Sound Quality Vs. Measurements

Status
Not open for further replies.
Administrator
Joined 2004
Paid Member
Hi Andrea,
Some type of audio analyzer, Sound Technology would be the most common. Keysight, RS are others. I use an excellent RTX 6001 designed by Jens on this forum. I think I will upgrade the power supply. For software I use ARTA and M.I.

Additionally, you would need a really good distortion meter and a good oscilloscope, and I also use spectrum analyzers. A low distortion oscillator or two and a wide band AC voltmeter too. I just got an old HP 3400A, but normally use a Keysight 34465A DVM. You would need a really good DVM anyway to set bias currents, and I often use a Keysight 34461A for that, I used to use HP 34401A - an excellent meter still more than good enough for this.

I just got a pair of Klipsch THX 6000 speakers I use on the bench, and my main system uses a pair of PSB Stratus Gold (original). Not the best you can get, but very competent and reasonable.

Additionally, I have a ton of test equipment used for assessing capacitors and resistors, so three LCR meters, some DVMs god to 7 1/5 digits and special jigs for matching and grading transistors. Power supplies, an AC power supply (Keysight 6812C), frequency counters, oscillator analyzer (HP 6372A), RF generators and FM stereo generators, Jitter meters for tape and CD, laser power meters, GPS disciplined oscillators for the bench 10 MHz reference clock, network analyzer, current probes for DC and AC. Plus, test equipment to check my test equipment.

I constantly upgrade and improve my equipment as the chance presents itself. In the last two years, I have spent over $50,000 in upgraded test equipment. I wouldn't do that if it wasn't needed. partially education as well as I experiment with circuits and components. I won't be spending that much again in the near future!

-Chris
 
Hi Chris,

it looks like me and my co-developer we own such tools:
- Quant Asylum QA401 Audio Analyzer (not much different than the RTX 6001)
- Keithley 2001 DMM (7 1/2 digits AC 20Hz - 2MHz)
- National Instrument GPIB-USB-HS
- Lecroy WavePro 954 oscilloscope (1 GHz bandwidth 2 GS/s sample rate 8 GS/s single channel mode)
- Lecroy HPF1000 fet probe
- DG8SAQ USB-Controlled Vector Network Analyzer
- Advantest R3132 Spectrum Analyzer
- Clio Pocket measurement system
- TimePod 5330A High-Performance Phase Noise and Allan Deviation Test Set (now Micosemi 5120A)
- several MTI-Milliren 5MHz reference OCXO
- DRIXO 5.6448 MHz reference oscillator (one of the best available on the whole market, in class with Wenzel BT ULN and Oscilloquartz BVA)
- several LCR meters, Function generators, Lab power supplies and so on

Well, we are merely hobbyist.

Then I wonder which audio companies don't own such cheap tools, I don't know any.

I'm really curious, which audio companies were you referring to?

Andrea
 
Jam was given a hearing test by a prospective high end audio company boss. It was just a few years ago. Various audio components were switched around to make a new system configuration. Jam was the only one who could name each component change by ear.

I'm not sure that counts as a hearing test, they are usually conducted by audiologists not someone in the high end audio business. I suppose if the components themselves weren't high end then it might have some merit considering what some think to be the performance level of high end products.
 
Yes, I took my elderly mother to an audiologist to get her hearing aids adjusted. After the test was over we went to his office to discuss options. I picked his brains for every technical nugget I could get. Also, read some published research on hearing aid technology development.

What those guys normally measure is one thing: Threshold SPL at various frequencies. They then use hearing aids with digital multiband compressors to bring all frequencies up to their threshold SPLs without ever going too loud.

Some hearing aids also communicate with each other over bluetooth to synchronize directional cues related to arrival time.

For people with very severe hearing loss where it is not possible to raise SPL high enough without causing further hearing damage, what works for some people is for the hearing aids to have a digital pitch shifter to shift all incoming speech related audio frequencies into the frequency band which the patient can still hear. It doesn't sound like real speech at all after that, but some people's brains can adapt to understanding spoken language from that very distorted signal.

As can be seen, little of it has to do with judging sound quality. Also, unless hearing loss is severe, turning up the volume can reach threshold SPL at most frequencies.
 
Last edited:
<snip>

The time he gave for sound memory was about 15 seconds.

Often it depends on the definition of terms, the echoic memory - which is the auditory memory essentially - has a storage time of a couple of seconds, surely less than 15 seconds. The terms depend on the memory model people are using, if the acoustical information is processed then it is transferred to other parts of our memory and lasts longer.

