one speaker compared to another.

I may be oversimplifying this. If timbre is much more complex than waveform/frequency then how can we graph that so we can get that overlay? If the overlay is extremely complex, can we use a computer program to compare the plots instead of our eyes?

I do not want to reinvent a wheel. If there is already a way to measurably compare timbre, I would love to know about it. Empirical evidence that say an Academy is a great center LaScalas would be more influential to me than the subjective opinions of many individuals.

Some of you are going to beat me up a bit on this. Please be kind.

To me, the most revealing plots are waterfall impulse response plots for the whole speaker in an anechoic chamber or one tested outside on a clear field without nearby sound reflectors. You basically get the most information that you probably can get in one plot.

However, as you pointed out, it really isn't that simple. For instance, you would like to know if there are cabinet reflections/refractions and significant amounts of SPL being sprayed into the room at different angles from the front of the speaker's cabinet that significantly affect the timbre.

I got probably the best lesson of all in the anechoic chamber in Hope when Roy demonstrated just two different horns on top of a Jubilee bass bin (EQed flat, etc.). The energy coming off the K-510 horn at frequencies lower than ~2 KHz in the vertical axis, and well as in the horizontal axis in below ~800 Hz really affected the timbre of the speaker in the chamber, which eats most of that stray energy, but you could walk around the speaker while it played music, and could hear much different timbres off-axis, while doing the same experiment with the K-402 produced a much different result, due to the the latter horn's ability to control it's off-axis energy all the way down to the crossover point. Once you walked around the speaker until you just couldn't see the inside of the K-402 horn, the sound basically shut down about 10-20 dB, and sounded very muted and a bit bass-heavy, due to the wider polar response of the bass bin.

What got really interesting is when we pulled the same speakers one at a time into the auditioning room right next to the chamber: the K-402 configuration Jubilee sounded about the same but with better lf response, while the K-510 sounded like a completely different animal since it was pumping a significant amount of band-passed energy (mostly midrange) into the nearby walls, ceiling and floor of the speaker, just like all other speakers do, including all other Klipsch speakers.

Remember the discussion of the precedence effect and Haas effect in the Corner-Horn Imaging FAQ? Not all sound delays are created equal by the human ear. Roy worked on getting the "salt and pepper" EQing for the K-510 down while we listened to the test CD through that speaker configuration, and he really was never able to make the K-510 Jubilee sound like the K-402 Jubilee. This was the real ear opener. I'll never forget that lesson and how much I learned that day.

Bottom line for me: you really need a lot more than a two-dimensional waterfall plot to "see timbre".

Chris

Edited October 31, 2013 by Cask05

[Schu]

I have prefect pitch... comes from playing acoustic guitar for over 40 years. I have never had a single issue between my ref and heritage were a NOTE was at a different pitch between the two speakers types.

when I used MCACC to get a rough set up of my speakers, I was very surprised at the sonic differences between the her. and ref., it was extremely evident.

since EQ'ing the speakers, I can still hear differences, but it's not so significant that I think that it is a mistake to run Ref. center and Her. fronts.

of course this is NOT critical listening, but in the HT realm.

Edited October 31, 2013 by Schu

[mustang guy]

HHHHmmm,,,interesting plot,,,fundamentals of most musical instruments are below 5K and their harmonics are above 5K ,,,,,I WISH THAT IMAGE WAS LARGER,,,,,

The article was here: http://ep-forum.com/smf/index.php?topic=3264.0 (timbre adjustment of a Fender Rhodes electronic piano), and there is a larger version of the image there. The accompanying article was posted here: http://ep-forum.com/smf/index.php?topic=3263. Since this plot was of a synthesized piano, I suppose the harmonics would be different than an acoustical piano.

This sounds similar to voice recognition. Very interesting.

You may be on to something. Voice and timbre are interchangeable. Why not use voice recognition software to match speaker timbre? I will go look for some public domain voice recognition software.

To me, the most revealing plots are waterfall impulse response plots for the whole speaker in an anechoic chamber or one tested outside on a clear field without nearby sound reflectors. You basically get the most information that you probably can get in one plot.

However, as you pointed out, it really isn't that simple. For instance, you would like to know if there are cabinet reflections/refractions and significant amounts of SPL being sprayed into the room at different angles from the front of the speaker's cabinet that significantly affect the timbre.

I got probably the best lesson of all in the anechoic chamber in Hope when Roy demonstrated just two different horns on top of a Jubilee bass bin (EQed flat, etc.). The energy coming off the K-510 horn at frequencies lower than ~2 KHz in the vertical axis, and well as in the horizontal axis in below ~800 Hz really affected the timbre of the speaker in the chamber, which eats most of that stray energy, but you could walk around the speaker while it played music, and could hear much different timbres off-axis, while doing the same experiment with the K-402 produced a much different result, due to the the latter horn's ability to control it's off-axis energy all the way down to the crossover point. Once you walked around the speaker until you just couldn't see the inside of the K-402 horn, the sound basically shut down about 10-20 dB, and sounded very muted and a bit bass-heavy, due to the wider polar response of the bass bin.

What got really interesting is when we pulled the same speakers one at a time into the auditioning room right next to the chamber: the K-402 configuration Jubilee sounded about the same but with better lf response, while the K-510 sounded like a completely different animal since it was pumping a significant amount of band-passed energy (mostly midrange) into the nearby walls, ceiling and floor of the speaker, just like all other speakers do, including all other Klipsch speakers.

Remember the discussion of the precedence effect and Haas effect in the Corner-Horn Imaging FAQ? Not all sound delays are created equal by the human ear. Roy worked on getting the "salt and pepper" EQing for the K-510 down while we listened to the test CD through that speaker configuration, and he really was never able to make the K-510 Jubilee sound like the K-402 Jubilee. This was the real ear opener. I'll never forget that lesson and how much I learned that day.

Bottom line for me: you really need a lot more than a two-dimensional waterfall plot to "see timbre".

You've just added a new dimension to the problem. Off axis timbre. Even if two horns have identical timbre on-axis, moving off-axis would change that. Only one person gets to sit in the perfect seat in a movie. Everyone else will be off-axis by almost every speaker. How in the blazes could a horn that is not identical have the same off-axis timbre?

I have prefect pitch... comes from playing acoustic guitar for over 40 years. I have never had a single issue between my ref and heritage were a NOTE was at a different pitch between the two speakers types.

when I used MCACC to get a rough set up of my speakers, I was very surprised at the sonic differences between the her. and ref., it was extremely evident.

since EQ'ing the speakers, I can still hear differences, but it's not so significant that I think that it is a mistake to run Ref. center and Her. fronts.

of course this is NOT critical listening, but in the HT realm.

It sounds as if MCACC tries to timbre match speakers. Perhaps it is only trying to get them all to play at the "reference level". Some speakers are more capable than others, and pardon the dig, but MCACC might blow a fuse if it tried to EQ a Bose. You are right, HT listening is all about timbre matching the front 3 speakers so the moving sounds seem more realistic. Quite frankly, a timbre mismatch doesn't deter from a good movie with good visuals. You will kick me, but I have a Bose center with a pair of Heresy's at my lake house. I needed something small, and it was in a box in a closet, so I grabbed it. It sounds awful, but I don't get up there much, so I haven't had a chance to replace it.

[mustang guy]

This sounds similar to voice recognition. Very interesting.

The only stuff I could find on voice recognition biometrics was Batvox. I don't know what it would cost, but as they say if you have to ask it's too much. All the other voice recognition software was concerned with speak recognition and translation. Voice biometrics technology would be the best application for timbre matching speakers.

[Islander]

A big factor in speaker sound and timbre is the room it's in. Most rooms are not perfectly symmetrical, nor is the size and placement of the furniture, so even closely-matched speakers usually vary a bit once they're in position.

When I sit in the listening position and run a test tone all around the speakers, the timbre shifts are noticeable, but I don't worry about it.

[mustang guy]

A big factor in speaker sound and timbre is the room it's in. Most rooms are not perfectly symmetrical, nor is the size and placement of the furniture, so even closely-matched speakers usually vary a bit once they're in position.

When I sit in the listening position and run a test tone all around the speakers, the timbre shifts are noticeable, but I don't worry about it.

You are correct. However, voice mismatched speakers make it that much worse. Timbre matching is still as relevant as speaker placement and acoustical treatment.

I'm the guy that actually has a Bose center to two heresy's. Now that is Heresy! BTW, it sounds like crap. I need to take a picture of it next time I'm over so you all can make proper fun of me.