Hi-resolution versus 16 bit 44.1 Khz

[March Audio]

4 hours ago, andyr said:

But given that theory ... it must not be possible to get any better SQ than CD standard (16bit / 44.1kHz) - right?

 

Which IME ... is simply not the case

 

Yes that is the case if you make the assumption that digitisation and replay is performed perfectly.

 

The problem is that your estimation is based on and influenced by many other confounding factors, none of which has anything to do with the sample rate.

 

Edit: to be fair my view is that if cd had been 18 bit 48kHz it would what I would refer to as "blameless", but 16bit 44.1kHz is adequate for vast majority of recordings.

Edited January 20, 2023 by March Audio

[March Audio]

4 hours ago, andyr said:

My (listening) experience is that there is - but nobody seems to be able to explain why.

 

But what is your listening experience based on?  Have you ever performed any controlled tests comparing 44kHz v 96KHz recordings?

 

Most "hi res" recordings have been remastered.  They arent even the same master tape played back on the same tape deck as the 44kHz version.

 

They simply don't sound the same, but that's got nothing to do with sample rate.

Edited January 20, 2023 by March Audio

[March Audio]

11 hours ago, LHC said:

 

If you are interested in this question may I suggests you follow the work and outputs of the AES TECHNICAL COMMITTEE on 'HIGH RESOLUTION AUDIO' (https://www.aes.org/technical/hra/) 

 

In 2016 this HRA commissioned a meta-study on scientific evidence for hi res audio resulting in this paper by JOSHUA D. REISS (https://qmro.qmul.ac.uk/xmlui/bitstream/handle/123456789/13493/Reiss A Meta-Analysis of High Resolution 2016 Published.pdf)

Quoting its finding "Results showed a small but statistically significant ability of test subjects to discriminate high resolution content, and this effect increased dramatically when test subjects received extensive training."

 

I think a colloquial way of stating Reiss finding is that 'some people can hear the differences in some high res audio content some of the time'. AFAIK there is no consensus explanation of this finding; and to this day remains an open scientific question. Unfortunately this does not appear to be a very active area of scientific research, but notably there are ongoing Japanese researchers publishing in this area. For example the following Japanese paper (https://web.archive.org/web/20210715091831id_/https://www.jstage.jst.go.jp/article/jjppp/38/3/38_2003br/_pdf) has this abstract:

 

"High-resolution audio has a higher temporal and/or depth resolution than compact disks (CDs). Several researchers have suggested that inaudible high-frequency components of high-resolution audio might influence listeners’ psychophysiological states. Two types of digital sound sources with the identical frequency structure
(＜22 kHz) were created at different sampling rates (192 kHz and 44.1 kHz) by filtering out inaudible highfrequency components of an original sound source, which was an excerpt of natural environmental sounds recorded in 192-kHz/24-bit format. Twenty-four university students listened to the three sounds in a random order for five minutes each. Results indicated that the 192-kHz sound source compared to the 44.1-kHz sound source induced higher power in the theta (4.0‒8.0 Hz) and slow alpha (8.0‒10.5 Hz) bands of the electroencephalogram. However, no apparent differences were found in sound quality or subjective mood. These results suggest that digital sound sources with a higher sampling frequency than CDs influence the physiological state of listeners, although the difference might not be consciously perceived."

 

Another study by the Japanese in 2020 (https://www.nature.com/articles/s41598-020-78889-9) Abstract

"High-resolution digital audio is believed to produce a better listening experience than the standard quality audio, such as compact disks (CDs) and digital versatile disks (DVDs). One common belief is that high-resolution digital audio is superior due to the higher frequency (> 22 kHz) of its sound components, a characteristic unique to this audio. This study examined whether sounds with high-frequency components were processed differently from similar sounds without these components in the auditory cortex. Mismatch negativity (MMN), an electrocortical index of auditory deviance detection in sensory memory, was recorded in young adults with normal hearing (N = 38) using two types of white noise bursts: original sound and digitally filtered sound from which high-frequency components were removed. The two sounds did not produce any MMN response and could not be discriminated behaviourally. In conclusion, even if high-resolution audio is superior to the standard format, the difference is apparently not detectable at the cortical level."

