Thanks for the test track. Don't know about leaving it here, I'll ask some other mods. If it was a free download, should not be a problem.
I also see it as at -19dB RMS, at least the loud section at the end - a good choice. But we'll still get complaints that it isn't adequate, or not dynamic enough, or that it's not related to the -12dB track, or something. It's been a bumpy road so far.
I also see it as at -19dB RMS, at least the loud section at the end - a good choice. But we'll still get complaints that it isn't adequate, or not dynamic enough, or that it's not related to the -12dB track, or something. It's been a bumpy road so far.
and some will say that Tchaikovsky sucks.
one other very dynamic recording (in the extreme sense) is Aaron's Copland "Fanfare for the Common Man" from the Wilson Audio demo disc. I believe it's Telarc. it starts with the kettle drums and if one forgets to adjust the volume...
dynamics to make you jump out of your seat are also found on Karajan's Legendary Decca recordings. some of the best classical recordings that I've heard (but I'm not the biggest classical fan).
but I digress...
I thought about adding a test track too, something classical and very dynamic, the trouble is the copyright involved. I've got a high res 24/176.4 recording of Tchaikovsky's Hopak that was free to download from some website ages ago, I'm guessing that that would be okay? It's not too long and has some very large peaks relative to the quiet start and is rather fun to listen to.
I have converted it down to 44.1k 16bit and compressed it to a reasonably high quality FLAC and here's a download link. I also altered the level slightly so that the peak signal level of it hits 0dBfs too. How long the download will exist for I don't know
Tchaikovsky Hopak.flac
Pano may wish to download it and convert it perhaps into something different.
What , giving everyone the same recording to test ..... NOOOOOO! Sounds like a bad idea, I tell yah !!!
Don't do it , it's a simple test , stop it ....
.
you lost me on that one...You should try it push, make sure you have good batteries ......
It was free at the time, I don't know whether it was only a promotional thing, but I found the link through some 'high res' thread over on some music forums.
The thing about the track is that the first 15-20 seconds of it average around -38dB, which is where you're likely to set your volume control so that the strings are at a nice volume. Then the last 15 seconds for the climax average -17dB with the bass drum pushing the level up to almost digital zero. I think a piece of music showing almost 40dB of 'real' dynamic range is quite good.
Of course lots of kettle drums are another thing entirely and the peak in the Tchaikovsky track only happens once. I am sure we could come up with better tracks and I personally would have preferred it if the track en massé had averaged around -10dB (normall for processed pop), but I think that might be a tough challenge. I could of course record myself playing the piano and do it deliberately as to accentuate the dynamic range...but that again wouldn't have any sustained long term average levels.
Now if someone wants me to rip the audio from the final fight scene in the incredibles...that's very VERY dynamically impressive, but probably only if you've got 20hz extension
I literally had a smile on my face for the entire movie just from the way it sounded when I finally got the 'current' system together. Of course that would also be illegal.About them test tracks. In the hydrogenaudio archives (or was it r3mix?) there should be a discussion about the legality of using and distributing samples from copyrighted music. Conclusion was that you could distribute a max. 30sek? sample for codec testing(or anykind I guess?) purposes. I am not a lawyer, not to mention the ACTA/SOPA/PITA fiasco lately, so take this what it's worth - elderly-women-in-the-sauna-talk
On topic now - seems that this crowd is not into movies? Luckily the whole chain from production to playbakc is standardized, whitch guarantees 20dB dynamic range. Just get an amp so that max_no_clip = 105dB and you're good to go (well, except speaker dynamic compression and various levels of harshness). On music? Anything with more headroom than K-10 I consider a good record. Sad really.
Hi Pano
I want to say I had mixed feelings about posting about this from the beginning.
I really like the DIY area, that is my background. When I was a teen, a person I met and was mentored by, gradually put me on a different and better set of tracks in life. Some years later when I realized the impact he had, I gave him an invention I had, an acoustic levitator transducer which he then demonstrated to the new group of kids he was mentoring.
