Best Practice - Choosing the right speakers
I consider myself extremely lucky that my first years in the industry from 1990 to 1996 were at Graham’s Hi-Fi in North London. Whilst working with that fantastic bunch of people, I was schooled daily in the nuance of great sound. One of the things, however, that I never really understood at the time was how one system could work so well in one room, yet sound poor in another. I now know that this is all about the speaker/room/listener interface and interaction. We have come a long way since 1990 with digital EQ now being commonplace, and a far wider range of domestically acceptable acoustic solutions becoming available. These allow us to create an environment where a loudspeaker can work optimally in-room and potentially have the sonic icing on the cake provided by digital EQ.
Back in 1990, the subjectivist vs objectivist argument was raging. Subjectivists argued that so long as a speaker sounded good, why do measurements matter? Objectivists argued that measurements are the only scientific way to predict how a speaker sounds. Well, now we know that they were both right! The subjectivist argument falls over because a speaker that may sound incredible in one room with a certain listening position, put it in a different room may sound very different. The reason for this is that a speaker does not beam sound just to you at the listening position. Depending on frequency, a speaker radiates sound differently as you move away from the on-axis (the speaker is pointing straight at you and you are level with the speaker’s acoustic axis) sound. The total sound arriving at your ears is a combination of direct (straight from the speaker without reflecting off anything) and reflected (off one of more surfaces, with the reflected sound arriving at your ears after the direct sound). Sounds arriving at your ears within around 30ms after the direct sound are called early reflections.
The combination of on and off axis response is probably the single most important aspect of a speaker’s performance and plays a huge part in defining how it sounds in real rooms. Because the off-axis response is being affected by the room, this explains why a speaker can sound so different when used in different rooms.
Different manufacturers have different philosophies. These are amplified by marketing departments who focus on a specific aspect of a speaker’s design as being ‘unique’ or particularly special. What I have learnt in the past 30 years is that no particular design philosophy, materials selection or driver configuration is necessarily ‘better’ or ‘worse’ than any other. Every speaker design is a balance of compromises. The trick is to choose a speaker based on the specific requirements of a system and room.
To go into detail about selecting speakers would take a 1000 page book. Here, however, is an outline of the most important things to consider.
How big can the speaker be?
When it comes to low frequencies (anything below around 150Hz), there is no substitute for size. Some of my favourite speakers ever have been small 2-way designs. Amazing for certain types of music, rubbish for dub techno played at club volume levels.
The reality of real rooms means that it is often difficult to accommodate speakers especially if they are deep. Whilst hiding them behind an acoustically transparent screen is easy, if the speaker is 400mm deep, that extra space might be a tough sell to a client in an already small room.
How loud should the speaker play?
Ah, the old ‘reference level’ chestnut. Reference level is not 105db as mostly discussed, it is 85db with 20db headroom. What difference does that make I hear you ask? Well, few of us listen to pink noise, m-noise or sine waves continuously at 105db. 105db is the peaks. Unfortunately, most speaker manufacturers do not give us the specifications to calculate long term, and short term sound pressure level capability. A speaker that has specs of 200W and 91db/w @ 1m really does not tell us enough to accurately calculate how loud for how long the speaker can play. As part of CEDIA Standards efforts, the RP1 workgroup are creating a recommended practice called Performance Facts. This will be a list of engineering specifications that a manufacturer should supply so that we as cinema designers can make informed engineering decisions. We hope to publish the audio (video will come as part two) part in around September of this year.
The amplifier also makes a big difference to how loud a speaker will go without distorting. If the speaker is active (has x-over before the power amps with the amp usually built in), the manufacturer should be able to give accurate specs about its maximum SPL capability. If the speaker is passive, if possible spec an amp with 3db (twice the power) headroom to ensure that it never clips, and distortion is kept low. You can then limit maximum SPL on the processor.
What is the speaker’s coverage?
The chances are that the room you are designing has more than just one seat. As discussed above, as you move away from a speaker’s acoustic axis, its frequency response will change. It is therefore really important to specify a speaker that covers the audience area with similar sound at all seats.
This is especially important for overhead speakers used in immersive audio formats. These often point straight down (Nice and easy for interior design but generally poor for performance) meaning that the listener can be 50º off axis. Also be careful with wide L & R (often used in 9.1.6 and above arrays) speakers. Again, manufacturers do generally not supply the data that allows you to make predictions about a speaker’s coverage. The heat map data shown below allows you to accurately predict a speaker’s coverage and off-axis frequency response. This will ensure that your design delivers great performance at every seat.
Graphic courtesy of Perlisten and an example of excellent measured performance.
How well behaved is the speaker off-axis?
A speaker’s early reflections window frequency response (if you’d like to read up more on what this means, get Floyd Toole’s book ‘Sound Reproduction’, and read this standard - ANSI-CTA-2034A) will give a great prediction of its in room steady state frequency response. The Early Reflections Directivity Index will give you an at-a-glance idea of how well behaved the speaker is off-axis. A not too bumpy Early Reflections DI is a good indication that a speaker is predictable and will sound similar in different rooms. The ERDI shown on the above plot is an example of excellent performance.
How does it sound?
Now we get subjective. A Mercedes, BMW, Lexus or Audi are all superbly engineered. Drive one of them and you’ll likely love it. Drive them all and you’ll definitely have a favourite. That doesn’t make the others bad, it’s just that you like the balance of design compromises made by one of them. It’s the same with speakers. PMC, Meyersound, Waterfall, Steinway Lyngdorf, Ascendo, Wisdom Audio, Alcons, JBL, Perlisten, B&W etc (This list is neither exhaustive, nor a recommended list. It’s just illustrating a point) all make excellent speakers. They all, however, sound ‘different’. Many of these differences can be explained by looking at the specs (if available). If the early reflections window frequency response is available, that will give you a great idea of a speaker’s in-room tonality.
Different speakers also exhibit different impulse responses. This is how a speaker reacts to a short sound impulse and shows how the sound rises and how the subsequent mechanical energy decays. Many factors affect this - Driver materials and design, cabinet materials and construction, crossover etc.
A speaker’s phase response is also important. This is how in-phase different frequencies are relative to each other. A flat phase response is one of the most difficult things to achieve in audio.
But doesn’t electronic EQ just fix everything?
No. EQ is amazing and every system needs it. At low frequencies, it’s absolutely essential. At mid and high frequencies, it’s best to get it mostly right in the physical domain (speaker choice, speaker mounting, speaker angling and room acoustics) and then use EQ minimally.
This is a BLOG post and just touches on the subject. I hope it has shown you that the way a speaker sounds is a combination of many objectively measurable engineering aspects; some of them easier to predict in a design than others. At HTE we can work as much or as little on the room’s technical design as is required by the project. As an industry, we tend to be rather tribal when it comes to speaker specification. One brand rarely covers every use scenario and it’s important to design, before specification is done.
A designer is a problem seeker. An engineer is a problem solver. By correctly engineering a system using product specs to predict its performance, together we can deliver consistently amazing and immersive results.