Any thoughts on headphones for comparisons of loudspeaker tonality? I'm comparing loudspeakers vs headphone overall frequency response. I do realize its apples vs oranges but I've yet to find a pair of headphones locally and under $300 that don't sound peaked at bass and high treble.
Not an easy task. I don't think that any headphones I have ever heard had the same tonal balance as well balanced speaker. And the fact that the same headphone can sound different on different heads doesn't help.
I have tried to get a better idea of headphone balance by measuring with in-ear microphones and comparing to a speaker system that has a good tonal balance both by ear and measurement. All that has really done is guide me to headphone EQ settings.
Certainly someone will have had better luck than I. Finding a tonal balance reference headphone would be nice.
I have tried to get a better idea of headphone balance by measuring with in-ear microphones and comparing to a speaker system that has a good tonal balance both by ear and measurement. All that has really done is guide me to headphone EQ settings.
Certainly someone will have had better luck than I. Finding a tonal balance reference headphone would be nice.
As above, the headphone is only half of the sound reproduction system. The other half is the 'load' the headphones are connected to, namely the listener's pinna with its elastic cartilage, and different shape and length ear canals. Well designed headphones are engineered to match the 'average' response when coupled with a range of anatomies that represent the variation in humans. Cheaper headphones are probably engineered to have a loudness contour built in to improve sales, not acoustics.
Koss ESP.9 electrostatics. They can be had for under 300. They will probably need to be taken apart and cleaned, refoamed and have the ear cushions stuffed or replaced. They are tight on the head but I think they sound similar to my electrostatic speakers.
About the best are the Sennheiser 650's but they need equalization. All headphones do, and all headphones need different eq for each person.
You can read over on ASR suggested eq for various cans they tested. There are also entire forums dedicated to headphones. There is a standard target for headphone response based on average ears but few manufactures can hit it as it is rather complex. A headphone does NOT measure flat.
I think it is a total waste of time though. You can learn a lot more a lot quicker with a calibrated mic. Much cheaper and you can repeat the measurements. Free software and a laptop. I have been told, that the iPhone mic is very well understood and there aps that work quite well.
You can read over on ASR suggested eq for various cans they tested. There are also entire forums dedicated to headphones. There is a standard target for headphone response based on average ears but few manufactures can hit it as it is rather complex. A headphone does NOT measure flat.
I think it is a total waste of time though. You can learn a lot more a lot quicker with a calibrated mic. Much cheaper and you can repeat the measurements. Free software and a laptop. I have been told, that the iPhone mic is very well understood and there aps that work quite well.
Great project to EQ headphones to Harman headphones curve
https://github.com/jaakkopasanen/AutoEq
Results a lots of headphones EQ curves
https://github.com/jaakkopasanen/AutoEq/tree/master/results
https://github.com/jaakkopasanen/AutoEq
Results a lots of headphones EQ curves
https://github.com/jaakkopasanen/AutoEq/tree/master/results
I think you can learn an awful lot by listening to real acoustic sources un-amplified, say a drum kit, or a piano, or a voice, and then listen to recorded drums, pianos and voices over loudspeakers. And then if you want to understand what loudspeakers and headphones do to sound, hold a coffee cup near your ear while listening to music replayed through loudspeakers, then a drink bottle, then a bucket. You'll either get the picture, or you won't; hopefully not the latter...
I have the cheaper little brother, the HD559. Those measure - on my ears - fairly close to what I get from a flat speaker system in a large room. At least above 1000 Hz. Below there is a beautiful hump centered at 150 Hz. That is easy to EQ with a wide filter having a Q of about 0.25 But if that 125 Hz hump is EQed to flat on the graph, then the headphones sound too thin. Perception is tricky! (and so is HRTF)
@dhaman Thanks for the links.
Thank you for your well though out replies. What I am trying to accomplish is controversial at best and downright damned by audiophiles at worst. Crutchfield has implemented a speaker comparison system that records various speakers of the same ilk in a controlled environment. They offer headphones that you may rent to compare these sound files so that you get an acoustic idea as to which speaker you prefer. I am wondering if this is sheer folly and if it may actually deter speaker sales due to the variables. If a prospective buyer hears these sound files via substandard headphones or perhaps via ones that are peaked in frequency may they reject said speaker because of its perceived sound?
