• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Scientific Measurements vs Real World Usage

piercer

New Member
Joined
Apr 13, 2024
Messages
4
Likes
9
Hi all,

I am a very scientific person and really do get the fact that DACs with similar noise measurements below 120db aught to sound the same.

But I had a thought...

The measurements on this forum seem to be very much about what is produced when an ideal signal is fed in - e.g. a 1KHz sine wave. Its good to have a reference point but what happens when you feed a real musical signal in? Do different DACs respond differently to different frequency mixes and sudden changes to the frequency content of a signal? For example, are there any tests that have a baseline piece of music that gets fed in and then the output waveform is measured and compared - If you had very similar measuring DACs from the sine wave perspective would they all end up with identical wave forms coming out when actually playing music?

Thanks for any information on this.

Conrad
 
The measurements on this forum seem to be very much about what is produced when an ideal signal is fed
Here's Amir's Multitone signal, before FFT analysis:
Multitone waveform.png
 
Hi all,

I am a very scientific person and really do get the fact that DACs with similar noise measurements below 120db aught to sound the same.

But I had a thought...

The measurements on this forum seem to be very much about what is produced when an ideal signal is fed in - e.g. a 1KHz sine wave. Its good to have a reference point but what happens when you feed a real musical signal in? Do different DACs respond differently to different frequency mixes and sudden changes to the frequency content of a signal? For example, are there any tests that have a baseline piece of music that gets fed in and then the output waveform is measured and compared - If you had very similar measuring DACs from the sine wave perspective would they all end up with identical wave forms coming out when actually playing music?

Thanks for any information on this.

Conrad

First, a variety of test signals are used, not just 1kHz sine. Second, ANY arbitrary waveform is a combination of sine waves. So yes, for any half-decent measuring DAC, the output music waveforms will be practically identical.
 
Thats what multitone tests are for. The test signal is composed of 20? sinewaves across the audible spectrum, which is similar to a complex music signal.

Edit: Looks like Im slow on the draw today.
 
There’s usually good correlation between other tests and the multitone results, which is a generalized representation of actual music material.
 

The Sony AVR has a bigger drop from 1 kHz tests compared to Multitone versus the Yamaha AVR. A lot of the dedicated DACs do a great job on both tests. The Sony example is one that shows the limits of 1 kHz as the only test signal. Of course, this is “known” based upon the simple statement: If you only needed the first graph, people/Amir wouldn’t take the time to show all of the other graphs…
 
Seems to me a very scientific person would... never mind maybe read this or not
 
To add: Simple tests like the 1khz tone and/or multitone are considered suitable for evaluating music performance because music itself can be expressed as a series of pure tones - you may have heard of the Fourier Theorem and/or Fourier Transform. Not only can you recreate music mathematically out of pure tones, signals actually behave this way.

So the logic goes, if the equipment can handle a pure tone, and if it can handle several pure tones across a wide bandwidth, it can handle music, because that's a mathematically accurate description of what music is.
 
Why don't you look at the sweeps and multitone? Far more useful ,you can can see what they do where the music bangs (look at -30dB where the main bulk is up to -15dB for well recorded music) ,IMD vs Level for more tones,Multitone32,etc.
 
Perform subjective measurements first. If a difference in sound is determined to exist, and the 1 KHz technical measurements don't show a difference, then a different technical test would need to be devised to account for the difference in sound.
 
If you had very similar measuring DACs from the sine wave perspective would they all end up with identical wave forms coming out when actually playing music?

Oh yeah, I did this with real music too:

No difference.

Yes, difference
 
I think we can be kind in our interpretation, it is a common thought for a beginner. I know because I had it too just a few months ago. Its reasonable to want to incorporate 'real world usage' in to the lab tests, it is just a shame it requires some knowledge on signals to understand why a multitone test accurately precludes this need.
Since it is not necessarily obvious to beginners, then we will continue to see this sentiment.
 
Oh yeah, I did this with real music too:

No difference.

Yes, difference
Discussion didnt seem to reach a conclusion on that, did it?
 
To add: Simple tests like the 1khz tone and/or multitone are considered suitable for evaluating music performance because music itself can be expressed as a series of pure tones - you may have heard of the Fourier Theorem and/or Fourier Transform. Not only can you recreate music mathematically out of pure tones, signals actually behave this way.

This is true for signals and it is also true for transfer functions of linear, time-invariant systems. However, DACs are not LTI systems. Hence, multi-tone test signals are important in order to characterise their transfer function a bit better and closer to how they behave with complex signals.

I'd wish for frequency response plots, meaning both amplitude and phase response. They may not reveal much of concern, but we just don't know, because nobody ever plots the phase response.
 
This is true for signals and it is also true for transfer functions of linear, time-invariant systems. However, DACs are not LTI systems. Hence, multi-tone test signals are important in order to characterise their transfer function a bit better and closer to how they behave with complex signals.

I'd wish for frequency response plots, meaning both amplitude and phase response. They may not reveal much of concern, but we just don't know, because nobody ever plots the phase response.
You can do that with some of pkane's software. Most have some things going on above 10 khz with regards to phase.
 
You can do that with some of pkane's software.

I'm sure even a lowly AP Analyser could do it :p

It just baffles me (somewhat) that no reviewer ever shows it.
 
Discussion didnt seem to reach a conclusion on that, did it?

My ears, my measurements.

We cannot prove that it’s not simple correlation but it is interesting to see that the differences I heard correlated with the measurements. I still encourage those who believe there are differences to try to quantify them with measurements themselves.

But it answers the questions that the original poster asked. Here’s some data showing how two products with “close” 1 kHz test tones can different greatly on multitone testing. It probably doesn’t happen with 120 dB testing.

I also show what happens when you measure the output of two DACs that should be well past the point of diminishing returns.

I have even done my ABX testing of CD and DSD layers.

At the end of the day though, I say: put all your money in speakers and then DSP.

Not just internet arguing or debating — I listened and measured my own gear.
 
musical signal
As the name implies, a multitone stimulus signal consists of multiple sine waves (or tones) at different frequencies combined together. Any number of tones can be used, but 3 to 30 tones spaced logarithmically across the audio band is typical. This feature alone offers several advantages over traditional single tone testing:

It allows the performance of a DUT to be evaluated over a range of frequencies with a single measurement. The characteristics of a multitone waveform (e.g., spectral content, histogram, crest factor, etc.) give it a much closer resemblance to typical audio program material like music or speech, than a single sine wave. Multitone distortion is generally better at detecting real-world problems involving clock jitter and sample rate conversion than traditional THD+N measurements.


JSmith
 
This is true for signals and it is also true for transfer functions of linear, time-invariant systems. However, DACs are not LTI systems. Hence, multi-tone test signals are important in order to characterise their transfer function a bit better and closer to how they behave with complex signals.

I'd wish for frequency response plots, meaning both amplitude and phase response. They may not reveal much of concern, but we just don't know, because nobody ever plots the phase response.
This is the kind of thing I was thinking of. I am well aware that all waves can be represented as a sum of sin waves but was wondering what impact the time varying nature of such a sum would have on a DAC and how phase shifting and 'smearing' may only be apparent when such things are considered and measured.

TBH I was unaware of the multi-tone test and will look into that.
 
Back
Top Bottom