2010 Winter Symposium Round Table - Review
ESP - Evaluating Speaker Performance
Moderator: Dr. Floyd E. Toole
As engineers, we hate guesswork. It is wasteful of time, energy and materials. Yet, in spite of our considerable technical knowledge and skills in designing loudspeaker components and systems, we are often forced to rely on something resembling ESP - the out-of-body Extra-Sensory Perception that is the topic of twilight-zone plots.
Measurement data are essential in the design of transducers and the loudspeaker systems created with them. Technical measurements are highly repeatable. Subjective evaluations provide a measure of whether the designs are successful but, unless the tests are done blind, with some degree of control, the results are highly variable.
Our customers, whether they are in the OEM or the retail market have expectations of how they want their products to sound. The historical problem has been that they have difficulty expressing those expectations in ways that can be translated into product descriptions and quantitative design objectives. Engineers are forced to Estimate Speaker Parameters based on their interpretations of incomplete information. In the end, many simply design the best product they know how to and cross their fingers. Fortunately, human adaptation allows most customers to think that their dreams have been realized. ESP sometimes works, but not always.
From the customer perspective, the non-standardized specifications we provide on our products, whether they are raw transducers or complete systems, are often inadequate to permit any predictions of sound quality in reflective spaces: cars, living spaces or auditoriums. In the consumer world, most loudspeaker specifications are insults to our intelligence. Forced to rely on reviews in the popular press and “listening tests” in the few stores that can provide them, these poor souls truly need ESP if they are to make good choices.
We can do much better than this. As I explained at length in the book “Sound Reproduction”, we really know a lot about how to design good loudspeakers, and how to use them to advantage in reflective spaces. The relationship between anechoic measurements and subjective opinions is no longer mysterious. More work can still be done, no doubt, but we are at a point where it would be useful to have the knowledge we have disseminated through the industry, so that engineers can spend less time on ESP and more time focusing on delivering better performance within the constraints of a design, and customers need not rely on ESP when choosing loudspeakers for real world applications. It is time for ALMA International, an organization dedicated totally to loudspeakers, to decide on, and promote better, more relevant, technical data for its own use, and for its customers.
Attendees were placed on one of the following five tables based on market segment, and addresed the following questions:
- How do you Evaluate System/Speaker Performance? This would focus on what data is measured and how. For example: frequency response, Impedance, Phase, Polar Pattern, sensitivity, and so on.
- What data do you publish? This would cover not only the data published but also how it is displayed - smoothing, resolution, x/y format and axis labeling, and so on. Another aspect of this is what data do you not publish and why?
- How do you correlate the measured data to your customer’s requirements and/or satisfaction? How do we inform professional customers and consumers that these data sets are useful, what they infer in terms of sound quality, what they mean in terms of selecting appropriate amplifiers, etc?
- How do you insure that you duplicate this performance long term with EOL testing? i.e. How do you choose limits that maintain the integrity of the design, just avoiding audible or audibly distracting faults?
The notes generated by each group in the session can be downloaded by clicking on the link to the right of the group name.
|
Download Notes |
|
Download Notes |
|
Download Notes |
|
Download Notes |
|
Download Notes |
ALMA Winter Symposium 2010 Review:
Main | Training | Papers | Tutorials | Round Table | Panel Discussion | Industry Issue | Exhibit Hall | Members Dinner