[cranky]

6dB is twice the electrical power. Twice the sound power (human perception) is not 2x but 2^10. Or 10dB. If you wanted 5 times the (acoustic) power it would be 5^10.

Edit: About speakers being linear or not, that is dependent on many factors. Some professional speakers will run up to 130dB without breaking a sweat all day, which is enough to damage your hearing. Of course monstrous amounts of power and very careful design is required to hit those levels. This is not applicable for home audio. At home you normally don't exceed 85dB (or shouldn't) at your listening position. This means that good recordings with peaks of up to 15-20dB may be enjoyed without going deaf.

Edit again: Here's another link that explains the confusion between x2 and ^10. http://www.sengpielaudio.com/calculator-levelchange.htm Of course the best source to learn from is the Master handbook of Acoustics (Walton).

[vina]

Originally Posted by cranky

6dB is twice the electrical power. Twice the sound power (human perception) is not 2x but 2^10. Or 10dB. If you wanted 5 times the (acoustic) power it would be 5^10.

I went through the links last night and understood as much - the author of one of the articles uses this as an example ("I want to double the volume - so you make it 10dB more")

Originally Posted by cranky

Edit: About speakers being linear or not, that is dependent on many factors. Some professional speakers will run up to 130dB without breaking a sweat all day, which is enough to damage your hearing. Of course monstrous amounts of power and very careful design is required to hit those levels. This is not applicable for home audio. At home you normally don't exceed 85dB (or shouldn't) at your listening position. This means that good recordings with peaks of up to 15-20dB may be enjoyed without going deaf.

If you look at the plots given in the first article you sent, the impedance of subwoofers changes significantly with frequency. This is a case of impedance mismatch - a calibrated pre-distortion scheme before the driver can make the overall response flat, but since it can not do anything about peak power, overall volume will reduce.

Also loudness perception will reduce too (simply because there is no compression near the peaks of the music)



For a custom design, given that most car cabins are less than 2mx2mx1m in size, and have a lot of absorbent material in the car, frequencies less than 100Hz will have a hard time forming standing waves in any dimension of the car. I may be wrong here (please correct me in that case) but low frequency boxes should be very much possible if they are sealed on one side and hae a moving diaphragm on the other side. Rather than creating a pressure wave, you'll be compressing and decompressing the air in the entire cabin (the ear of course wouldn't know the difference)

For high frequencies - absorbent materials should be good, otherwise not much of a problem.

Mid-range can and will set up standing waves - is there any standard solution?

Originally Posted by cranky

Edit again: Here's another link that explains the confusion between x2 and ^10. Loudness volume sound level change factor of perceived loudness formula calculate power level noise levels volume logarithm dependence three four fold loudness sound - by what factor does level decrease dependence comparison decibel levels 3 dBSPL 6 Of course the best source to learn from is the Master handbook of Acoustics (Walton).

great links thanks for all of them.

[DerAlte]

Originally Posted by vina

Well I always thought speakers' job is to reproduce the input perfectly - the audio signal processing guys put a lot of effort into getting the timber etc. right. ...

O @ vina bhaiya, o @vina bhaiya, at least don't make falooda out of history of sound reproduction. The quest for the best speaker possible (and hence best reproduction possible) precedes even rudimentary signal processing (RADAR / SONAR application) by decades. You are talking of the very recent phenomenon of trying to "get the timbre that I want" (like getting the timbre of valve amps using DSP), whereas the traditional approach relied (still does) on getting the timbre / sound signature right by appropriate material science and engineering. The price of a speaker reflects in some measure the recovery of the investment that went into getting it right by research.

Originally Posted by vina

...I'm confused: If the impedance match is good ...
Also bigger diaphragm = better impedance matching ... how is it related to mechanical limits ...

Err ... what exactly are you trying to get at by using the term "impedance matching"? The conventional usage of the term has only to do with efficiency of electrical power transfer, so it is difficult to relate to if you use it in the context of mechanical power transfer (cone to air; cone is more colloquially used in the context of speakers than 'diaphragm')

Originally Posted by vina

... position of speakers vs. listeners, but also there are uncertainties like how will the waves flow - for example how does the roofing material behave - does it absorb, if so how much, can you get a standing wave pattern ...

