What objective measurements qualify as "High Fidelity"?

[Holmz]

Let's continue with this:

Electronics (e.g., amplifiers and DACs)

1. Signal-to-noise ratio (SNR): 115 dB min. throughout specified power range
2. Total harmonic distortion (THD): -115 dB max. throughout specified power range
3. SINAD - Not important so long as SNR and THD meet thresholds
4. Intermodulation distortion (IMD): -115 dB max throughout specified power range
5. Stereo separation: -70 dB min. 20 Hz - 20 kHz throughout specified power range
6. Frequency response: 20 Hz - 20 kHz +/- 0.1 dB ( for amplifiers, throughout specified power range with a resistive load and with a load having a complex impedance - this would need to be defined).
7. Output impedance (amplifiers): 0.1 ohm max.

Speakers and speaker/subwoofer combinations

1. On-axis frequency response: 20 Hz - 20 kHz +/- 1.5 dB @ 1W/1m
2. Horizontal off-axis frequency response: -6 dB +/- 1.5 dB 500 Hz - 15 kHz @ 1W/1m @ a specific angle that is greater than 50 degrees
2. Vertical off-axis frequency response: -3 dB +/- 1.5 dB 500 Hz - 15 kHz @ 1W/1m @ a specific angle that is greater than 20 degrees
3. THD: 0.5% max 100 Hz - 20 kHz @ 96 dB; 1.0% max 40 Hz - 100 Hz @ 96 dB; 3.0% max 20 Hz - 40 Hz @ 96 dB
4. Dynamic range (compression): 0.5 dB max 20 Hz - 20 kHz comparing 102 dB to 76 dB @ 1m


Feel free to propose change/update these specifications and to propose new specifications.

What about the time domain, and transient response?
There is not mention of that in your list.

Products like DIRAC seem to be addressing it, and they seem to have people buying the stuff who claim it changes the sound of their systems.
If it is changing the sound, then we can argue whether it is better or worse, but as they are based upon some metrics, it would seem that those metrics are already extant.

 

[terryforsythe]

What about the time domain, and transient response?
There is not mention of that in your list.

Tell me what you think those specifications should be, and I will add them to the list.

EDIT: Or, copy the list in a new post and add what you think should be added. Then I will copy everything over to the first post in the thread. That will provide a place where everyone can see the current status of the specifications without having to dig through the thread.

 

[terryforsythe]

I agree - but more than not being able to pass the test... I don't believe we actually know precisely what is lacking to be able to pass that test.

I think one major issue is reproducing percussive instruments. I have never heard a recording where the drums/symbols sound real. Maybe transient response?

 

[dlaloum]

What about the time domain, and transient response?
There is not mention of that in your list.

Products like DIRAC seem to be addressing it, and they seem to have people buying the stuff who claim it changes the sound of their systems.
If it is changing the sound, then we can argue whether it is better or worse, but as they are based upon some metrics, it would seem that those metrics are already extant.

I feel like (just guessing) the time domain and transients are where the ESL57 works its magic...

And that area, isn't well defined in the usual metrics used.

 

[Holmz]

Tell me what you think those specifications should be, and I will add them to the list.

Most of the speaker measurements lately have included impulse response.
The problem is that many say that impulse response doesn’t matter.

So it is hard to get a “we” consensus on it.
From a mathematical perspective, it is difficult to claim that the output in faithfully matching the input (fidelity), when bands of teh signal are flipped in phase.

I think one major issue is reproducing percussive instruments. I have never heard a recording where the drums/symbols sound real.

Yes those are impulsive time domain signal and generally broadband.

Maybe transient response?

Probably…
You may want to try speakers that do those sounds better.

 

[Sal1950]

Electronics (e.g., amplifiers and DACs)

1. Signal-to-noise ratio (SNR): 115 dB min. throughout specified power range
2. Total harmonic distortion (THD): -115 dB max. throughout specified power range
3. SINAD - Not important so long as SNR and THD meet thresholds
4. Intermodulation distortion (IMD): -115 dB max throughout specified power range
5. Stereo separation: -70 dB min. 20 Hz - 20 kHz throughout specified power range
6. Frequency response: 20 Hz - 20 kHz +/- 0.1 dB ( for amplifiers, throughout specified power range with a resistive load and with a load having a complex impedance - this would need to be defined).
7. Output impedance (amplifiers): 0.1 ohm max.

Nice numbers but can you supply any bias controled DBT to support those as being any particular threshold?

 

[dlaloum]

I think one major issue is reproducing percussive instruments. I have never heard a recording where the drums/symbols sound real. Maybe transient response?

The ESL57 did a good job there too, within its stringent SPL and frequency limits... - I think transient response metrics may well be key. (Quad did advertise their speakers ability to "pass a square wave".... which is directly related to transients...)

 

[terryforsythe]

Most of the speaker measurements lately have included impulse response.
The problem is that many say that impulse response doesn’t matter.

So it is hard to get a “we” consensus on it.
From a mathematical perspective, it is difficult to claim that the output in faithfully matching the input (fidelity), when bands of teh signal are flipped in phase.

I think impulse response probably does matter.

Once we post it people can discuss it. If no consensus is reached, we can always take a vote on it.

 

[terryforsythe]

Nice numbers but can you supply any bias controled DBT to support those as being any particular threshold?