 

While these may seem like negative results thankfully research continues (albeit slowly). 

This "psychological" research is highly dubious and has not been replicated. 

 

Even if they did manage in a research environment to effect a linear record and replay system with a bandwidth of 96kHz (I say they didnt), you certainly won't find that in any domestic hifi speaker.

 

A meta-analysis of 17 studies published between 1980 and 2016 showed that discrimination performance was only slightly better than chance

 

.....,..........

 

However, no apparent differences were found in sound quality or subjective mood.

 

 

Edited January 21, 2023 by March Audio

[Addicted to music]

Just because the numbers are higher doesn’t mean its bad.  Sometimes higher numbers; bit rate and sampling rate that increases FR capability can also mean that the equipment can pick up other audio FR spikes, usually above 20khz that’s not wanted.   Go to around 3 mins in in this video, that spike at 45khz and some are above is picking up other electronic gear emitting noise pollution that you can’t  hear but if you have the tools you can see that it’s doing it’s job right to specs.

 

 

 

Edited January 20, 2023 by Addicted to music

[MLXXX]

7 hours ago, March Audio said:

A meta-analysis of 17 studies published between 1980 and 2016 showed that discrimination performance was only slightly better than chance

A case of damning with faint praise.  All that effort at trying to find a correlation across multiple studies, and that's all that was achieved!

 

7 hours ago, March Audio said:

Edit: to be fair my view is that if cd had been 18 bit 48kHz it would what I would refer to as "blameless", but 16bit 44.1kHz is adequate for vast majority of recordings.

I agree re use of 48kHz as it makes the reconstruction filtering significantly less of an issue.  There can be very slight audible differences between DACs fed a 44.1kHz sample rate stream, attributable to different filtering choices.  (Very subtle differences, audible in careful A B comparisons.)

Edited January 20, 2023 by MLXXX

[Addicted to music]

8 hours ago, March Audio said:

This "psychological" research is highly dubious and not been replicated. 

 

Even if they did manage in a research environment to effect a linear record and replay system with a bandwidth of 96kHz (I say they didnt), you certainly won't find that in any domestic hifi speaker.

 

A meta-analysis of 17 studies published between 1980 and 2016 showed that discrimination performance was only slightly better than chance

 

However, no apparent differences were found in sound quality or subjective mood.

 

 

Here is someone who is young, done the music training and has a qualification and she can hear up to 18khz only pick 4 out of 6    Interesting video that’s been posted b4.

 

 

 

 

 

 

Edited January 20, 2023 by Addicted to music

[March Audio]

2 hours ago, MLXXX said:

A case of damning with faint praise.  All that effort at trying to find a correlation across multiple studies, and that's all that was achieved!

 

I agree re use of 48kHz as it makes the reconstruction filtering significantly less of an issue.  There can be very slight audible differences between DACs fed a 44.1kHz sample rate stream, attributable to different filtering choices.  (Very subtle differences, audible in careful A B comparisons.)

 

In my experience the only filters that make an audible difference are the slow ones that filter into the actual audible band below 20kHz.

 

Filter ringing is a total red herring.  I will demonstrate why later.

Edited January 21, 2023 by March Audio

[March Audio]

2 hours ago, MLXXX said:

A case of damning with faint praise.  All that effort at trying to find a correlation across multiple studies, and that's all that was achieved!

 

Indeed.  "Slightly better than chance" is in the area of experimental error.

Edited January 21, 2023 by March Audio

[March Audio]

1 hour ago, Addicted to music said:

 

Here is someone who is young, done the music training and has a qualification and she can hear up to 18khz only pick 4 out of 6    Interesting video that’s been posted b4.

