Speaker design questions
A recent Hi-Fi magazine featured a buying guide with hi-fi audio components described by one or two lines. A more comprehensive review would no doubt result in many more words, however the fact is that no two descriptions were the similar.
If Hi-Fi is about accuracy, and speakers accurately produce the recorded sounds, how can Hi-Fi speakers all sound different? Bookshelf speakers may be limited in their capability, but what about floor standers costing tens of thousands of dollars? Should they sound almost the same with subtle differences? How subtle?
Has anyone built floor standers or bookshelf speakers for that matter that sound better than the commercial efforts, and have tested better?
When designing, building and refining a speaker, to what degree does tuning the speaker to one's preferences actually distort the performance?
A recent Hi-Fi magazine featured a buying guide with hi-fi audio components described by one or two lines. A more comprehensive review would no doubt result in many more words, however the fact is that no two descriptions were the similar.
If Hi-Fi is about accuracy, and speakers accurately produce the recorded sounds, how can Hi-Fi speakers all sound different? Bookshelf speakers may be limited in their capability, but what about floor standers costing tens of thousands of dollars? Should they sound almost the same with subtle differences? How subtle?
Has anyone built floor standers or bookshelf speakers for that matter that sound better than the commercial efforts, and have tested better?
When designing, building and refining a speaker, to what degree does tuning the speaker to one's preferences actually distort the performance?
Speakers -- even the best ones -- still have a long ways to go. Every speaker is a set of compromises.
I'd guess that we are maybe 10-20% of the way to where we will be someday. It is entirely possible to make valid compromises to get a very good speaker that sounds quite different from an equally valid speaker that has a different set of compromises.
dave
The papers linked here may help understand loudspeaker design.
It does use mathematical analysis of speakers, however:
Read Research - Articles
It does use mathematical analysis of speakers, however:
Read Research - Articles
If you aren't being cynical, then accuracy always has compromises, and its' certainly possible for different speakers to sound differently. Also, oscilloscopes don't buy speakers. Human's do, and what the emotional value of a dollar is, or a sound is quite variable.
If you are being cynical, then the High End industry is nothing more than fashion made by the cynical for the ignorant.
If you want to see how different speakers can sound look through the results in speakermeasurements.com They are the only one's who publish distortion and compression figures regularly. The compression figures are especially telling of the great from the mediocre.
The other thing is that speakers are designed for different rooms, tastes, etc.
There is no single standard of recording that goes from microphone location all the way through living room. This also greatly affects what is "best" or "perfect." There's some loosely coupled relationship between the recording engineers and the music and movie consumers, each trying to guess what the other is thinking.
Having said this, we've gone through this exercise before, but the markup on speaker drivers once in a cabinet can be anywhere from 5x to 20x depending on the label. Much more about reporting in audio in my mind, is about maintaining the status of the high end speakers and justifying their price.
As for tuning to your preferences, that's what DIY is all about. Regardless of the standard you use, it's all a set of compromises made for the builder's tastes. My measurement microphone does not buy capacitors. I do. When I do, I make myself happy above all. Do I have the best sounding speakers in an apartment in San Francisco as a result? You bet I do.
Erik
If you are being cynical, then the High End industry is nothing more than fashion made by the cynical for the ignorant.
If you want to see how different speakers can sound look through the results in speakermeasurements.com They are the only one's who publish distortion and compression figures regularly. The compression figures are especially telling of the great from the mediocre.
The other thing is that speakers are designed for different rooms, tastes, etc.
There is no single standard of recording that goes from microphone location all the way through living room. This also greatly affects what is "best" or "perfect." There's some loosely coupled relationship between the recording engineers and the music and movie consumers, each trying to guess what the other is thinking.
Having said this, we've gone through this exercise before, but the markup on speaker drivers once in a cabinet can be anywhere from 5x to 20x depending on the label. Much more about reporting in audio in my mind, is about maintaining the status of the high end speakers and justifying their price.
As for tuning to your preferences, that's what DIY is all about. Regardless of the standard you use, it's all a set of compromises made for the builder's tastes. My measurement microphone does not buy capacitors. I do. When I do, I make myself happy above all. Do I have the best sounding speakers in an apartment in San Francisco as a result? You bet I do.
