I have two curves, one is file1, the other file2. They plot output level vs time. I take the difference between the two curves. So I get delta vs time. From here I can compute the mean, which will constitute an estimator of the "distance" of file2 from file1. As cmp can print the differing bytes, it gives me the starting material for the calculation.

Suppose file1 is the undistorted signal (it is not because it is lossy compression in the first place, but just suppose). I'm not saying the value computed above measures distortion. But is one of several estimators. Fourier transform could also be used.

I don't know that this is necessarily true, but my belief is based on an intuitive assessment of observed speaker damage rather than testing my intuition to see if it's reproducible. One of the things that happens when you overdrive an amp is you clip the signal. This causes an unnaturally erratic signal at the speaker that I've attributed to damaged voice coils and damage to the cone either near the coil or at the web. That's simplified... there's a lot more going on... but I don't believe that your tweeters are the only concern.

I would be looking for a hardware solution rather than strictly an OS solution... something to limit the input signal to your power amp. A quick search came up with this, but there are probably commercial devices available as well. That won't prevent distortion from over driving your sound card in your speaker and introducing distortion, so maybe it would be worth using the normalization as well as putting in a hardware "safety valve".

Paul

Using an audio tool like normalize can have a bad effect on your setup if you are trying to minimize damage to your speakers when you do not know what it does and do not know how to use normalize correctly.

You may want to read the following about normalizing

http://www.hometracked.com/2008/04/2...normalization/

Read the following how tweeters or woofers can be damage.

http://sound.westhost.com/tweeters.htm

You should follow your audio path with an oscilloscope to make sure the test signal does not clip. It is best to adjust the volume of the PCM and master volume of the sound card, so the audio does not clip. Anybody using a microphone should always do a sound test. A sound test checks if their microphone volume is set at a proper volume. With out doing a sound test, the chances of clipping will be a lot higher compared to doing a sound test. You always want head room, so a scream can be done if with out hurting the speakers too much.

Equalizing causes a lot problems if there is no head room. You need the head room to equalize. If you are equalizing a frequency to +3 dB, you will need to turn down the volume by half or preferably two-thirds. If you do not do this, you will literally will have 60 watt amplifier even though it is a 30 watt amplifier. I will let you image what +6 dB of equalizing will do to your setup given that the power supply of your amplifier can handle that amount of power.

A multi-way loudspeakers uses tweeters that are under powered for the rated loudspeaker. This is fine to do because the amount watts that goes into a tweeter is only a faction of what goes into the woofer. A 4-way loudspeaker is not better than a 3-way loudspeaker. Another reason why the tweeter blows is the crossover frequency is too low or the filter order is low. For a good crossover for a tweeter, the crossover frequency should be an octave higher than the resonant frequency of the tweeter. Also the filter order should be at least 2nd order. For high output or for dance clubs, expect 4th order or 6th order. Using MOSFET for the output devices for amplifiers can have problem with some tweeter types like piezo tweeters. They will need a resister in series to minimize damage. Some tweeters may need a low pass filter at around 21 KHz to 25 KHz.

MP3 or MPEG audio is a lossy compression. WAV and CDDA are a raw audio as you can get for raw audio. If you repeatedly fix and save into an MPEG-3 file, there will be loss every time. Sure a high bit rate will minimize the effects of the audio getting worst for every save. A higher bit rate will not increase fidelity when the information is not there in the original. If you need a visual, use a JPEG file and fix and save the file several times. There will always be a loss until you can not tell what you are looking at. When you increase the DPI for a JPEG file, it is the same as increasing the bit rate. If there is no information in the original to do this, it will just put in crap and leave it at that. Of course the DPI of JPEG is 72, so it does not support any higher. If you are short on space, use FLAC or any lossless compression. IMHO, using MPEG-3 audio is the worst to compare with because what it does to audio.