 

 

 

 

 

 

 

This test is interesting but not really about sample rate.  It's comparing wav to mp3.  Plus it would need to be performed multiple times on the same song to provide meaningful data.  Just one result per song can be a guess.

[aussievintage]

It would help people understand frequency response, and what matters, if they could sit and listen to the results of a very sharp cutoff low pass filter as they move it down the spectrum.  In my Puredata software I have a plugin that implements a chebychev lowpass filter.  I set it for 8 stages, which is 16 poles, nice and steep.

 

I use it to filter out high frequencies, particularly on bad recordings,  and old 78 rpm discs.  But if I feed it a good signal, let's say CD digital, it's amazing how much you can lop off the top before it is audible, old ears etc conceded.

 

I always remind myself, there's a reason frequency response graphs are  log, not linear.   That bit from 10kHz to 20kHz is rightly only a small section.  Most music is between 100Hz to 10kHz - bass below that, and noise and effects above that (exaggeration I know, but only to make a point  

[LHC]

13 minutes ago, March Audio said:

 

Indeed.  "Slightly better than chance" is in the area of experimental error.

 

Potentially correct. But it has been seven years since the meta-analysis and AFAIK no one has identified the 'experimental error' that led to this statistical 'anomaly'. 

 

 

[March Audio]

1 hour ago, LHC said:

 

Potentially correct. But it has been seven years since the meta-analysis and AFAIK no one has identified the 'experimental error' that led to this statistical 'anomaly'. 

 

 

Well thats probably because no one has bothered to look because the result is so clear.  The result is providing no evidence that people can hear a difference, quite the contrary in fact.

 

If people could genuinely hear a difference the result would be way better than "chance".  When you have a minuscule number of outliers in the results you investigate reasons for the anomalies as there is usually a confounding factor rather than any genuine reason.

 

So a very slight indication above chance is likely to be an anomaly.  You would need to look in detail at each studies methods. I would also expect that the more studies you add into the meta group, the more it would tend to "chance".

 

 

Edited January 21, 2023 by March Audio

[LHC]

Speaking in broad generality scientists love 'anomalies' because they could represent signals of new discoveries or lead to innovations. So when scientists encounter an anomaly in their results it is followed by a fury of effort to try and explain it away using conventional means, for example it could be a result of experimental errors, data handling/analysis errors, or calculation errors. One recent example is the B-meson decay anomaly that teased physicists of the possibility of new physics beyond standard model; after more data collection and more thorough studies this anomaly is no more and previous discrepancy attributed to errors (https://www.scientificamerican.com/article/hint-of-crack-in-standard-model-vanishes-in-lhc-data/). This is par for the course.

 

So I naturally assumed this would be the same for Josh Reiss (now president of AES) meta-analysis finding of a statistically significant anomaly in hi res experiments, and should lead to a fury of effort in the audio research community to explain it away. Apparently this is not the case and I could only find a trickle of related research since. The general lack of interests (when compared with the broader Science norm) is puzzling. 

 

 

[LHC]

To be fair I should mention the online survey done by Mark Waldrep and accepted as a AES paper in 2020

https://www.aes.org/e-lib/browse.cfm?elib=20954 (behind a paid wall). Abstract:

 

"The perceptibility of high-resolution versus CD standard audio has been the subject of research and debate since the introduction of hi-res audio distribution formats twenty years ago. The author conducted a large survey to determine whether experienced listeners could differentiate between a diverse set of twenty native high-resolution PCM stereo recordings and down conversions of the same masters at 44.1 kHz/16-bits fidelity. Participants were encouraged to audition the files using their own systems, which ranged from modest, headphone-based personal setups to audiophile quality rooms costing in excess of $50,000 to professional studio environments. They were not allowed to use analytical tools or other non-listening means to assist in their observations. Over 400 responses were received from professional audio engineers, experienced audiophiles, casual music enthusiasts, and novices aged eleven to eighty-one years. The online survey submissions show that high-resolution audio was undetectable by a substantial majority of the respondents regardless of experience level, equipment cost, or process with almost 25% choosing “No Choice.”. However, some evidence exists that specific genres and recordings produced moderately higher positives."