I asked what can I do to repay him, he said you will be able to answer questions, to help is what you can do for me. When Robert Oaks Jordan passed away, we lost an unsung audio pioneer but a person who changed many lives. Anyway, that’s why I post on some DIY forums occasionally and when I am working on something new, it is for me, essentially the same feeling I had playing in the sand box. I figure, most of the people who would be interested in this stuff are probably drawn to that feeling too.
I have had some time to think about “what it takes” and actually maybe the real issue is “do you have enough” for what you’re playing. Using the sine at the frequency you chose makes sense too, nothing I am talking about can be seen at the signal level without equipment most don’t have. The last thing I wanted to do was make anyone less likely to do the test or be curious.
In that vein, it has occurred to me there may be another test you could try or add, at least for people with two requirements, one, they have some ability to make an attenuator and that their amplifier is a conventional one where one side of the output is ground. The attenuator is one that lowers the amplifier output voltage down to approximately the same as the input voltage and of an input impedance that will not go up in smoke.
There is a freeware program an acquaintance has written which is for comparing music files for differences.
The idea is that it takes two files, aligns them in time and by inverting one and adding them, it adjusts to a null or residual difference. Thus, if the two signals really are identical, there is nothing left but what comes out is the difference signal. If one uses a sound card, most are identical L to R (except for latency) so this should work fairly well for this.
Audio DiffMaker
http://www.libinst.com/Detecting Differences (slides).pdf
Now loudspeakers have far too many issues to use this with but if one used the input signal and the output voltage (through an appropriate attenuator) as the other, the same null would result, the difference between them you can hear.
While playing music X at level X, one can see if the amplifier residual has spikes or transient like stuff in it which happens when the output doesn’t track the input (including things like distortion). The cool part is you can hear it, the not cool part is there are no measurements, just auralization of it via the output file.
Best,
Tom
I want to say I had mixed feelings about posting about this from the beginning.
I really like the DIY area, that is my background. When I was a teen, a person I met and was mentored by, gradually put me on a different and better set of tracks in life. Some years later when I realized the impact he had, I gave him an invention I had, an acoustic levitator transducer which he then demonstrated to the new group of kids he was mentoring.
I asked what can I do to repay him, he said you will be able to answer questions, to help is what you can do for me. When Robert Oaks Jordan passed away, we lost an unsung audio pioneer but a person who changed many lives. Anyway, that’s why I post on some DIY forums occasionally and when I am working on something new, it is for me, essentially the same feeling I had playing in the sand box. I figure, most of the people who would be interested in this stuff are probably drawn to that feeling too.
I have had some time to think about “what it takes” and actually maybe the real issue is “do you have enough” for what you’re playing. Using the sine at the frequency you chose makes sense too, nothing I am talking about can be seen at the signal level without equipment most don’t have. The last thing I wanted to do was make anyone less likely to do the test or be curious.
In that vein, it has occurred to me there may be another test you could try or add, at least for people with two requirements, one, they have some ability to make an attenuator and that their amplifier is a conventional one where one side of the output is ground. The attenuator is one that lowers the amplifier output voltage down to approximately the same as the input voltage and of an input impedance that will not go up in smoke.
There is a freeware program an acquaintance has written which is for comparing music files for differences.
The idea is that it takes two files, aligns them in time and by inverting one and adding them, it adjusts to a null or residual difference. Thus, if the two signals really are identical, there is nothing left but what comes out is the difference signal. If one uses a sound card, most are identical L to R (except for latency) so this should work fairly well for this.
Audio DiffMaker
http://www.libinst.com/Detecting Differences (slides).pdf
Now loudspeakers have far too many issues to use this with but if one used the input signal and the output voltage (through an appropriate attenuator) as the other, the same null would result, the difference between them you can hear.
While playing music X at level X, one can see if the amplifier residual has spikes or transient like stuff in it which happens when the output doesn’t track the input (including things like distortion). The cool part is you can hear it, the not cool part is there are no measurements, just auralization of it via the output file.