Otoh music seems to have a more direct path to longer lasting memory even without conscious processing.

If the mentioned 15 seconds were the relevant time span when evaluating audio reproductions, it wouldn't make any sense to chase for higher quality, as we would have forgotten when listening the next time.
 
The basis for the so-called "great debate" are the known thresholds of hearing. Through the decades a certain group of people asserted that the reproduction devices (despite loudspeakers and rooms) were already better than human listener's hearing abilities.

If listening experiments had shown that the participants couldn't detect THDs below a certain number and other experiments have shown that the participants couldn't detect linear distortion below a specific number and could not detect frequencies above a specific bandwidth (just to name some parameters), then it led to the conclusion that every device with a measured performance better than the threshold in each category, must result in a reproduction that could not get any better.

Therefore it was concluded that audible differences between amplifiers was a matter of the past since the mid seventies; audible differences between CD-players should have been non existent right from the beginning.
(For example, a lot of people were very upset when the first german audio magazine began to describe audible differences between CD-players)

Although the hypothesis based on the thresholds seemed to be plausible, it neglected the sometimes quite profound span in listener abilities, as the threshold number is an estimation of the mean and the distribution spreads to both sides (a physiological lower bound exists, though) and the question if results from single parameter experiments could be extrapolated to evaluations where a lot of parameters are varied at once.

Next step led to the discussion about proper experimental procedures for sound listening tests ...... :)
 
Would it help if I bring in a tube pre-amp, to increase H2 and H4 alone at 2% to 4% distortion? I have also tubes for preamp which can give more H2 and fairly equal H3 and H4.

Ionmw,
thank you for sharing your thoughts.
Personally, I think there`s something inexplicable about the H13.

To keep distortion down, choose recordings where just one instrument is playing. Large orchestras have a propensity to create an interharmonic mess.
 
What those guys normally measure is one thing: Threshold SPL at various frequencies.

As can be seen, little of it has to do with judging sound quality.

Considering flat(ish) frequency response is one of the prime requisites to sound quality then a flattish hearing response would be required to judge it properly. Doesn't Floyd Toole use young people with good hearing to conduct his tests?
 
Could you please look back at my proposed plots in Post 20101 (page 2011)? ...
I am sorry Ionmw, I have looked at that post but still can not figure the head and tail of what is being plotted. Looks like Amp input vs output, but can not derive the idea about a flat response from them. Furthermore, I'd expect an amp with +/- 40V supply should easily deliver 75W into 8 ohm, why would you need +/- 80V to get 60W? And what problem do you have that you expect an addition of 2-4% H2 can help?
 

And once again, the discussion was masterfully shifted from answering the question "Can, with a certain confidence level, a human hear X" to "if not, it doesn't mean that there isn't somebody that can hear X" (without identifying these individuals or group). And of course, questions about the relevance of the existence of such individuals or group will always remain not answered.

This false dichotomy and faulty logic, deserves the generic name of "FUD", since it is promoted solely to instill a sense of doubt among the audio enthusiasts (lest skilled in the meanders of formal logic, and with little to no appetite for nitpicking) when listening to qualified voices of reason.
 
Considering flat(ish) frequency response is one of the prime requisites to sound quality then a flattish hearing response would be required to judge it properly...

Maybe so if one were tasked with judging frequency response. Trivially easy to measure FR, so no need to use a human for that task. Try picking a task humans are good at, perhaps such as whether human voices sound nicely musical in a way that most people will probably like; how do you measure that with an AP, given it doesn't have a test for that?

Trust me, measuring FR, THD, DNR, etc., won't tell you everything that matters to human perception. Its a myth that they those performance metric classifications cover everything that matters. But we may think they do because we have not fully conceptualized the problem space. We only think in terms of the mental models we already have. WYSIATI. WYSIATI | Jeffrey Saltzman's Blog
 
Last edited:

TNT

Member
Joined 2003
Paid Member
Maybe so if one were tasked with judging frequency response. Trivially easy to measure FR, so no need to use a human for that task. Try picking a task humans are good at, perhaps such as whether human voices sound nicely musical in a way that most people will probably like; how do you measure that with an AP, given it doesn't have a test for that?[/url]

Do you find the people you communicate in daily life having musical voices?

Would you like to add or subtract something to them while trying to reproduce them in an audio system? No? Added or subtracted tings show up in an AP as ..... added or subtracted "things" ;)

//
 
Status
Not open for further replies.