Erik
Dave
Well it's just that anyone starting out will feel that with all the effort will result in only 10-20 % of the goal of accurate sound reproduction.
It won't bring your favourite musicians into your living room if you just close your eyes.
Your answer makes sense, though, there is much 'headroom' for improving one aspect of performance over the other, but not all.
Well it's just that anyone starting out will feel that with all the effort will result in only 10-20 % of the goal of accurate sound reproduction.
It won't bring your favourite musicians into your living room if you just close your eyes.
Your answer makes sense, though, there is much 'headroom' for improving one aspect of performance over the other, but not all.
Last edited:
Proof is in the pudding. You can create a pretty convincing and enjoyable illusion.
If you think that 10-20% from a state of the art speaker is not sufficient, consider that when the recording is made likely less than that is being captured.
HiFi has a lot more potential, but that does not preclude us from enjoying what we can do.
dave
See, the idea of a musician in my room, drinking my booze, making my women swoon and trashing the place before they leave, isn't my cup of tea.....
I don't think people realize just how varied what we want out of our music can be. I don't always have the desire to have "realistic" dynamic range sessions in my living room. I want to enjoy music often though. If I were making speakers to sell, should I design for the full concert dynamic range, or for speakers that sound good for people to enjoy a variety of listening moods?
Best,
Erik
I don't think people realize just how varied what we want out of our music can be. I don't always have the desire to have "realistic" dynamic range sessions in my living room. I want to enjoy music often though. If I were making speakers to sell, should I design for the full concert dynamic range, or for speakers that sound good for people to enjoy a variety of listening moods?
Best,
Erik
The thing with speakers is that they operate in 3D space. A channel of an amplifier or a preamp or other purely electronic device has one output point where you can measure the frequency response or relative distortion, you connect your measurement gadget to that point, and measure it, and there it is for what you're measuring at least.
But a speaker has an INFINITE number of output sensing places --- it will be different at every point in space (further, nearer, higher up, lower down, to the right, to the left...) you care to measure at. And it outputs to ALL of those places at the same time, so you can't just measure where you will sit and say that's it, because the sound that goes out to other directions will excite room objects and surfaces and you WILL hear all of it. If you doubt that, turn a speaker away from you in a room, play some music, and notice that it sounds different but not tremendously less loud than when pointed at you. And if you move the speaker or stuff in the room, there will be a difference in the measured response (and nearly always the audible response, too). So that alone makes loudspeakers way trickier than other components (and is why there's all the interest in things like radiation patterns, dipoles, controlled directivity devices, arrays, etc).
And it gets worse. You are probably listening to music in stereo, so whatever you hear comes through only two separate audio signal channels. Being a human hearing live music, though, you would normally hear the event in a space (the same space you are in) so the music is heard not just from one direction, but envelopes you from all around you (because the recorded event happened in a space with acoustic surfaces, too, and that space was excited by original acoustic energy that originated from 3D objects having many more than just two output points). For realistic envelopment, you need to have energy coming from around you, and stereo channels going directly to your ears only (even if you can do that by absorbing everything) can't give that. With stereo at least, the room has to provide the envelopment. But every acoustic pattern (directions, responses, delays) will be different with every speaker design and every room, none will be the same as in the original space, but some will be more convincing that others... and with different realism for most people, too.
Still worse, loudspeakers exhibit much, much worse nonlinear distortion (harmonic, IMD), much worse phase distortion (hardly any speakers can reproduce a recognizeable square wave, nearly any amplifier can), and much worse frequency response (the true unsmoothed frequency response of a loudspeaker won't be flat, compared to an amplifer. In a room, with all the reflections mixed in, it will look like a grassy field with that someone has been playing in with handgrenades.
That's why speakers will virtually never sound the same, even expensive ones, even ones designed by the same designer! They aren't (relatively) very accurate to begin with in any particular direction, and when put into a room they at best make a decent illusion rather than a reproduction.