[davewantsmoore]

23 hours ago, andyr said:

So what is all this extra data (being used) for?

Storing the frequencies between 20khz and 40khz.

 

 

But... typically.

• They are extremely quiet
• They are highly distorted by your record player
• They are highly distorted by your speakers
• ... but even if they were loud and undistorted, they would still be inaudible.

It isn't necessarily the wrong choice to work at a high(er) sampling rate (it's a complex discussion, with lots of ifs and buts) ..... however the 4 dot points above should not be forgotten, and they are all too quickly discarded by people who "heard a high sample rate thing which sounded good and make the leap of faith that 'more is better' "

[davewantsmoore]

17 hours ago, andyr said:

 

Shure, Trev - but the topic is (surely?) ... is there any better SQ than that which is delivered by 16bit/44.1kHz CD-standard recordings!

 

My (listening) experience is that there is - but nobody seems to be able to explain why.

 

If you take the better sounding recording... and you convert it to 16/44, then it will sound just the same (with certain caveats).

 

ie.   you should not attribute the higher sound quality to the container it is delivered to you in.

 

The actual reason(s) for the goodness of the sound you are referring to .... could be a number of things.   It will depend on the specifics of the recording.

[davewantsmoore]

1 hour ago, LHC said:

So I naturally assumed this would be the same for Josh Reiss (now president of AES) meta-analysis finding of a statistically significant

"statistically significant"  ?

 

So he says.

 

Consider that "just better than chance" .... may quite likely mean that if we include more trials, that will flip to "just worse than chance".

 

In my dictionary "significant" has a different meaning.

[davewantsmoore]

20 hours ago, andyr said:

Surely this would have a minimal number of recorded data points

No, the signal entering the recorder still contains noise at all frequencies.... and that would be recorded.

[MattyW]

On 17/1/2023 at 8:08 PM, Niktech said:

Is anyone really noticing a benefit to listening to 24 bit 48 KHz and above to listening to 16 bit 44.1khz?

Some tracks I’ve listened too at 16 bit 44.1 KHz sound way better than the the hi -res. Some hi-res has so much air and treble it’s unlistenable  to my ears- it’s like nails down a chalkboard .

 

It would seem what is more important than the resolution is the quality of the Master, i.e. un-butchered, and of course your equipment. I guess this in someways ties in with the desktop / streaming thread. What are forumites preferences here?

 

 I did do a search to see if we’ve discussed/debated this previously on SNA, but I’ve had a few vodka beverages now and may have overlooked it. I did see one dated 2012. 
 

Cheers
 

 

 

I’m in complete agreement with everything you’ve said here. I even downsample all content to 16 bit 44k to lessen the impact streaming from my Roon server to Roon end points as I’m using Wi-Fi otherwise I run into issues when the weather is humid.

[davewantsmoore]

2 hours ago, LHC said:

The general lack of interests

I genuinely don't think this is true.

I think everyone has just done their research and made their conclusion.

 

It is actually not that (relatively) difficult a thing to test, you can use sound sources which have significant HF spectrum, and record them very carefully .....  you can play them back using special HF transducers, and increase the HF level to many many many times the original levels.   At that point when you still find it is inaudible, it seems like a reasonable conclusion that it won't be audible with normal recordings on normal playback systems, at normal levels.

[March Audio]

37 minutes ago, davewantsmoore said:

In my dictionary "significant" has a different meaning.

In mine too.  There is no significant indication that people can hear differences.  There is highly significant indication that they can't.

 

2 hours ago, LHC said:

The general lack of interests (when compared with the broader Science norm) is puzzling. 