I think your speakers are crap if the tweeter is easily damage by some small problems like occasionally clipping. I doubt this is a Linux issue because your speakers can easily damage using Windows or Mac. Though I do not use a pre-amplifier in my setup because my sound card outputs plenty enough of voltage for my amplifier although my amplifier has an variable gain controller to help to keep the clipping to a minimum and keep the distortion down. The amplifier that I use is in an AV receiver.

Not really, it depends on the frequency you are equalizing as well as the Q factor (bandwidth).
For example if you have a 36 band equalizer this means your Q factor will be low (1/3) and boosting any of the high frequencies by 3dB will have very little effect on the total resulting power which depends on the whole spectrum except if you are listenings to a pure sine wave.
Well, you could expect that the people who made your speakers know the basics otherwise you bought crap.

About 4-way vs 3-way (or any comparision of n-way) it is true that the ideal loudspeaker is a one-way point source but such a speaker is a myth.
You have 4-way speakers that are better then 3 or 2-way speakers while is opposite is also true.
Imho, in the end it all comes down to buy speakers that you like.
Personally I like honesty in a speaker and own a set of expensive studio monitors, but there are people that might prefer lower qualities and if they do they should buy what they like to hear even if it's cheap. Don't spend big bucks on something someon told you to buy but buy what you like to hear (don't listen to your wife who only wants a nice looking speaker where she can fit some stuff on like a nice vase).
Yes, and in the same way you could explain why you would want to raise the bitrate after having modified a audio track.
If you 'retouch' a picture you might wanna higher the quality afterwards to not loose what you did.
Keeping in mind that if you already have a high quality it wouldn't matter because you wouldn't be able to see it anyway (example a picture which is has a higher resolution then what it will be output to).
Maybe it's not speaker realted but his soundcard is poor. If the soundcard has a bad low pass filter you might get lots of crap in the high frequencies (harmonics) which can result in damaging the component.

I remember a client having troubles with some high quality studio monitors because his digital piano had a very poor low pass.
We could even hear the HF distorting on some notes.
Luckily on such speakers the HF components are always nicely overpowered and no damage was done, but we replaced them anyway and the maker of the piano paid for the replacements out of shame

Hi again:

My audio files are nearly all of them FLAC or Monkey Audio, with some very high bitrate MP3s. See this:
I would never trust this file (a FLAC file) to the normalize command or any other normalization method, unless it were highly sofisticated. So, normalization is out of the question for me.

And what are the audio components you'll be reproducing this jewell with? That's no fair question. I may have a much better hi-fi tomorrow, who knows. Well, I wanted to exclude normalization from the set of possible solutions. That's all. Regards.

P.S.: I had two Tanoi Monitor Gold. They were coaxial speakers. You would be astonished at the modesty of the boxes they came packaged in. And a single little paper inside with the specifications.

I think it depends on the fidelity of your system ... but you're right, what if you get a better system tomorrow. I dunno, maybe you could try simply not modifying the file at all. Instead play it with a media player that has an equalizer, and reduce the frequencies at which the tweeter is invoked (higher frequencies). Even mplayer has an equalizer.

I think I've been causing a good confusion here. My tweeters have break once in 12 years. Furthermore, I have two Yamaha tweeters that I intend to substitute for the present ones.

Because of this, and because it would be sinful to attenuate the highs, equalization of the recording characteristic is neither an option for me.

The true problem is all about large signals causing distortion and distortion under certain conditions putting in risk the tweeters. True, 12 years without noticeable damage, and possibility of powerful Yamaha tweeters. But still, I want to avoid risk.

If I would have known you like to listen to classical music I would have strongly discouraged you from normalizing anything.
Trying to normalize such music gives you a high chance of clipping, which is something you certainly want to avoid.

Ok, here's the only solution then: turn down the amp.

Your statements completely goes against reputable authors about equalizing audio. Equalizing any frequency about +3 dB doubles the output of the amplifier. If the speaker driver does not have enough cone excursions, it will bottom out. You can use an equalizer that have more than 36 bands and it still will have problems that I just said. You have to set where master 0 dB above where you equalize to be sure the speaker drivers does not damage from excessive amount of output.

