Why you can't trust audio measurements

[andyr]

11 minutes ago, March Audio said:

 

Oh I'm quite happy to make recordings played through this device for everyone to listen to.

 

 

If you're allowed to do that without infringing copyright - go right ahead.  But you would need to post 2 recordings - one with the 'black box' in play ... and one without it in the signal chain (just so people can make the comparison).

 

11 minutes ago, March Audio said:

Just to be clear, this isn't some kind of compression device is it?  Or something with active gain control?

 

I don't know what a 'compression device' is, Alan.  But it certainly doesn't have active gain control.

 

11 minutes ago, March Audio said:

We are specifically testing the effects of bandwidth and only bandwidth.

 

I don't understand the words you've written, Alan.

 

AFAIAC ... you will be confirming that the device does soften the leading edge of transients.  I have no idea how this relates to bandwidth - this was a concept raised by you/Grant.

 

[March Audio]

14 minutes ago, andyr said:

 

If you're allowed to do that without infringing copyright - go right ahead.  But you would need to post 2 recordings - one with the 'black box' in play ... and one without it in the signal chain (just so people can make the comparison).

 

 

I don't know what a 'compression device' is, Alan.  But it certainly doesn't have active gain control.

 

 

I don't understand the words you've written, Alan.

 

AFAIAC ... you will be confirming that the device does soften the leading edge of transients.  I have no idea how this relates to bandwidth - this was a concept raised by you/Grant.

 

Andy,

 

You have been specifically talking about bandwidth and the effects of bandwidth on in band transient response.  You interjected on Grant and my conversation.

 

For you to now claim you dont understand what I have written indicates you didnt understand what you were saying in the first place

 

Before I waste any any time on this, why dont you provide a description of how the device works and a circuit diagram.

 

Dont tell me, it adds copious amounts of specific distortion?

Edited November 19, 2022 by March Audio

[POV]

So after all that, can someone please explain in plain language what the leading edge of a transient is, and how one could tell the difference via listening between a dull edge and a sharp edge?

[andyr]

27 minutes ago, POV said:

... and how one could tell the difference via listening between a dull edge and a sharp edge?

 

If you can understand how a picture looks when it's been taken by a real camera (which has lens-focussing capabilities - so not a phone camera) when the lens is in perfect focus ... vs. when it is slightly out of focus, I would've thought you should be able to understand the difference between sharper delivery of the leading edge of a transient ... and less sharp.

 

And btw - apart from looking at a pic that's been taken - how does one measure the degree of out of focus-ness of a picture?

 

[andyr]

20 minutes ago, March Audio said:

 

So Andy is muddling up several things.  He has said that increasing bandwidth beyond 20kHz increases in band (below 20kHz - the stuff you can actually hear) transient speed.  It doesnt.

 

We will now probably have to endure a page of "i didnt say this or that" from him

 

 

Good one, Al. 

 

Please show me the post where I said "increasing bandwidth beyond 20kHz increases in band (below 20kHz) transient speed".

 

[March Audio]

1 hour ago, andyr said:

Alan - I have never mentioned 'bandwidth' in my posts

 

Except of course you did:

 

"I am aware of this - for instance, cartridges which have a bandwidth of, say, 50kHz rather than 20kHz ... have a faster rise time and deliver (the leading edge of) transients better."

 

Which by definition you are talking about audible sounds which are below 20kHz

 

@sir sanders zingmore asked us to be respectful of each other, but its really difficult when all you seem to do Andy is play silly buggers.

Edited November 19, 2022 by March Audio

[El Tel]

It's getting harder and harder to do any testing from suggestions on this thread. It's a real shame as I felt we might be getting to a mutually acceptable approach for a couple of areas of interest.

 

Firstly a great suggestion to blind test a switch and it seems Alan has been ghosted on that offer, and now a different gizmo that surfaced and has now been snuffed out.

 

For the first time in a long time I thought we may have a really great opportunity for people to come together and get to the heart of the matter on a couple of things.

 

It seems we are destined for division rather than honest and frank exchanges based on shared experiences.

 

Such a shame.

[rocky500]

1 hour ago, POV said:

So after all that, can someone please explain in plain language what the leading edge of a transient is, and how one could tell the difference via listening between a dull edge and a sharp edge?

Found something here that goes into a bit of detail & how they describe it

https://joeysturgistones.com/blogs/learn/what-are-transients-why-do-they-matter-to-your-mix

 

Edited November 19, 2022 by rocky500

[stereo coffee]

1 hour ago, March Audio said:

 

You have two concepts here:

 

1. A technical definition

2. An individual subjective interpretation of a sound characteristic using ill defined language.

 

Technically it is how fast the leading edge of a signal rises.  A sine wave is slow compared to a square wave for example.  A low frequency is slow compared to a high frequency.  How fast a signal can rise is dictated by the system bandwidth.

 

 

 

 

When really, could I say,   its a vast number of other criteria, all contributing as well,such as rectifier design, removing ripple from supplies as well as listening intently to how the circuit responds to feedback, choice of semiconductor devices   minimising reactance and presenting a uncomplicated signal path without out of band distortions or circuit distortions. Lots of criteria to keep the midnight candle burning.... and then it has to work - in case i forgot that 

 

 

  

  

[sir sanders zingmore]

1 hour ago, March Audio said:

 

 

 

@sir sanders zingmore asked us to be respectful of each other, but its really difficult when all you seem to do Andy is play silly buggers.

 

Difficult or not, you need to (remain respectful, that is).

Blaming others won't cut it, sorry

[March Audio]

2 minutes ago, sir sanders zingmore said:

 

Difficult or not, you need to (remain respectful, that is).

Blaming others won't cut it, sorry

Dont you think its disrespectful when other certain members use semantics and games just to argue without contributing to the thread?

[MLXXX]

On 18/11/2022 at 4:20 PM, andyr said:

You can take a picture - of a face, say - using  makes the notes more clearly definedthe best lens it's possible to buy ... and you will see a very detailed picture of that face; every blemish, enlarged pore etc. ... will be clearly displayed.

 

Or ... you can put a gauze filter over the lens - which softens the image you see; ie. it doesn't have the same clarity as when there is no gauze filter.

 

That's what I meant when I wrote "delivers the leading edge of a transient better".  The notes are more clearly defined.

 

2 hours ago, andyr said:

If you can understand how a picture looks when it's been taken by a real camera (which has lens-focussing capabilities - so not a phone camera) when the lens is in perfect focus ... vs. when it is slightly out of focus, I would've thought you should be able to understand the difference between sharper delivery of the leading edge of a transient ... and less sharp.

 

And btw - apart from looking at a pic that's been taken - how does one measure the degree of out of focus-ness of a picture?

 

The photographic situation can be measured as lines of resolution visible in a test pattern that includes broadly spaced lines and progressively more closely spaced lines (e.g. so many lines per inch or per centimetre).  The resolution achieved can be assessed by eye as was done in the early days of analogue television by looking at a test pattern displayed on a CRT screen.

 

Theses days with digital images (including videos consisting of a series of still images captured at a particular frame rate) we can talk of the OTF (optical transfer function). The OTF can be used to describe the extent of contrast achieved for picture content that alternates between black and white at various different spatial repetition rates.   (See the Wikipedia description of OTF at https://en.wikipedia.org/wiki/Optical_transfer_function . )

 

To achieve a very detailed image (be it a test pattern or of a real world scene) you need either:

(1) An inherently higher frequency of spatial response, which naturally gives high contrast for small details. (So, a camera with many megapixels of sensor resolution, a high quality lens, an accurate focus adjustment for the subject being photographed, and a file compression protocol that does not lose too much detail.)   Or,

(2) Digital enhancement (exaggeration) of edges in the image, to boost the subjective contrast where the lens has introduced blur through its imperfections, or a poor focus setting,  or because the optical sensor has an insufficient density of photoreceptors, or where a digital file compression algorithm has been very aggressive.  Too much of this kind of image enhancement will look harsh and unnatural.

 

With audio, you can achieve a very basic equivalent of (2) above with treble boost.  So if your system is struggling to reproduce 20kHz you can add treble lift to try to compensate for that.  However there will be a limit to how much treble can be clawed back in that way.

 

If you want an optically softer image you can reduce the OTF with an algorithm that imposes an averaging of the values of adjacent pixels, or by simply rescaling to a lower resoution pixel grid (e.g. 640 x 360 instead of 1920 x 1080). With audio for a more mellow sound, you can apply cut to the treble.

 

*  *  *

 

An important basic concept to understand with audio transients is that they cannot exceed the bandwidth of the reproduction system. Conventional CDs are only good to about 20kHz (or slightly higher) because of the need for anti-aliasing filtering. Any waveform shape at a repetition rate of 20kHz that rises of falls more steeply than a 20kHz sine wave will be rounded back to the rise and fall times of a 20kHz sine wave. (So, for example, something that started out as a 20kHz square wave would end up as a 20kHz sine wave when played back with a conventional CD player.)

 

Even with vinyl discs the upper frequencies are rolled off to avoid cutting a groove with modulation excursions too great in amplitude, and too rapid,  for the playback stylus to track safely.  So even if a cartridge is measured to be capable of responding to small excursions at say 30kHz cut into a specialist test disc, a production pressing for music typically will not include any such frequency content, or if it is present it will be at a very low level.  A practical implication is that a very fast rise time transient cut into a vinyl disc is very likely to be limited in its rise time, or "rounded", to avoid stylus tracking issues when the disc is played back. The full theoretical upper frequency limit of the vinyl disc will not be utilised. Of course even if it were utilised, it would not be expected to be heard as a difference.  To [young adult] human ears a 20kHz sine wave is a thin tone. There is no timbre because we cannot hear any overtones of 20kHz. A square wave [produced with an ultrasonic speaker] would still sound like a thin pure 20Khz sine wave, to a young adult human listener.

 

Edited November 19, 2022 by MLXXX

[aussievintage]

9 minutes ago, sir sanders zingmore said:

 

Difficult or not, you need to (remain respectful, that is).

Blaming others won't cut it, sorry

 

and yet you allowed a post of Andy's where he cast aspersions on Alan's character saying he would misrepresent the test results?????

[sir sanders zingmore]

14 minutes ago, aussievintage said:

 

and yet you allowed a post of Andy's where he cast aspersions on Alan's character saying he would misrepresent the test results?????

 

It slipped through ... it's been removed

[sir sanders zingmore]

22 minutes ago, March Audio said:

Dont you think its disrespectful when other certain members use semantics and games just to argue without contributing to the thread?

 

I repeat (for the last time)

Difficult or not, you need to (remain respectful, that is).

Blaming others won't cut it, sorry

 

 

[andyr]

1 hour ago, March Audio said:

 

Except of course you did:

 

"I am aware of this - for instance, cartridges which have a bandwidth of, say, 50kHz rather than 20kHz ... have a faster rise time and deliver (the leading edge of) transients better."

 

Which by definition you are talking about audible sounds which are below 20kHz

 

 

Yes, I did say that, Alan.  It was an (obviously vain) attempt to explain to someone what "delivering the leading edge of transients better" was.

 

You focussed on that post ... rather than on earlier ones of mine.

 

I'm afraid the conversation about leading edge of transients has gone on so long that I've forgotten where it started ... so I will attempt to keep out of this thread from now on.

 

Edited November 19, 2022 by andyr

[stereo coffee]

37 minutes ago, MLXXX said:

 

 

The photographic situation can be measured as lines of resolution visible in a test pattern that includes broadly spaced lines and progressively more closely spaced lines (e.g. so many lines per inch or per centimetre).  The resolution achieved can be assessed by eye as was done in the early days of analogue television by looking at a test pattern displayed on a CRT screen.

 

Theses days with digital images (including videos consisting of a series of still images captured at a particular frame rate) we can talk of the OTF (optical transfer function). The OTF can be used to describe the extent of contrast achieved for picture content that alternates between black and white at various different spatial repetition rates.   (See the Wikipedia description of OTF at https://en.wikipedia.org/wiki/Optical_transfer_function . )

 

To achieve a very detailed image (be it a test pattern or of a real world scene) you need either:

(1) An inherently higher frequency of spatial response, which naturally gives high contrast for small details. (So, a camera with many megapixels of sensor resolution, a high quality lens, an accurate focus adjustment for the subject being photographed, and a file compression protocol that does not lose too much detail.)   Or,

(2) Digital enhancement (exaggeration) of edges in the image, to boost the subjective contrast where the lens has introduced blur through its imperfections, or a poor focus setting,  or because the optical sensor has an insufficient density of photoreceptors, or where a digital file compression algorithm has been very aggressive.  Too much of this kind of image enhancement will look harsh and unnatural.

 

With audio, you can achieve a very basic equivalent of (2) above with treble boost.  So if your system is struggling to reproduce 20kHz you can add treble lift to try to compensate for that.  However there will be a limit to how much treble can be clawed back in that way.

 

If you want an optically softer image you can reduce the OTF with an algorithm that imposes an averaging of the values of adjacent pixels, or by simply rescaling to a lower resoution pixel grid (e.g. 640 x 360 instead of 1920 x 1080). With audio for a more mellow sound, you can apply cut to the treble.

 

*  *  *

 

An important basic concept to understand with audio transients is that they cannot exceed the bandwidth of the reproduction system. Conventional CDs are only good to about 20kHz (or slightly higher) because of the need for anti-aliasing filtering. Any waveform shape at a repetition rate of 20kHz that rises of falls more steeply than a 20kHz sine wave will be rounded back to the rise and fall times of a 20kHz sine wave. (So, for example, something that started out as a 20kHz square wave would end up as a 20kHz sine wave when played back with a conventional CD player.)

 

Even with vinyl discs the upper frequencies are rolled off to avoid cutting a groove with modulation excursions too great in amplitude, and too rapid,  for the playback stylus to track safely.  So even if a cartridge is measured to be capable of responding to small excursions at say 30kHz cut into a specialist test disc, a production pressing for music typically will not include any such frequency content, or if it is present it will be at a very low level.  A practical implication is that a very fast rise time transient cut into a vinyl disc is very likely to be limited in its rise time, or "rounded", to avoid stylus tracking issues when the disc is played back. The full theoretical upper frequency limit of the vinyl disc will not be utilised. Of course even if it were utilised, it would not be expected to be heard as a difference.  To [young adult] human ears a 20kHz sine wave is a thin tone. There is no timbre because we cannot hear any overtones of 20kHz. A square wave [produced with an ultrasonic speaker] would still sound like a thin pure 20Khz sine wave, to a young adult human listener.

 

We are thank goodness not listening to sine waves or destroying our audio equipment with square waves. Rather music content clearly has frequencies above what we hear, but what is needed is sufficient bandwidth to not disturb their proper reproduction. 

 

Neither are tonal controls needed that add emphasis - without  concurrently also adding deemphasis - thus validating the inventions of Murray Crosby that are useful, distinct from adding distortion .