That's certainly an interesting idea. At least they are doing something to allow direct comparisons and a level playing field. Do you know much about the system and recording techniques, is it described somewhere?
Perhaps play some music through your speakers, record it with a decent mic, then play the recording through your speakers to double the distortion. I like JVC flats, but I don't have any other headphones to compare them with.
The Crutchfield speaker compare idea is very interesting, especially in relation to online shopping in the Covid pandemic. Crutchfield have made a serious attempt at doing this properly to eliminate errors so that comparisons are fair, employing the services of not one but two PhD engineers do develop the idea. SpeakerCompare™
Acoustic engineers have used a similar technique for at least four decades to compare the sound of rooms and auditoria and even simulating the sound of acoustic spaces that haven't yet been built. David Griesinger has many decades worth of binaural recordings of real auditoria and rooms and has developed the algorithms and techniques for in-ear equalisation of headphones to achieve near perfect remote auralizeation. David Griesinger
It's not clear what prior art that Crutchfield's Dr Rick Wright and Dr Gary Gibbs drew upon, as a cursory Google Scholar and AES author search didn't reveal any published papers by the pair either together or separately. I'm going to have a listen to some of their speaker systems when I get a chance. I can see one serious and avoidable pitfall that Crutchfield has built in to their comparison approach, but I will reserve judgment until I've had a play.
Acoustic engineers have used a similar technique for at least four decades to compare the sound of rooms and auditoria and even simulating the sound of acoustic spaces that haven't yet been built. David Griesinger has many decades worth of binaural recordings of real auditoria and rooms and has developed the algorithms and techniques for in-ear equalisation of headphones to achieve near perfect remote auralizeation. David Griesinger
It's not clear what prior art that Crutchfield's Dr Rick Wright and Dr Gary Gibbs drew upon, as a cursory Google Scholar and AES author search didn't reveal any published papers by the pair either together or separately. I'm going to have a listen to some of their speaker systems when I get a chance. I can see one serious and avoidable pitfall that Crutchfield has built in to their comparison approach, but I will reserve judgment until I've had a play.
Thank you again for the references and support for some sort of subjective experience for speaker comparison. I did attempt to use a very good MIC the AT 2020 USB+ to compare to the original recorded source. It was fairly close to the original but as noted above "distortions" as in room acoustics, became part of the recording. I placed the MIC at 1 meter yet did capture the room as well. I suppose going outside would assist in removing most of he "distortions" but I would lose the room bass boost and the speaker will sound thin. Ill try to place absorptive panels around the MIC sides and see if that minimizes the acoustic space distortions yet leaves the room bass boost. Overall I'm just trying to allow the prospective buyers a reasonable way to compare models without relying on reviewers. That's a topic for another thread however...
From Audioadvice.com: "In-ear headphones have small drivers, so most can’t produce the types of highs and lows that on-ear or over-ear headphones can. They also pump the music straight into your ear-drum, so they’re less spacious or “open” sounding than other types of headphones." I tend to agree with this. You wont find many studios using ear buds for mastering...
Shinjutsu Audio said "I am wondering if this is sheer folly and if it may actually deter speaker sales due to the variables. If a prospective buyer hears these sound files via substandard headphones or perhaps via ones that are peaked in frequency may they reject said speaker because of its perceived sound?"
It looks like when you select "Listen Now" they prompt you to choose from a list of headphones that they've measured (EG Beyerdynamic DT770Pro). And then I presume they have a 2-part EQ, one to adjust for the speaker and the other to adjust for your headphones. So if what you want to do is use their tool then just pick up a pair they have set up in their tool. As long as you have a headphone they've measured (and your sample is is consistant with theirs), I expect their tool would account for the performance of your headphone.
One might wish to listen to a variety of virtualized speakers on the Crutchfield tool and compare them with one's own DIY efforts. I can't tell for sure what they're doing, but it sort of looks like they might be EQing to the anechoic measurement of the speaker, which can be fairly different from listening in a room with a reverberant field. So the Crutchfield tool might be nice for comparing speakers within the tool, but I'm not certain how much I'd trust their processing to sound like speakers really would in your room.