But a speaker has an INFINITE number of output sensing places --- it will be different at every point in space (further, nearer, higher up, lower down, to the right, to the left...) you care to measure at. And it outputs to ALL of those places at the same time, so you can't just measure where you will sit and say that's it, because the sound that goes out to other directions will excite room objects and surfaces and you WILL hear all of it. If you doubt that, turn a speaker away from you in a room, play some music, and notice that it sounds different but not tremendously less loud than when pointed at you. And if you move the speaker or stuff in the room, there will be a difference in the measured response (and nearly always the audible response, too). So that alone makes loudspeakers way trickier than other components (and is why there's all the interest in things like radiation patterns, dipoles, controlled directivity devices, arrays, etc).
And it gets worse. You are probably listening to music in stereo, so whatever you hear comes through only two separate audio signal channels. Being a human hearing live music, though, you would normally hear the event in a space (the same space you are in) so the music is heard not just from one direction, but envelopes you from all around you (because the recorded event happened in a space with acoustic surfaces, too, and that space was excited by original acoustic energy that originated from 3D objects having many more than just two output points). For realistic envelopment, you need to have energy coming from around you, and stereo channels going directly to your ears only (even if you can do that by absorbing everything) can't give that. With stereo at least, the room has to provide the envelopment. But every acoustic pattern (directions, responses, delays) will be different with every speaker design and every room, none will be the same as in the original space, but some will be more convincing that others... and with different realism for most people, too.
Still worse, loudspeakers exhibit much, much worse nonlinear distortion (harmonic, IMD), much worse phase distortion (hardly any speakers can reproduce a recognizeable square wave, nearly any amplifier can), and much worse frequency response (the true unsmoothed frequency response of a loudspeaker won't be flat, compared to an amplifer. In a room, with all the reflections mixed in, it will look like a grassy field with that someone has been playing in with handgrenades.
That's why speakers will virtually never sound the same, even expensive ones, even ones designed by the same designer! They aren't (relatively) very accurate to begin with in any particular direction, and when put into a room they at best make a decent illusion rather than a reproduction.
I've been telling my wife that it's been the musicians doing all that!
Last edited:
Dave
Thanks for the link. Now I can compare the measurements with the reviews and draw some conclusions.
What is 'compression' - I did not see any parameter named compression on the speaker measurement pages.
The other question was - how do the speakers you constructed compare to the commercially produced ones in terms of price/performance?
Thanks for the link. Now I can compare the measurements with the reviews and draw some conclusions.
What is 'compression' - I did not see any parameter named compression on the speaker measurement pages.
The other question was - how do the speakers you constructed compare to the commercially produced ones in terms of price/performance?
Compression is the opposite of linearity. It means that as the input voltage goes up the output goes out uniformly. Here's an example from the Focal Sopra.
Chart B: Difference @ 95dB from 70dB, 50Hz - 20kHz (measured @ 2m)
So, the idea is that they measure the frequency response at 70 dB (probably at 1 kHz) then they add 25 dB (70 dB + 25 dB = 95 dB) to the input signal and measure the frequency response. You can then compare the difference in the FR. From this chart, you can see that around 10 kHz instead of getting 25 dB more output, you only get 22 dB more.
It's really difficult to compare price/performance to commercial models since those go from absolute crap to pretty good for the same prices. Now, parts cost to purchase price is more easily measured.
If you'd like a technical view of them however you can find a pretty detailed discussion in this thread.
Best,
Erik
Chart B: Difference @ 95dB from 70dB, 50Hz - 20kHz (measured @ 2m)
So, the idea is that they measure the frequency response at 70 dB (probably at 1 kHz) then they add 25 dB (70 dB + 25 dB = 95 dB) to the input signal and measure the frequency response. You can then compare the difference in the FR. From this chart, you can see that around 10 kHz instead of getting 25 dB more output, you only get 22 dB more.
It's really difficult to compare price/performance to commercial models since those go from absolute crap to pretty good for the same prices.
Now, parts cost to purchase price is more easily measured. If you'd like a technical view of them however you can find a pretty detailed discussion in this thread.
Best,
Erik
Last edited:
Here was a useful thread on DIY vs. commercial prices.
http://www.diyaudio.com/forums/multi-way/284938-looking-suggestions-bookshelf-build.html
Best,
Erik
http://www.diyaudio.com/forums/multi-way/284938-looking-suggestions-bookshelf-build.html
Best,
Erik
Thanks Erik. So the compression measure actually measures if an input signal produces the same output level across the frequency band.