Best,
Tom
Tom I still don't think you've grasped the concept of how the test works, not 100% because what you're suggesting is far more complicated then Pano's test, but in the end it would end up being no more effective.
If you were going to go to the trouble of actually recording the output of ones amplifier then something like Audio DiffMaker seems like the wrong piece of software to use and a little bit more complicated then is necessary (it also seems prone to problems, especially when I tried it out). If one wanted to record the output of their amplifier then all they'd need to do is record it playing music through an appropriate resistor divider and then analyse the peaks with something like audacity. If they are clipped it will be quite obvious I'd imagine. All of this of course does require one scales down the voltage appropriately because getting it wrong could fry the input to ones sound card! I've done that a few times when I've been lazy, but in my case all I need to to is replace a 50pence opamp
If you were going to go to the trouble of actually recording the output of ones amplifier then something like Audio DiffMaker seems like the wrong piece of software to use and a little bit more complicated then is necessary (it also seems prone to problems, especially when I tried it out). If one wanted to record the output of their amplifier then all they'd need to do is record it playing music through an appropriate resistor divider and then analyse the peaks with something like audacity. If they are clipped it will be quite obvious I'd imagine. All of this of course does require one scales down the voltage appropriately because getting it wrong could fry the input to ones sound card! I've done that a few times when I've been lazy, but in my case all I need to to is replace a 50pence opamp
With the music sample, basically you listen to it and you can do it two ways.
The first way is set the volume control to the position you would set it to when normally listening to a piece of music like that. After having set the volume control play one of Pano's test tones, either the -12 or -18 will will do. Measure the amplifier voltage and then multiply it by the correct factor, 4x for the -12dB tone and 7.994x for the -18dB value. This will tell you the maximum amount of rms voltage swing you need to reproduce the peaks of a 0dB sine wave at what you could call your normal listening volume for a dynamic piece of music.
The second way is to do the same as above, but set the volume control to the loudest you would ever want to listen to the system at with that piece of music. Then repeat the above test. This time the result will tell you how much voltage swing you need for a dynamic piece of music played as loud as you're ever likely to play it.
The first way is set the volume control to the position you would set it to when normally listening to a piece of music like that. After having set the volume control play one of Pano's test tones, either the -12 or -18 will will do. Measure the amplifier voltage and then multiply it by the correct factor, 4x for the -12dB tone and 7.994x for the -18dB value. This will tell you the maximum amount of rms voltage swing you need to reproduce the peaks of a 0dB sine wave at what you could call your normal listening volume for a dynamic piece of music.
The second way is to do the same as above, but set the volume control to the loudest you would ever want to listen to the system at with that piece of music. Then repeat the above test. This time the result will tell you how much voltage swing you need for a dynamic piece of music played as loud as you're ever likely to play it.
As he says:
"While playing music X at level X, one can see if the amplifier residual has spikes or transient like stuff in it which happens when the output doesn’t track the input (including things like distortion). The cool part is you can hear it, the not cool part is there are no measurements, just auralization of it via the output file."
There are some types of amplifier distortion (crossover distortion, for instance) that are most apparent at low voltage levels of output like many here have measured, and actually become less noticeable when level is high.
The type of test Tom suggests would give you an idea of the linearity of your amplifier, and if the non-liniearities cause objectionable sound quality, as you will only hear the "garbage".
Art
I still don't understand.
how is that type of measurement more telling than using a scope/sound board, FFT analysis etc?
and how is xover distortion in the amp related to this topic?
anyway, if you wish just go ahead with the test. chances are that that you won't find anything relevant. with decent amps distortions are way more subtle than anything that would show using the run of the mill sound card and not show in THD/IMD plots.
it may be a whole different story with a very good SC like the Lynx etc and you may find some differences that are averaged out with FFT measurements.
how is that type of measurement more telling than using a scope/sound board, FFT analysis etc?
and how is xover distortion in the amp related to this topic?
anyway, if you wish just go ahead with the test. chances are that that you won't find anything relevant. with decent amps distortions are way more subtle than anything that would show using the run of the mill sound card and not show in THD/IMD plots.
it may be a whole different story with a very good SC like the Lynx etc and you may find some differences that are averaged out with FFT measurements.