It looks like when you select "Listen Now" they prompt you to choose from a list of headphones that they've measured (EG Beyerdynamic DT770Pro). And then I presume they have a 2-part EQ, one to adjust for the speaker and the other to adjust for your headphones. So if what you want to do is use their tool then just pick up a pair they have set up in their tool. As long as you have a headphone they've measured (and your sample is is consistant with theirs), I expect their tool would account for the performance of your headphone.
One might wish to listen to a variety of virtualized speakers on the Crutchfield tool and compare them with one's own DIY efforts. I can't tell for sure what they're doing, but it sort of looks like they might be EQing to the anechoic measurement of the speaker, which can be fairly different from listening in a room with a reverberant field. So the Crutchfield tool might be nice for comparing speakers within the tool, but I'm not certain how much I'd trust their processing to sound like speakers really would in your room.
I've now had a play with the Crutchfield SpeakerCompare™. Forgetting about their choice of music tracks, which stopped me from wanting to listen to too many systems, I didn't like the sound of any of the speakers I listened up to and including the $5,500 Revels. The acoustic recordings reveal serious resonances in all of the systems I listened to, and none came close to the systems I have at home for clarity. Part of the issue is that anechoic listening is not pleasant and not how music is normally listened to, so the comparisons are relative, but not relevant to real listening in real rooms. The music samples they supply are not to my taste at all, and I could not get my own music tracks to play through unfortunately.
I don't understand the use of an anechoic chamber instead of gated measurements for measuring the loudspeakers. An anechoic chamber has a cut-off frequency approximated by a wavelength of 4 times the wedge depth, wavelengths longer than this are not absorbed to any great degree. A gated measurement (i.e. where the microphone is cut before the first reflection arrives) in the same volume can usually give a lower cutoff frequency for deeper valid bass frequency measurements. Maybe inherent limited resolution at of SpeakerCompare™ at low frequencies masked the differences between the weight of different systems.
As a side note, I once worked for a loudspeaker manufacturer who insisted on using his "anechoic" chamber. The anechoicity of his treasured room started to fall apart below 2kHz: that is the frequency at which the inverse square law started to fail. In a room of the same dimensions, a gated MLS measurement would be accurate down to under 100Hz. Hopefully Crutchfield have built a better anechoic chamber than the one I had to use.
For SpeakerCompare™ to be able to offer meaningful comparisons, there needs to be a way to choose a room to listen in from a number of standardised rooms of various volumes representing typical listening environments. For each volume there needs to be a brightness selection to represent the amount of damping in the room. And the speaker measurements need to include three dimensional data, i.e. say 15º measurements for all 360º in all three plains.
This kind of stuff is very old hat, maybe so old it has been forgotten about. The AES has Standard AES56-2008 (r2019): AES standard on acoustics - Sound source modelling - Loudspeaker polar radiation measurements, which is used to characterise the 3D performance of loudspeaker sound so the data can be used for auralising in simulated acoustic spaces. This is standard daily fare for professional acousticians.
Right now SpeakerCompare™ is a sideshow alley novelty, not a hifi tool IMO.
I don't understand the use of an anechoic chamber instead of gated measurements for measuring the loudspeakers. An anechoic chamber has a cut-off frequency approximated by a wavelength of 4 times the wedge depth, wavelengths longer than this are not absorbed to any great degree. A gated measurement (i.e. where the microphone is cut before the first reflection arrives) in the same volume can usually give a lower cutoff frequency for deeper valid bass frequency measurements. Maybe inherent limited resolution at of SpeakerCompare™ at low frequencies masked the differences between the weight of different systems.
As a side note, I once worked for a loudspeaker manufacturer who insisted on using his "anechoic" chamber. The anechoicity of his treasured room started to fall apart below 2kHz: that is the frequency at which the inverse square law started to fail. In a room of the same dimensions, a gated MLS measurement would be accurate down to under 100Hz. Hopefully Crutchfield have built a better anechoic chamber than the one I had to use.
For SpeakerCompare™ to be able to offer meaningful comparisons, there needs to be a way to choose a room to listen in from a number of standardised rooms of various volumes representing typical listening environments. For each volume there needs to be a brightness selection to represent the amount of damping in the room. And the speaker measurements need to include three dimensional data, i.e. say 15º measurements for all 360º in all three plains.
This kind of stuff is very old hat, maybe so old it has been forgotten about. The AES has Standard AES56-2008 (r2019): AES standard on acoustics - Sound source modelling - Loudspeaker polar radiation measurements, which is used to characterise the 3D performance of loudspeaker sound so the data can be used for auralising in simulated acoustic spaces. This is standard daily fare for professional acousticians.