I had no idea speaker drivers did this. Speaker drivers are so ill-behaved beasts, it seems unlike amplifiers.
Your comment on commercial prices and the other thread echoes what I have read from many sources.
There is wide variation in speaker prices which got me wondering- $2000 to $30,000 for floor standers, and speakers costing $500,000 or more. Given the above formula, parts would cost about 500,000/13 = $ 38,461. The rest is sheer intellectual property : wages + test equipment + anechoic chambers and such?
I had no idea speaker drivers did this. Speaker drivers are so ill-behaved beasts, it seems unlike amplifiers.
Your comment on commercial prices and the other thread echoes what I have read from many sources.
(Post #5 on the "Looking for suggestions... " thread.
There is wide variation in speaker prices which got me wondering- $2000 to $30,000 for floor standers, and speakers costing $500,000 or more. Given the above formula, parts would cost about 500,000/13 = $ 38,461. The rest is sheer intellectual property : wages + test equipment + anechoic chambers and such?
Thanks Erik. So the compression measure actually measures if an input signal produces the same output level across the frequency band.
I had no idea speaker drivers did this. Speaker drivers are so ill-behaved beasts, it seems unlike amplifiers.
Your comment on commercial prices and the other thread echoes what I have read from many sources.
(Post #5 on the "Looking for suggestions... " thread.
There is wide variation in speaker prices which got me wondering- $2000 to $30,000 for floor standers, and speakers costing $500,000 or more. Given the above formula, parts would cost about 500,000/13 = $ 38,461. The rest is sheer intellectual property : wages + test equipment + anechoic chambers and such?
Compression / Linearity tests measure whether the output tracks the input evenly across the frequency spectrum. Add 3dB at any point and the output should also be +3 dB. Add 10 dB at any frequency and the output should be +10 dB at that frequency relative to some baseline.
As for relative costs, who knows?
I mean, it's rare a maker does what Sony did, which is pull all drivers from a single brand in a way that we can calculate from off the shelf parts that easily. This is one of the major reasons manufacturers make customs, so that we can't compare so directly. Sometimes those custom drivers are just new labels, sometimes changes are more significant. For instance, Magico has a custom Beryllium + Diamond coated tweeter as well as carbon fiber + graphite drivers made by ScanSpeak we can't possibly compare to their regular drivers. Wilson uses Focal tweeters that are about 10x heavier than what goes into the Sopra. If you want a midrange (cost wise) speaker with Focal tweeters, Wilson is the way to go. The very best we can do is bound the problem across one dimension and say, based on what evidence we have, that at best, the price of a commercial speaker is about 5x to 20x the driver cost.
Given that formula, my current stereo speakers are around $800 in drivers each, or $1,600 for the pair. I would have had to spend $8000k to $32,000 for speakers similar to mine. Of course, I also overspent on the crossovers, and this does not include cabinetry. Here's a very nice brand, with really fun cabinetry and stands for around $6500, the Lawrence Audio Violins:
An externally hosted image should be here but it was not working when we last tested it.
I must say, they're beautiful. They use Aurum Cantus drivers with a retail driver price of around $500 per speaker , or $1,000 per pair. Using our formula we'd say it's between $5k and $20k a pair. The actual $6,500 price tag is near the low end of the range. Not too shabby considering the lovely woodworking and design, and that a stand is included. Looking at crossover pictures online, it seems they use the Mundorf MKP caps for a ribbon tweeter, and undervalued IMHO. Shows very good taste while being frugal. The Mundorf MKP's are also among the least expensive "boutique" capacitors available. Compare that to the Sony SS AR-1's which are around 20x and big monolithic beasts.
Best,
Erik
Last edited:
Buying based on looks is fine, but there must be some way of measuring capability. Auditioning speakers also works.
I would think the ultimate test would be to have an A/B switch between a live band playing in a distant room through audio quality cable and monitor speakers and a recording of a live band playing through your speakers.
Of course a lot will depend on the monitor speakers. Maybe it's all about the monitor speakers in the recording studio?