Audio diffmaker isn't really suitable for comparing wave forms from which to try and figure out system distortion. It isn't accurate enough for that, at least when I tried it, it wasn't. The results kept varying quite significantly from one repeat of a test to the next and this most certainly wasn't system related.
If the levels of the recorded sound, from the amplifier, were precisely matched to the recording level of the music (I think diff maker corrects for levels anyway?) then diffmaker would start showing gross differences if the peaks were consistently clipped throughout the musical work. The trouble here is I don't know how diffmaker sets levels, it might screw it up if the peaks from the amplifier are clipped. So as Tom says you would need to precisely match the levels with your resistor divider on the output of the amplifier otherwise it could create trouble. This seems an overly complicated way to check if ones amplifier is clipping with music when reproduced at a certain volume level from a digital source.
If the levels of the recorded sound, from the amplifier, were precisely matched to the recording level of the music (I think diff maker corrects for levels anyway?) then diffmaker would start showing gross differences if the peaks were consistently clipped throughout the musical work. The trouble here is I don't know how diffmaker sets levels, it might screw it up if the peaks from the amplifier are clipped. So as Tom says you would need to precisely match the levels with your resistor divider on the output of the amplifier otherwise it could create trouble. This seems an overly complicated way to check if ones amplifier is clipping with music when reproduced at a certain volume level from a digital source.
Never have so many read so much into so little.
Why do people think this is a test of amplifier or loudspeaker quality, distortion, power supplies, dynamics, compression? It isn't any of those things, no not at all.
It's just a simple test that can be done accurately with simple tools to tell you what peak voltages you can expect.
Once you've measured, you can calculate whether or not you amp is close to clipping - that's all. Nothing more. All those other parameters are nice to know, but this test will NOT tell you that.
This was meant as a simple test to tell you what voltages to expect at your speakers. It's turned into an exorcise for reading into it things that just aren't there.
On another note. It looks like the 2-5V has pulled ahead by a nose! Where will they be at the finish line?
Why do people think this is a test of amplifier or loudspeaker quality, distortion, power supplies, dynamics, compression? It isn't any of those things, no not at all.
It's just a simple test that can be done accurately with simple tools to tell you what peak voltages you can expect.
Once you've measured, you can calculate whether or not you amp is close to clipping - that's all. Nothing more. All those other parameters are nice to know, but this test will NOT tell you that.
This was meant as a simple test to tell you what voltages to expect at your speakers. It's turned into an exorcise for reading into it things that just aren't there.
On another note. It looks like the 2-5V has pulled ahead by a nose! Where will they be at the finish line?
Pano, you seem disgruntled, Don't be. My little aspie brain has analyzed your test for any flaws and it is, for lack of a better term, perfect. (5th should enjoy that)
There will always be problems explaining how to do something. People tend to inject their own bias into instructions. That is not your fault, it's just humans communicating in our own inefficient ways.
Tom seems to want to help by adding more variables and conclusions to cover bases that are applicable to his reality. Maybe he should start 'Tom's helpful testing suggestions' thread. I'd read it.
enjoy the day,
revb.
There will always be problems explaining how to do something. People tend to inject their own bias into instructions. That is not your fault, it's just humans communicating in our own inefficient ways.
Tom seems to want to help by adding more variables and conclusions to cover bases that are applicable to his reality. Maybe he should start 'Tom's helpful testing suggestions' thread. I'd read it.
enjoy the day,
revb.
Your ears? The idea is to play it at what you feel is a realistic level on your system. I don't know where you would realistically set the SPL on this piece. Maybe 100-105dB peak, as would be heard in the good seats?
With the -12dB tone after setting your level?
From the -12dB tone, yes. That's peak, not RMS. An amp that can do 60WPC clean on a sine wave could hit your peaks. You might want a little more headroom if you play that loud often.