I did not do any comparison between 4-way and 3-way which you are stating. I am just stating that people should not be too impress of a 4-way when a 3-way can do the same. A lot of manufactures goes for marketing instead of thinking about the quality of the sound. If people think a higher number for n-way is better, marketing will just go crazy to suit that thinking.

So you are saying that let us all save in a lossy format and any fixing that we do save at a higher bit rate than what we had when we got the file. This is the stupidest thing I have heard for a recommendation. When I record or doing any fixing, everything is saved in raw or WAV. Then if like how it sounds, converted into a format that people are familiar with even it will ruin my hard work when outputting into lossy like the crappy MPEG-3 audio.

Lossy compression is pieces of shit to deal with if you want fidelity. It is best to deal with the real source that did not get touched by any lossy compression. The statement of saving at a higher bit rate is just stupid. I was first on your side so you do not feel stupid stating. Now I change my mind, saving at a higher bit rate then before is with worst recommendation after fixing the lossy file to begin with.

Any sound card have this problem. Loudspeaker designers really does not take the conservative route when designing. I do this and it works well. I try to design the loudspeaker so it can handle two to three times. For subwoofer, up to five times.

I feel bad for those 'reputable authors'. Are those guys listening to pure sinewaves or white noise??
Anyone knows or should know how a constant Q equalizer works and what the repercussions are on AUDIO playback.

A typical setting for pure titanium HF would be to boost everything above 10-12kHz by far more then 3dB (sometimes up to 10dB, but this is rare) and this wouldn't be a constant Q equalising but a high shelving, even after doing this you will see only very little power difference on your amps when playing audio tracks (even in active crossed system the HF amplifier which would typically start at 1-2kHz in a 2-way system would show only a fraction of that very high frequency modification).

Seriously, just test it while playing a audio track, wath your VU and you'll see, only LF changes will have big influence on the total power.

You stated that a 4-way is not better then a 3-way and I just added that any speaker no matter how many ways is dependent on how well it's designed.

You didn't read all my posts.
I am just saying that if you get some materials in a lower rate and change it like adding more magnitude you might wanna use a better rate afterwards since compression mostly soften those transients thus making it more sensitive to quality loss. Just switching rates for sake of switching rates would be stupid.
Of course you should always have limits because at a certain degree it becomes useless.
(I explained that in a earlier post)

I hate any kind of lossy compression but having raw tracks still is expensive when it comes to storage.

A common problem with sound cards and any D/A converter is that they need a good low pass to avoid those HF harmonics which are just trash and exist in any unfiltered D/A conversion.
If you have a D/A converter you should have a low pass at least as low as half the sampling rate otherwise you will end up with tons harmonics you never wanted in the first place.

And it's not up to the speaker builder to take this into account since it's not their components that create this problem.


I have no idea why you felt being so attacked while I mostly was just elaborating what you stated and don't get why you had to attack me like that and indirectly call names.
Anyway, I will consider this topic finished since any further comments will just drive this topic into a flame war between Hi-Fi and professional audio which doesn't help the OP.

I find your disquisitions extremely interesting guys, and very didactic. Now at this point, I see the problem under better light and it splits in two.

(a) There are loudness differences among the several audio files I have. These are files I play when I want to listen to them. These differences are the lesser part of the problem.

(b) There are hundreds of little audio files played by applications in order to signal some event. Example, the desktop environment crashed. A window will popup and a certain sound will be heared. Some of these files have the signal recorded at very high levels. Now suppose you've chosen to make a halt in your work and listen The Chattanooga Choo Choo, which has been recorded at a low volume level. So you have the ALSA mixer output controls near the top, and may be your preamp volume knob in a rather advanced position. But program A is doing a job and, when A finishes the job, signals the event by a loud sound. Speakers integrity aside, it'll be very disagreable to your ears.