The distortion and noise values I took from some of Amir's statements on those subjects. The rest of the values I chose based on how I think an excellent system would perform.

But, I am not set on any of the values. I just wanted to get something down as a starting point. I fully expect all of the specifications to change with more input.

 

[Holmz]

I think impulse response probably does matter.

Once we post it people can discuss it. If no consensus is reached, we can always take a vote on it.

I am not inclined to vote on something that is mathematically provable.
It is not a consensus that I want to participate in.
It is like voting on whether relativity or gravity exist, or whether the Earth is flat.

We can look at the input signal and the acoustic output, and compare them to come up with some %age of agreement or %age of distortion.
Just, as shown in your list… many people do not seem to care. (They do not even make the list of things to talk about.)
It is all solely back to, ”within +/- X dB from 20-20kHz”, and SINAD —> infinity..

It is basically a bunch of compromises in a speaker system… and people trade away cost, frequency, and time domain response to achieve some holistic “speaker”.
Sometimes those work well, but end up being difficult loads to drive.

And we have not even gotten around to the radiation pattern and the room.
The room and radiation pattern mostly matter as they are part of the implemented system, but they are only connected together at the end user.

 

[Sal1950]

But, I am not set on any of the values. I just wanted to get something down as a starting point. I fully expect all of the specifications to change with more input.

OK, but if you intent to build a database on "what is High Fidelity" you have to be able to accurately and reliably predict what we can or can't hear. Just throwing up some arbitrary numbers will answer nothing. I know there have been many studies over the decades but this is your baby and the homework is up to you. Outside of the lunatic fringe, it's pretty well accepted that the engineering of audio electronics has been a solved problem for a number of decades when properly executed.

 

[terryforsythe]

I am not inclined to vote on something that is mathematically provable.
It is not a consensus that I want to participate in.
It is like voting on whether relativity or gravity exist, or whether the Earth is flat.

OK. Let's scratch the consensus and vote, and leave it to science.

We can look at the input signal and the acoustic output, and compare them to come up with some %age of agreement or %age of distortion.

Sounds good to me.

And we have not even gotten around to the radiation pattern and the room.

I touched on the dispersion pattern, but if that needs to be expanded, let's do it.

One goal is for people to be able to recognize what equipment meets "high fidelity" specifications when shopping for equipment.

As far as the room, that would be a different section we can add to guide people on proper setup.

 

[terryforsythe]

Outside of the lunatic fringe, it's pretty well accepted that the engineering of audio electronics has been a solved problem for a number of decades when properly executed.

The engineering of electronics may be solved, but not all electronics manufacturers properly implement those solutions. There is much confusion, even in this forum.

 

[Holmz]

…
I touched on the dispersion pattern, but if that needs to be expanded, let's do it.

That sorta depends on the room. If I am putting speaker in a tiled room, I probably want a narrow radiation.
And if it is one person, then narrower is fine.

One goal is for people to be able to recognize what equipment meets "high fidelity" specifications when shopping for equipment.

It is not my job to educate people.
They go in hear a bright, screechy, and exciting speaker and buy it.
Then later find that they are fatigued listening to it.
If I was in the sales or marketing department, then maybe I would care.

And the large majority just like their ear bids and the music that way.
They are my enemy as they are the ones that the loudness wars catered to.

As far as the room, that would be a different section we can add to guide people on proper setup.

Why?
If they are interested they will find the info from somewhere.

I cannot drag someone by the scruff of the neck and make them listen.
Even when people are on the same team in a work environment, it is hard to get a consensus and common direction.

 

[terryforsythe]

Just throwing up some arbitrary numbers will answer nothing.

The numbers are not arbitrary.

 

[terryforsythe]

It is not my job to educate people.

I would like your input if you have expertise in this subject. But if you don't want to contribute, that's fine.

 

[dlaloum]

OK, but if you intent to build a database on "what is High Fidelity" you have to be able to accurately and reliably predict what we can or can't hear. Just throwing up some arbitrary numbers will answer nothing. I know there have been many studies over the decades but this is your baby and the homework is up to you. Outside of the lunatic fringe, it's pretty well accepted that the engineering of audio electronics has been a solved problem for a number of decades when properly executed.

electronics yes - transducers ... no!

 

[Holmz]

I would like your input if you have expertise in this subject. But if you don't want to contribute, that's fine.

What’s the point?
Are we going to solve the problem of what people should get?
Salesmen do that.
Or is this a savour complex, or Quixotic approach to helping others?

It is like trying to get everyone to drink a Malbec because they are a good value grapes.
When some people just like Pinot or Cab.

If I was able to understand your purpose or reasoning, it would help me.

 

[Holmz]

electronics yes - transducers ... no!

Transducers themselves are tractable.

It is the speaker that starts to get spiced up and becomes intractable.

• Cabinet
  • Resonances
  • Size
  • Bracing
  • Baffle steps
• XO-order
• Box order
• XO implementation
• etc.

 

[dlaloum]

Transducers themselves are tractable.

It is the speaker that starts to get spiced up and becomes intractable.

• Cabinet
  • Resonances
  • Size
  • Bracing
  • Baffle steps
• XO-order
• Box order
• XO implementation
• etc.

Yes - I was being imprecise using the term "transducers"...