I would think the ultimate test would be to have an A/B switch between a live band playing in a distant room through audio quality cable and monitor speakers and a recording of a live band playing through your speakers.
Of course a lot will depend on the monitor speakers. Maybe it's all about the monitor speakers in the recording studio?
I agree that looks are not everything. I was just saying that in addition to being high value, they throw in nice cabinetry. To do that at such a low price point is a good thing.
I'm not sure a lot of buyers would agree with your statement. I don't necessarily always want a live band playing. There's a reason why studio albums sell.
Of course, the ultimate guide to "value" is your own ears and gut. We may think this is all so scientific, but oscilloscopes and FFT software doesn't buy speakers. We do. I was not put on this planet to make some FFT algorithm happy, I buy and design what I like.
You should too.And yes, how studio monitors are balanced, and which studio monitors are being used has a great deal to do with what gets recorded, but one of the best recording engineers I know of, Cookie Marenco over at Blue Coast records uses absolute crap speakers.
Which is weird becuase they were also selling a very nice USB DAC, the UFO in Blue. Shame mine died after 2 months. Anyway, I digress. My point was that it's kind of a guessing game between us speaker buyers and the recording engineers. I don't know how Cookie can use such speakers and select such a nice sounding DAC and produce such great tracks.Best,
Erik
Last edited:
This is more and more about the recording process. Of course there is recording quality which I understand varies, and I am only just beginning to sense the differences between recordings when more accurate audio systems are used.
I once saw a some information posted on audio recording and where the microphones are placed and so on. I think it may have a bearing on speaker design. It was funny that the huge kick drum is recorded using a small microphone - so the sound from the drum membrane flows into a tiny area in the microphone and the playback is supposed to create the same effect on an 8 inch speaker. Something audio information must go missing, I think.
If different audio recording engineers place greater emphasis on one part of the sound spectrum, or a certain instrument, or set of instruments, such as bass and drums, then some speakers may play those tracks better than others.
I have not yet got into analysing stereo, though there are some test sites and much information, however I prefer to hear synthesizers appear in both left and right chanells equally, spread wide across the sound stage. This is most probably an artificial effect caused by mixing. Some recordings place the vocalist up close and personal, for example in folks songs or tracks with acoustic instruments. So a set of speakers may be more suited to producing this effect than others. The live band effect really is not what they are aiming for, however there are some excellent live recordings out there.
Eric, you apparently have contacts with the recording industry, do you have any info on the recording process vs Hi-Fi audio reproduction? There seems to be a major disconnect between the two.
I once saw a some information posted on audio recording and where the microphones are placed and so on. I think it may have a bearing on speaker design. It was funny that the huge kick drum is recorded using a small microphone - so the sound from the drum membrane flows into a tiny area in the microphone and the playback is supposed to create the same effect on an 8 inch speaker. Something audio information must go missing, I think.
If different audio recording engineers place greater emphasis on one part of the sound spectrum, or a certain instrument, or set of instruments, such as bass and drums, then some speakers may play those tracks better than others.
I have not yet got into analysing stereo, though there are some test sites and much information, however I prefer to hear synthesizers appear in both left and right chanells equally, spread wide across the sound stage. This is most probably an artificial effect caused by mixing. Some recordings place the vocalist up close and personal, for example in folks songs or tracks with acoustic instruments. So a set of speakers may be more suited to producing this effect than others. The live band effect really is not what they are aiming for, however there are some excellent live recordings out there.
Eric, you apparently have contacts with the recording industry, do you have any info on the recording process vs Hi-Fi audio reproduction? There seems to be a major disconnect between the two.
Last edited:
Found info on recording studios:
Multimedia Audio Course - The recording studio - Outline of a recording studio - Sound engineering - Audiosonica
This does not sound good:
"The first thing we notice is that there are two rooms: a recording room (where the recordings of sounds takes place) and a control room (where the sounds are handled and manipulated)"
Did he say "manipulated"?
Multimedia Audio Course - The recording studio - Outline of a recording studio - Sound engineering - Audiosonica
This does not sound good:
"The first thing we notice is that there are two rooms: a recording room (where the recordings of sounds takes place) and a control room (where the sounds are handled and manipulated)"
Did he say "manipulated"?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.