For me, (b) is the major part of the problem, and the only one I want to center my attention at for the time being. (a) speaks about sources (audio files and say, red book CD e.g.) which I'll call a-sources. Those little files spoken about in (b) I call them b-sources. As a first idea, I could do this: not to play any a-source except by means of an ad hoc script which first cancels output of all type-(b) audio events, then prompts the user to set his volume controls, then launches playback. Once playback is over, he prompts the user for type (b) audio events output restoration and proceeds in accordance. How do you see it?

On my part, I can say it's far from being a good solution, but I'd be glad to see it working. It of course poses a first consideration. How to cancel output of all type-(b) audio events and still have the player chosen to play the source from the script to have output. But maybe this is material for a new thread. I have yesterday started the thread http://www.linuxquestions.org/questi...-sound-850415/
The title is a bit misleading. I ask your counsel: do I go on with these matters using that thread or continue posting on this and do not post in the new one any more (SOLVED it)?

Why should I test it your way when audio is multiple of different waveforms and different amplitudes. Equalizing does cause damage if you are not careful. Equalizing can clip an amplifier making the audio worst than before. A VU meter does not measure everything. A full spectrum audio analyzer should be used to get a better idea what is going on.

You are putting words in my mouth. I said, do not waste your money on four way because it is not better.

Yes I did, but you did not read anything of my post carefully.

Normalizing is not magnitude. Magnitude has nothing to do with bit rate. Also magnitude is not the bit depth of the audio. Bit rate is the amount of information is stored at a given period which is sample rate. When a sound file is encoding into MPEG-3 audio with variable bit rate, the high frequency is monitored and using a Nyquist theory to get the sample rate.

Audio compressor not file compression does not lower quality. It lowers the threshold of the signal which is just a volume limiter. Using MPEG-3 audio repeatedly does not act like a compressor. If an MPEG-3 audio is turned into a picture and the audio file is repeatedly compress using MPEG-3 audio of the same bit rate, the quality of picture for each encode will get blocky to the point you will not be able to distinguish what was before start of the experiment.

In the old days raw storage is expensive. These days a terabyte hard drive is cheap which is as much as a computer game.

Who keeps on saying that the problem with sound cards is the they have to have low pass filter for the DAC. You do. Who has a problem with sound cards? You do. If you start paying for a good sound card, then you will not have any problems. Using a low pass filter for DAC will still have the high frequencies in the signal. They are just attenuated. Hopefully they are using at least a fourth order filter to attenuate the frequencies or else it will definitely be there in the amplifier. You do not understand. There is a lot more to sound card design than using a good low pass filter for DAC. There is the DAC it self and how the sound card is designed. Balance audio wiring should be used too. When you try to tell people to use a better sound card, they start bitching because a better sound card costs $100 to $600.

A speaker builder still have to think where the source is coming from. I take the conservative route for loudspeaker design while others take the extreme. I do not think of the high frequency harmonics and other harmonics through the range of the audio spectrum. I figure in how much equalizing and the maximum loudness it can handle. I tend to select tweeters that have the same power handeling as the woofer. I know this is over kill, but I am not govern by marketing.

You say :
Later I said :
And you respond with :
You try to deny it but then repeat it in the same sentence!!!

Nope, I repeat it differently since you do not understand. A good design loudspeaker does not require 4-way. Also it does not need a 3-way or 2-way. If it is then everybody will do it. A 3-way is good, but there are exceptions. A 3-way should be used when high volume is require because of power compression problems. IMHO, a 4-way or higher is just showing off and makes the marketing look good. If you want to say that 4-way or more-way is better, go ahead.

A 4-way requires a woofer, a mid-bass or midrange, a tweeter, and a super tweeter. A 3-way requires a woofer, a mid-bass or midrange, and a tweeter. A 3-way can easily handle the full audio spectrum up to 20 kilohertz or up to 23 kilohertz. The super tweeter used in 4-way probably can handle up to 23 kilohertz or could handle up to 30 to 40 kilohertz. You are paying more money for extra range that nobody will hear besides bugs and pets. However, a lot o companies does not put a lot of effort in multi-way systems because of marketing constraints.