- Joined
- Jun 22, 2001
- Posts
- 4,111
- Likes
- 1,336
Very early stuff was done on differtent head. Then we did a bunch on this head, but found there were some opportunities to make mistakes. These mistakes were primarily sealed headphones in that that if you didn't get a good seal you could get a significant drop off in the bass. I'm sure some of these still exist on the web site.
But the graphs for unsealed designes are pretty reliable and repeatable. I've seen us measure HD650s maybe a dozen times over the last couple of years and the graphs are spot on each time.
I want to caution two things: The graphs don't tell the whole story, not by a long shot; and generally they are legitimate measurements.
They are allways done the same way with the same amp and set to the same acoustic levels. It's worth mentioning that peoples impressions of a particular pair of headphones here is allways done with a variety of amps and ears, so I would contend that the graphs are a legitimate point of information to directly compare on pair of cans against another. If the graph shows a can as having more bass than another, you can bet it does. But it won't tell you if it's tight or muddy.
Someone asked if we use pink noise or a sweep: we do sweep the audio signal. We have tried averaged pink noise measurements and they turn out identical but a bit noisy. So we use sweeps as it is much faster and more accurate.
On the Fletcher-Munsen curves: What you want is flat response out in the far field and let the human hearing system to do what it wants. Not only are there Fletcher-Munsen curves but most people have places where their own ears have 5db difference left to right. Your head gets used to hearing your ears over time, whatever the response.
The problem with headphones is that they bypass the far field and couple directly to your ears, thereby removing some portion of the individuals acoustic filtering of the far field. So the headphone has to compensate for this short circuiting of the normal acoustic.
Our measurements are essentialy done at the ear drum. We then use compensation curves to dial out the ear canal resonances. The curves you see are essentially the frequency response as measured at the entry to the ear canal.
I'm of the opinion that we should smooth the curves SIGNIFICANTLY more than we do. It seems to me that the only information that we can really get out of a pair of cans is about the gross frequency response. The remaining details may be of some use but mostly obscure the general trend line. I'm considering getting rid of the finer detail in the graph.
But the graphs for unsealed designes are pretty reliable and repeatable. I've seen us measure HD650s maybe a dozen times over the last couple of years and the graphs are spot on each time.
I want to caution two things: The graphs don't tell the whole story, not by a long shot; and generally they are legitimate measurements.
They are allways done the same way with the same amp and set to the same acoustic levels. It's worth mentioning that peoples impressions of a particular pair of headphones here is allways done with a variety of amps and ears, so I would contend that the graphs are a legitimate point of information to directly compare on pair of cans against another. If the graph shows a can as having more bass than another, you can bet it does. But it won't tell you if it's tight or muddy.
Someone asked if we use pink noise or a sweep: we do sweep the audio signal. We have tried averaged pink noise measurements and they turn out identical but a bit noisy. So we use sweeps as it is much faster and more accurate.
On the Fletcher-Munsen curves: What you want is flat response out in the far field and let the human hearing system to do what it wants. Not only are there Fletcher-Munsen curves but most people have places where their own ears have 5db difference left to right. Your head gets used to hearing your ears over time, whatever the response.
The problem with headphones is that they bypass the far field and couple directly to your ears, thereby removing some portion of the individuals acoustic filtering of the far field. So the headphone has to compensate for this short circuiting of the normal acoustic.
Our measurements are essentialy done at the ear drum. We then use compensation curves to dial out the ear canal resonances. The curves you see are essentially the frequency response as measured at the entry to the ear canal.
I'm of the opinion that we should smooth the curves SIGNIFICANTLY more than we do. It seems to me that the only information that we can really get out of a pair of cans is about the gross frequency response. The remaining details may be of some use but mostly obscure the general trend line. I'm considering getting rid of the finer detail in the graph.