Right now SpeakerCompare™ is a sideshow alley novelty, not a hifi tool IMO.
Last edited:
Thank you johnmath for the well thought out reply. I'm going to try gated measurements in REW as that's what I use for measurements. What I am interested in is not a way to absolutely prove that via the internet and headphones the best speaker for your room. What I am interested in is using an internet tool to compare our differing models with each other. I have had quite a few requests to "put the speakers on You Tube so I can hear them".
I know this is not even close to ideal. The variables are beyond my control. I do think however the request seeks some sort of reassurance that they are not overly out of bounds in sound quality. Our speakers use some unorthodox or perhaps unique combinations of drivers: back loaded horns topped with compression drivers. I believe that some people may be unfamiliar with the sound of compression drivers and are uncomfortable with the concept. Horns may be perceived by some as screechy or colored in response so they seek reassurance.
I guess its somewhat similar to a movie trailer: a bit of film used to convince an audience that the film has merit. Its not the whole film and its also skewed towards the more exciting parts of the film. You may like the whole film or may not.
That beings me to traditional ways of choosing components. Hear it in your home, search for reviews or go on faith by looking at its technology or measurements. Each has flaws. Hearing it in your home is the gold standard. Unfortunately it is costly regarding shipping both ways should you not like the component. As a manufacturer "tire kicking" is not a way to keep costs down. I need not beleaguer you with droning on about reviewers. Its a truly subjective experience that is easily twisted and distorted by $$$. Measurements are a good indication of quality but the proof of the pudding is in the eating.
How does one get around these barriers to sales in the internet sales realm?
I know this is not even close to ideal. The variables are beyond my control. I do think however the request seeks some sort of reassurance that they are not overly out of bounds in sound quality. Our speakers use some unorthodox or perhaps unique combinations of drivers: back loaded horns topped with compression drivers. I believe that some people may be unfamiliar with the sound of compression drivers and are uncomfortable with the concept. Horns may be perceived by some as screechy or colored in response so they seek reassurance.
I guess its somewhat similar to a movie trailer: a bit of film used to convince an audience that the film has merit. Its not the whole film and its also skewed towards the more exciting parts of the film. You may like the whole film or may not.
That beings me to traditional ways of choosing components. Hear it in your home, search for reviews or go on faith by looking at its technology or measurements. Each has flaws. Hearing it in your home is the gold standard. Unfortunately it is costly regarding shipping both ways should you not like the component. As a manufacturer "tire kicking" is not a way to keep costs down. I need not beleaguer you with droning on about reviewers. Its a truly subjective experience that is easily twisted and distorted by $$$. Measurements are a good indication of quality but the proof of the pudding is in the eating.
How does one get around these barriers to sales in the internet sales realm?
This is extremely stupid. "Hear how these speakers sound through your speaker/headphones".... I don't even know where to begin....
I am not sure what to make of your comment. Auralisation of both rooms and loudspeakers has been practiced for decades by acousticians using headphones.
Would you think a client is going to commit $100 million to build a concert hall if they don't knowwhat it will sound like and if that the compare, soloist or singer is going to be heard clearly in the front row, the middle, the side, the back row and the balcony? Would a metropolitan railway station or international airport accept a >>$10 million tender for a PA system if they didn't know what speakers will sound like and whether the PA can produce intelligible announcements free of flutter echoes along the entire length of each platform and concourse?
How do they achieve this? By three dimensional measurements of the speakers to produce a data set for each of the loudspeakers used that can then be used to convolve the sound of the loudspeakers in the modelled acoustic space in software, to generate an image of what the system will sound like at any point in that simulated space. The result is listened to through calibrated headphones or earbuds to remove the room effect in which the evaluation takes place.
Professional loudspeaker manufacturers who want the opportunity to supply their product into new auditoriums and other spaces supply EASE data (EASE = Electro-Acoustic Simulator for Engineers) measured in accordance with "AES Standard AES56-2008 (r2019): Sound source modelling" so acousticians can simulate how those loudspeakers will sound in the the spaces they are planning but have not yet built, using software such as CATT Acoustic, EVERtims, Bose Modeller, ODEON, EASE, etc.
Last edited: