Making Musical Instruments Sound Natural

This blog post focuses on an important aspect to speaker measurement...Timbrel Balance.   

Timbrel Balance Definition: the characteristic quality of sound that distinguishes one voice or musical instrument from another or one vowel sound from another: it is determined by the harmonics of the sound and is distinguished from the intensity and pitch

Intensity and pitch in speaker measurement is simply the overall perceived frequency response of the speakers in your listening room.  It is a combination of the direct and reflected sound projected from the speakers.  A speaker designer will strive for a flat on-axis frequency response.  Attention is paid to the off-axis response however what is the ideal off-axis response?  Would the ideal speaker be omnidirectional?  Should it have a gradual narrowing at high frequency?  Should the directionality remain the same across the frequency spectrum?  What off-axis response presents the most authentic presentation in terms of maintaining the true timbrel balance of speech and instruments?  

 To answer this we need to understand what Power Response is.  Rather than try to explain it myself I think the folks at Audioholics have done a great job with their definition...  "Power Response is the sum of the total radiated acoustic output of a loudspeaker as measured in a sphere around the speaker at several incremental intervals  on- and off-axis in the far (reverberant) field.  This measurement essentially captures the total sound emitted by a loudspeaker at all frequencies, in all directions, and is therefore thought by proponents of this measurement method to be more representative of how a speaker will sound in an acoustically well-balanced listening environment than by what can be inferred from a simple on-axis anechoic frequency response measurement."

Conducting a power response of the entire output of a speaker is difficult to measure and so it's sometimes useful to simply look at only the measurements in front of the speaker within a certain horizontal listening window such as 0-60 degrees off-axis.  This will provide a very good picture of the perceived subjective frequency balance of the speaker since 50% of what we hear are the reflections within the room.  Moving forward I will refer to this type of measurement as the "average response".  

To demonstrate this I've measured the 'average response' of these two speakers...

Speaker #1 shown below...

Scanspeak 10F/8424G02 mounted in narrow baffle speaker.

Speaker #2 shown below...

Large format front horn using Qty:4 Scanspeak 10F8424G02 drivers. 

Speaker #1 Tested

I conducted off-axis frequency response measurements from 0-60 degrees off-axis in 5 degree increments.  Below is the resulting overlays of the various responses.  I used my MiniDSP to flatten the on-axis using PEQ.  

For interests sake this is what the off-axis coloured polar map looks like for speaker #1. 

Below is the is the 'average response' of all my off-axis measurements combined.  

Below is a comparison between the on-axis frequency response (red) and the averaged response (black) of the listening window.  As you can see the averaged response is shelved down by about -5dB starting at 5kHz.  

Speaker #2 Tested 

Below are the off-axis measurements for speaker #2. 

And again for interest's sake here is the off-axis coloured polar map for speaker #2. 

Below is the averaged response within the listening window. 

Below shows the comparison between the averaged response (black) and the on-axis response (red).  

Conclusion

It was shown that the averaged response of the non-horn speaker differs significantly to its on-axis response.  This is something that cannot be corrected unless you fundamentally redesign the speaker.  The timbrel balance will be altered as a result of the mismatched responses.  

Speaker design must not only scrutinize the on-axis frequency response.  They must also look at the averaged response across a wide listening window since this represents a very important aspect to how the speaker will sound.  

This test was conducted with the Scanspeak 10F/8424G02 which has an effective radiating diameter of 76mm.  However if we look at a regular bookshelf speaker with a 25mm dome tweeter the overall polar map looks very similar...

This is just another example of how a properly designed horn or waveguide improve the overal percieved sound quality.  There are other aspect such as distortion than can be looked at as well.  This concludes this test however for interests sake let's look at the distortion sweeps for Speakers #1 and #2. 

 

Speaker #1 Distortion sweep at normal listening level shows around %0.80 THD.

Speaker #2 Distortion sweep at the same listening level...

Distortion is drops to %0.10 through the midrange. (an 80% reduction in distortion)