Originally Posted by vina

... My point is, if you want to spend more money, speakers may be the wrong place ...

Hellos, K.I.S.S. principle please!?!? Further decomposition is inimical to solving the problem.

And, contrary to what you think, changing to better speakers is the first place where you gain the most - all the people here who changed from stock OE speakers to good components without changing anything else will vouch for it.

Originally Posted by vina

... a background in both signal processing and analog design tells me it is a waste of money ...

There you go again! Arrey baba, there are many products - whose comparatively very low cost will positively surprise you - which do the same (ignoring the ignorables) as that of systems that cost multiple of the car cost. And, 'poor ears' is a bad excuse. If one wants to appreciate good music, one can discover the nuances sufficiently without spending a pile. Where there is will, there is a way - knowledge of science and cost notwithstanding.

Originally Posted by vina

Twice the power will be 6dB on the sound scale. 2x here = 10x is a weird scale. ...

@cranky is trying to convey something else.

[vina]

Originally Posted by DerAlte

O @ vina bhaiya, o @vina bhaiya, at least don't make falooda out of history of sound reproduction. The quest for the best speaker possible (and hence best reproduction possible) precedes even rudimentary signal processing (RADAR / SONAR application) by decades. You are talking of the very recent phenomenon of trying to "get the timbre that I want" (like getting the timbre of valve amps using DSP), whereas the traditional approach relied (still does) on getting the timbre / sound signature right by appropriate material science and engineering. The price of a speaker reflects in some measure the recovery of the investment that went into getting it right by research.

Arre sir, I'm not trying to make falooda (though with icecreame that would't be too bad) of anything - the point is cheap speaker + DSP + very linear high power amplifier should do the same job. It may not be abough though - depends on a huge number of factors and I personally know mostly about the driver part only.

In the past good quality drivers were expensive and hard to get, DSP was not even an option. Today that is not so (I remember you talking about advancement in technology on another thread - I was referring to the same thing) cheap CMOS DSPs, and HVCMOS drivers give high power very linear drivers.

I myself designed one (100V peak output into 20ohms, linear all the way to 5MHz - diagnostic ultrasound so the measurements are not based on the fickle human ear) along with its receiver. For these applications, nobody cares about perception, you can't hear these frequencies anyway - linearity is the king (or you lose information via intermodulation) and so is SNR.

And since non-linearities still exist they use techniques like calibration, pre-distortion and equalisation.

Originally Posted by DerAlte

Err ... what exactly are you trying to get at by using the term "impedance matching"? The conventional usage of the term has only to do with efficiency of electrical power transfer, so it is difficult to relate to if you use it in the context of mechanical power transfer (cone to air; cone is more colloquially used in the context of speakers than 'diaphragm')

blame it on my background.

Almost all power transfer in the natural world happens through wave phenomenon (either mechanical waves like sound, or electromagnetic waves) - for small dimensions compared to wavelength you can approximate things as lumped components (e.g. the rigidity approximation in kinematics)

In a speaker power transfer happens at two point - coil+magnet transfers it to diaphragm (or equivalent) moving to and fro, and diaphragm via its movement transfers the energy to air.

Diaphragm + air acts as a spring-mass system (and rigid or semi-rigid diaphragm will also have their own springiness) hence the resonance at low frequencies (along with shape/size of cone - at higher frequencies further complications)

All of the mechanical resonance etc. will show up in the input impedance curve for the speaker - the resistive component of this curve (minus the resistance of the coil at DC) will correspond to the power transfer from electrical to sound, the inductive (or capacitive - if it ever gets there) component will only cause to reduce the output power.


Most drivers act as voltage sources (current limited) with very low output resistance (compared to the speaker) - impedance match is alrready non existent.

However if the impedance is varying with frequency too, then the power delivered will also vary.

Equaliser can help with this by reducing the power at frequencies where the impedance is lower (peak voltage is constrained, so equaliser will not help at high-impedance)



By the way, when I was in college, impedance matching, reflection coefficients etc. were used in many contexts (all related to wave propagation) - ultrasound, transmission lines, free space radio waves (including optical and beyond visible) ...

Originally Posted by DerAlte

Hellos, K.I.S.S. principle please!?!? Further decomposition is inimical to solving the problem.

And, contrary to what you think, changing to better speakers is the first place where you gain the most - all the people here who changed from stock OE speakers to good components without changing anything else will vouch for it.

that may be so, but then how many people actually changed the roof lining of their car for reasons of acoustics?

The context was Jinu spending time and effort in actually understanding acoustics, speakers and other details of the overall sound system (I would venture it includes his ear too) - if someone is going into that much detail then he might as well look at all the surfaces in his car too. This is like saying that get the acoustics of your concert hall fixed before you think about expensive equipment.


Of course, for most people (all people? - includes me anyway) all of that will be impossible, and the only reasonable course left is to replace the speakers.

Originally Posted by DerAlte

There you go again! Arrey baba, there are many products - whose comparatively very low cost will positively surprise you - which do the same (ignoring the ignorables) as that of systems that cost multiple of the car cost. And, 'poor ears' is a bad excuse. If one wants to appreciate good music, one can discover the nuances sufficiently without spending a pile. Where there is will, there is a way - knowledge of science and cost notwithstanding.

As for me - my ears can not detect (and mind doesn't care) what my eye can clearly see on the oscilloscope. CD quality music has a dynamic range of over 85dB, good quality compressed audio about 60-70dB but I can't distinguish with my ear alone. As a result - waste of money to spend more.

By the way, Philips did blind tests and MOST aficionados can not detect anything but loudness of music. In fact what most people call "good music" is actually pretty badly distorted - crest factor of naturally produced music can be over 20dB, crest factor of CDs in the market is between 6dB and 9dB. I think I'm aware of my ears' limitations and most people are not - that's the only difference (apart from the fact that this allows me to save cash)

there is another psychological play here. Once you know the true cost of many of the things you purchase, you may never buy stuff again. e.g. A pretty good crystal oscillator costs less than 10 cents in low volumes - I haven't bought a watch since I found that out.

Originally Posted by DerAlte

@cranky is trying to convey something else.

He clarified as much via the articles. Thanks. I read the articles after writing that post.

[navin]

Originally Posted by vina

Well I always thought speakers' job is to reproduce the input perfectly...then the cheapest of DSP processor can get whatever timber etc. you want via techniques like pre-distortion

My point is, if you want to spend more money, speakers may be the wrong place...a good environment if you figure out how to turn the situation into "man sitting inside a sound box"

If all speakers did a "perfect" job (lets assume our definitions of perfection are the same) then all speakers would sound the same. Atleast all CNO (Cost-No--Object) speakers would sound the same but a horn loaded Lowther system will sound very different from a 3-4 way tower using Accuton ceramic drivers. Ergo "perfection" is not achieveable and in fact tests have shown that a perfectly flat frequency response (for example) is not even desirable (very hard on the ears).

Yes the enviroment should also be considered but since the speaker is the one link in the audio chain that converts energy from one from to another (electrical to mechanical) it will have the most losses and hence this is where most of allocate most money to.

Lastly the man in the box is a terribel enviroment. When I was little (about 14 years old) I was small made (the paunch is a post 40s phenomenon) and actually crawled into a the back of a Marshall Stack to listen to the guitars of a Mumbai based Rock Band called Atomic Forest and it sounded terrible. I used to sit right at the foot of Ranjit Barot's (ex-Peoples) bass drum to determine the effect of direct sound vs reflected sound.

[vina]

Originally Posted by navin

If all speakers did a "perfect" job (lets assume our definitions of perfection are the same) then all speakers would sound the same. Atleast all CNO (Cost-No--Object) speakers would sound the same but a horn loaded Lowther system will sound very different from a 3-4 way tower using Accuton ceramic drivers. Ergo "perfection" is not achieveable and in fact tests have shown that a perfectly flat frequency response (for example) is not even desirable (very hard on the ears).

Yes the enviroment should also be considered but since the speaker is the one link in the audio chain that converts energy from one from to another (electrical to mechanical) it will have the most losses and hence this is where most of allocate most money to.

Lastly the man in the box is a terribel enviroment. When I was little (about 14 years old) I was small made (the paunch is a post 40s phenomenon) and actually crawled into a the back of a Marshall Stack to listen to the guitars of a Mumbai based Rock Band called Atomic Forest and it sounded terrible. I used to sit right at the foot of Ranjit Barot's (ex-Peoples) bass drum to determine the effect of direct sound vs reflected sound.

Navin

if you get an expensive speaker system, with equalisers, you can get a flat response. The all the "effects" you pay expensive fad designers for can be obtained for via DSP.


Man-in-box is not bad at all if the system is designed for that (concert hall acoustics is an example) - if you crawl into a loudspeaker and get deaf as a result that is your own doing, not even a bad design on the designers' part (they never would have thought you would do so).




Some ten years ago, a sound company approached Analog Devices and asked them to build amplifiers and drivers - they didn't know much about frequency response but they did know this - "harmony is when nothing is random, temperature = entropy = disorder = lack of harmony"

Next thing you know they ask Analog Devices that their chips must be made at Absolute Zero (i.e. 0 Kelvin).

ADI guys had a hard time explaining these fools that not only is this impossible, it'll be useless.

[cranky]

the impedance of subwoofers changes significantly with frequency. This is a case of impedance mismatch

Unfortunately it's not that.

The impedance is governed by the driver, the box and the air inside the box. It is also affected by seemingly minor things such as the shape of the baffle and how the driver is placed along it, and if the driver is countersunk or not.

If you take the exact same driver and put it in a different box, the repsonse will be different. This is why we hope that every amplifier is a perfect voltage source so it is not affected by impedance.

The voltage in your house doesn't change significantly even if you turn on every single thing because the impedance of the load is far higher than the impedance of the source. We try to achieve the same thing when designing an amplifier, specially for demanding loads like subwoofers. Some specific applications call for an amp with a high output impedance, but those are specific to the speaker and application in question.

if you get an expensive speaker system, with equalisers, you can get a flat response.

I think the problem is not about getting a 'flat' response, it's about where that response is achieved. 1 cm from the woofer cone? From the tweeter cone? 5m away? To add, no everyone's ear construction is different and 'flat' to a microphone may never be actually flat once you factor in a real room and a real listener.

The fact is that there is no theoretically perfect speaker. It is an electromechanical device, and is constrained by physics and available materials. We all know what we want from a speaker - perfect impulse response, zero distortion and coverage of every note audible to man with no deviation. This is not achievable - ever - at least not with current technology. If you study motor design (speaker motor, not the ones we usually discuss here), for example, you will realise that some of these goals in fact conflict each other. And this is before the cone has even moved.

[vina]

Originally Posted by cranky

Unfortunately it's not that.

The impedance is governed by the driver, the box and the air inside the box. It is also affected by seemingly minor things such as the shape of the baffle and how the driver is placed along it, and if the driver is countersunk or not.

If you take the exact same driver and put it in a different box, the repsonse will be different. This is why we hope that every amplifier is a perfect voltage source so it is not affected by impedance.

The voltage in your house doesn't change significantly even if you turn on every single thing because the impedance of the load is far higher than the impedance of the source. We try to achieve the same thing when designing an amplifier, specially for demanding loads like subwoofers. Some specific applications call for an amp with a high output impedance, but those are specific to the speaker and application in question.

I agree with the second part of your post (what is there to disagree?) however either I didn't understand what you wanted to say, or you are getting the concept of impedance completely wrong.

Impedance is not resistance - more precisely, impedance is not resistance only. Resistance corresponds to that part of impedance which models energy lost from the electrical network (in the form of heat, EM waves, sound ...).

For example, a speaker (or any other system) with an inductance of 0.1 Henry and resistance of 10ohms will be able to draw at best 0.1A from a perfect voltage source driving 160Hz sinusoid with 10Vrms.

The output power (neglecting heat losses) of the amplifier will be less than 0.1W as a result.

The same speaker/amplifier pair, at 32Hz on the other hand will get almost 0.4A of current, and output power will be over 1.6W.


When your speaker is near resonance, the impedance matching between its moving parts (diaphragm etc.) and air will be very poor, electrical input as a result will show a high complex (as in complex numbers, not as in complexity) impedance - limiting the current form the driver and as a result limiting the power output (in W, but not in VA).

Now changing the box, cone, even the room characteristics (for a small room with reflective surfaces for example) will influence acoustic resonance properties - and in turn the effective electrical input impedance.

[navin]

Originally Posted by vina

if you get an expensive speaker system, with equalisers, you can get a flat response. The all the "effects" you pay expensive fad designers for can be obtained for via DSP.

Originally Posted by cranky

The impedance is governed by the driver, the box and the air inside the box. ..This is why we hope that every amplifier is a perfect voltage source so it is not affected by impedance.

The fact is that there is no theoretically perfect speaker....you will realise that some of these goals in fact conflict each other. And this is before the cone has even moved.

Originally Posted by vina

Impedance is not resistance - more precisely, impedance is not resistance only.

Now changing the box, cone, even the room characteristics (for a small room with reflective surfaces for example) will influence acoustic resonance properties - and in turn the effective electrical input impedance.

1. Vina, Cranky never said impedance was only resistance or anything of that sort. I think both of you are saying the same thing.

2. The is NO perfect speaker. Period. You can ask Richard Vandsersteen, David Wilson, anyone. They will all tell you the same thing.

3. Lastly flat (as measured by measuring instruments) frequency response is not even desireable. It is vey hard on the ears. Using a DSP (or any other method) to achieve a flat response is hence redundant. Besides more important that just frequency resppnse is impulse and timbre. This is what makes one speaker prefered over another.

[vina]

Originally Posted by navin

1. Vina, Cranky never said impedance was only resistance or anything of that sort. I think both of you are saying the same thing.

that is why I wrote that I may not have understood what he was trying to say.

why I still wrote about impedance:
The example he gave of house supplies is irrelevant, and perfect voltage source. And also he wrote " This is why we hope that every amplifier is a perfect voltage source so it is not affected by impedance. " I wanted to clarify that even with a perfect voltage source he will not get flat response if the speaker goes into acoustic resonance.

Originally Posted by navin

2. The is NO perfect speaker. Period. You can ask Richard Vandsersteen, David Wilson, anyone. They will all tell you the same thing.

I don't have to ask anybody - high school physics tells me as much. When your frequency range spans 3 decades, it is not possible to have flat response. With half an octave it is very hard. That is why three speakers are used in general - and that is why even three are not enough.

Originally Posted by navin

3. Lastly flat (as measured by measuring instruments) frequency response is not even desireable. It is vey hard on the ears. Using a DSP (or any other method) to achieve a flat response is hence redundant. Besides more important that just frequency resppnse is impulse and timbre. This is what makes one speaker prefered over another.

I didn't say you can use DSP to get a flat response (though that is possible too). What I said was if you get a flat response amplifier onwards, then all the fancy stuff (timber for example, reverberations is another, echo is a third one ...) can be done in DSP. You don't have to pay any of the big names you have been dropping for anything.

To get a flat response in the first place you need caliberation tools and a good equaliser - no DSP needed (though, as I wrote above, DSP can help)



and why would a flat response not be desirable? I mean you may want your music to be adulterated by distortions, but if I want my Lata Mangeshar as she sounded at the recording studio, why the hell would I want some fad sounding overpriced fashion icon to change the frequency response of my system? And why would I later want to spend money to get a good equaliser merely to (while not even knowing this is going on) compensate for the non-flat frequency response of the speakers when the speakers are non-flat by design.

flat frequency response of the speakers and amplifiers doesn't mean "flat music" - all it means is that no specific tone is amplified more than the other.

[cranky]

To clarify - I thought *you* were talking about the electrical resistance mismatch between an amplifier and the driver's electrical bits.

When it comes to acoustic impedance, obviously it is a different ballgame and very difficult to estimate exactly - which was my point. What's worse is that it not only changes with frequency, but also level - so a sub being run at 80dB average presents a very different load than one being run at 90dB average because the air inside the box becomes a significant impedance issue.

Which as I see, you already know The point was that in all of this what you don't want is the amplifier misbehaving (this is an amp discussion, I suppose) because it couldn't cope with the load at least till a reasonable level.

[DerAlte]

Originally Posted by vina

... if you get a flat response amplifier onwards, then all the fancy stuff (timber for example, reverberations is another, echo is a third one ...) can be done in DSP. ...

@vina, at least try to be in the same world as the others in the discussion, bhai. You are getting it completely in reverse than the others in a car audio forum.

99.999% of the people who want to listen to music would find a user-controllable DSP an irritant at best. Even a majority of the DSP box manufacturers use it initially only to correct the bias imposed in the source - the OE HU - to make crappy OE speakers sound acceptable in an imperfect environment. The next most commonly used function in the DSP is to time-align the output from imperfectly positioned speakers.

You, on the other hand, seem to think that DSP is the do all - end all of the music world. A listener *does not want* a specific timbre, reverberation or echo just to be able to relate to a certain experience, for example a concert hall. If I want to listen to a Kishore Kumar song, I would like to immerse myself into the song - the lyrics, the notes, the beats ... - not to find out the difference had the song been sung on stage, even though the original recording was in studio.

THAT, my friend, is left for people who want to modify everything without getting up from in front of the PC. Why PC? PC users are thrilled to bits by the infinite controllability offered by various add-on DSP modules to modify the crappy sound that comes out from crappy equipment. And then the desire for 'modification' takes over their lives instead of music.

If you care to notice, practically none of the OTS audio mid-to-high_end equipment give such redundant facilities to users. The low-end mass-market guys do (everyone from Sony to Nippon to no-name you-know-where-it-came-from) - but usually that is marketing gimmick. Almost all users who simply want to listen to music cannot figure out whether they should like the reverb introduced or not.

K.I.S.S. principles, sir. Not everything that is theoretically possible is at all practically desirable. Navin has been trying to convey this - twice, in case you have not noticed.

[navin]

Originally Posted by vina

I didn't say you can use DSP to get a flat response (though that is possible too). What I said was if you get a flat response amplifier onwards, then all the fancy stuff (timber for example, reverberations is another, echo is a third one ...) can be done in DSP.

and why would a flat response not be desirable?..
flat frequency response of the speakers and amplifiers doesn't mean "flat music" - all it means is that no specific tone is amplified more than the other.

My understanding of DSPs is limited but what I understand is that most of these DSPs are PC based. This means that the music has to processed in a PC and that limits the music to the A/D and D/A converters available for a PC (I have no knowledge how good or bad these are as I have used only an Audigy unit that too years ago and sparingly). Hence I am not well versed to discuss DSPs but believe that (a) the current use of dedicated DSPs (like Genelec/Meridian) is limited and (b) that maybe DSP processing might have it's own limitations (for high end audio use).

As far as a flat response is concerned - a flat response while nice to see in a lab often has too much high frequency energy to make for a pleasant listening experience. This may also be because many recordings are poor equalised.

Off Topic:
Two processors I intend to get some day are the Behringer DEQ2496 and the Behringer DCX2496. The former is a DSP, the later a crossover; but both are cheap enough to be affordable by common folk like us.

Granted DSP has found great use in PA systems (JBL's Vertec for exmaple) and many companies like BSS and Rane produce DSP solutions for PA and line-array systems. However the acceptance for DSP in home audio is still limited.

[vina]

Originally Posted by navin

My understanding of DSPs is limited but what I understand is that most of these DSPs are PC based. This means that the music has to processed in a PC and that limits the music to the A/D and D/A converters available for a PC (I have no knowledge how good or bad these are as I have used only an Audigy unit that too years ago and sparingly). Hence I am not well versed to discuss DSPs but believe that (a) the current use of dedicated DSPs (like Genelec/Meridian) is limited and (b) that maybe DSP processing might have it's own limitations (for high end audio use).

DSP can be done via PC, but usually it is done via DSP processors (far cheaper, more powerful than a microprocessor). DSP for audio is much more wide spread than you would think - every single phone has it for example and so does every single Apple product.


DSP processors have no practical limitations save for the limitations of micropohone-amplifier-ADC {DSP} DAC - amplifier - speaker. In fact DSP is used to correct for a lot of imperfections in the rest of the chain. The limitations come from impossiblity of perfect calibration (especially at all power levels), different ear responding differently etc.


One limitation I have already written about is that perfect equalisation with DSP will not increase your peak power output from speaker, rather what it will do is if (for example) at 100Hz the speaker could output 70dBA, and at 50Hz it could output 65dBA - the DSP based equaliser will make it 65dBA everywhere - it can not increase the peak power, but it can decrease the output power everywhere.

This is obviously highly undesirable (especially if the limiting band is one where human hearing perception is poor to begin with, and the band sacrificed is one where human perception is sharp) - you may end up getting worse performance overall than you started with.



Make no mistakes - I was not saying that you should get flat response via DSP - I was saying you can (while sacrificing power and money)

ADC is needed for recording - not a factor for our discussion anyway. DACs today are so good (if you have money to spend, and electricity to burn) they can generate more dynamic range that human hearing.

Originally Posted by navin

As far as a flat response is concerned - a flat response while nice to see in a lab often has too much high frequency energy to make for a pleasant listening experience. This may also be because many recordings are poor equalised.

Labs spend lots of dollars to get flat response - capacitor microphones (the huge ones dangling in front of Kishore Kumar in old pictures) for example can give very flat response throughout the HiFi range (though low freqeuncy response is still not that good).

But that doesn't mean that with a system that gives flat response you have to hear music that is flat (perfectly flat and stationary audio will sound like noise anyway - because it is noise). A flat system will not make music flatter than it already is. What a flat system does is it gives the same amount of amplification to all frequencies - so it sounds exactly same as at recording.

On the other hand if a system's response is not flat, it'll add distortion that the recording studio never intended - that is not ture reproduction

What happens in studios is that they record everything with a flat response recording system BUT the sound coming from the instruments and vocal ccords already has lower energy at higher frequencies - so you do not get a flat frequency spectrum.


AFTER recording, a lot of work is done further on the sound. For example, for human conversation, crest factor (ratio of the lowest and highest sound power measured over a reasonably short time period - usually 20ms) is usually roughly 10dB. For most singers it is more (closer to 14dB) when they sing. Very accomplished singers that are renowned to have large range can have almost 20dB of crest factor while singing (from very faint to very loud).

However almost all CD (or DVD) cutting these days (unless it says "classical" on its cover) is done close to 6dB crest factor - people want loud - they get loud. Even the old songs are distorted in this way "digitally mastered" stuff is exactly this. This conversion is done using DSP at the studio.

Originally Posted by navin

... However the acceptance for DSP in home audio is still limited.

As I explained above, any audio equipment that is dealing with digital audio (and that means anything except tape and vinyl records) already has DSP, whether you like it or not. And contrary to your belief - more expensive systems have much more in DSP.

For example @cranky wrote about impedance changing with amplitude. Now this happens with all big speakers - it is called speaker non-linearity.

Let us say you somehow obtain a two-tone music file ( say 500Hz and 2KHz pure tones) and you play that file through this speaker at high volume, and the use a sensitive microphone and electronics to analyse what is coming out of the speaker.

If you did this - you'll find other frequencies (biggest power will be at 1.5kHz, next biggest at 2.5kHz) also present - this is what non-linear distortion does to you.

Now at high volume your ear may not perceive the difference - that is one way to go, or the designers of the speaker already know about this and inject a signal at 1.5kHz (and 2.5kHz) on purpose that is going to be out of phase (i.e. negative of) with the new 1.5kHz being generated - destructive interference will ensure they kill each other.


Speaker will be highly non-linear, but the DSP in the speaker driver will still allow a linear (but not necessarily flat in frequency domain) response.

[navin]

Originally Posted by vina

DSP processors have no practical limitations ...Make no mistakes - I was not saying that you should get flat response via DSP - I was saying you can (while sacrificing power and money)

Labs spend lots of dollars to get flat response - capacitor microphones (the huge ones dangling in front of Kishore Kumar in old pictures) for example can give very flat response throughout the HiFi range (though low freqeuncy response is still not that good).

But that doesn't mean that with a system that gives flat response you have to hear music that is flat (perfectly flat and stationary audio will sound like noise anyway - because it is noise). A flat system will not make music flatter than it already is. What a flat system does is it gives the same amount of amplification to all frequencies - so it sounds exactly same as at recording.

Speaker will be highly non-linear, but the DSP in the speaker driver will still allow a linear (but not necessarily flat in frequency domain) response.

Vina I dont for one second suspect that you cannot modify the response curve of a speaker or system with DSP to get exactly what suits your taste. However I also believe you can do this without DSP and it ain't that diffcult.

Sadly most recordings (especially the vintage ones) are terrible and hence are often more listenable on loudspeakers that are less "transparent".

We have digressed enough so let's get back to the topic on hand.