 

It's not puzzling at all.  We have significant evidence that people can't hear differences in study after study.  Why would you keep flogging a dead horse?

 

Edited January 21, 2023 by March Audio

[andyr]

46 minutes ago, davewantsmoore said:

If you take the better sounding recording... and you convert it to 16/44, then it will sound just the same (with certain caveats).

 

That's very bizarre concept Dave - why on earth would I want to do that?

 

(It's kinda like the concept of comparing the sound that two amplifiers make (from your spkrs) ... by running them both at a volume level which you can just hear.  

 

To me - given I'm now ripping all my LPs - the issue is simply:

• do I rip them at 48kHz
• or 96kHz?

 

Nothing I've read on this thread is convincing me that I should save disk space and only record at 48kHz.

 

46 minutes ago, davewantsmoore said:

The actual reason(s) for the goodness of the sound you are referring to .... could be a number of things.   It will depend on the specifics of the recording.

 

In my case ... the recording is the same - the only difference is the sampling speed I choose to record at.

 

Edited January 21, 2023 by andyr

[March Audio]

1 hour ago, LHC said:

To be fair I should mention the online survey done by Mark Waldrep and accepted as a AES paper in 2020

https://www.aes.org/e-lib/browse.cfm?elib=20954 (behind a paid wall). Abstract:

 

"The perceptibility of high-resolution versus CD standard audio has been the subject of research and debate since the introduction of hi-res audio distribution formats twenty years ago. The author conducted a large survey to determine whether experienced listeners could differentiate between a diverse set of twenty native high-resolution PCM stereo recordings and down conversions of the same masters at 44.1 kHz/16-bits fidelity. Participants were encouraged to audition the files using their own systems, which ranged from modest, headphone-based personal setups to audiophile quality rooms costing in excess of $50,000 to professional studio environments. They were not allowed to use analytical tools or other non-listening means to assist in their observations. Over 400 responses were received from professional audio engineers, experienced audiophiles, casual music enthusiasts, and novices aged eleven to eighty-one years. The online survey submissions show that high-resolution audio was undetectable by a substantial majority of the respondents regardless of experience level, equipment cost, or process with almost 25% choosing “No Choice.”. However, some evidence exists that specific genres and recordings produced moderately higher positives."

 

So more significant evidence that differences can't be heard.

 

This is what I would call a very "casual" study however. It shouldn't be taken too seriously. Participants in serious studies need to be screened for things like normal hearing etc.

 

EDIT:  I haven't read the study yet, but the comments from peers below the abstract in your link highlight that it is fundamentally flawed.

 

Edited January 21, 2023 by March Audio

[davewantsmoore]

8 minutes ago, andyr said:

That's very bizarre concept Dave - why on earth would I want to do that?

 

All I was trying to say, is that IF you did it.   It would sound the same.    ie. the quality of the sound is not related to the sample rate.

 

8 minutes ago, andyr said:

(It's kinda like the concept of comparing the sound that two amplifiers make (from your spkrs) ... by running them both at a volume level which you can just hear.  

Huh?

How is what I said, anything like that.   

 

8 minutes ago, andyr said:

Nothing I've read on this thread is convincing me that I should save disk space and only record at 48kHz.

Disk space wouldn't be my concern either.

 

8 minutes ago, andyr said:

In my case ... the recording is the same - the only difference is the sampling speed I choose to record at.

You should not expect that capturing the higher frequencies will sound better .... unless there is any specific quirks in your equipment which make them perform better with a higher sample rate.

 

If there was (such a quirk), you could still record at a lower sampling rate, and have your playback software "oversample" (increase the sample rate) at playback.

[March Audio]

34 minutes ago, andyr said:

Nothing I've read on this thread is convincing me that I should save disk space and only record at 48kHz.

 

So regardless of all the overwhelming scientific evidence, including evidence that is based on people listening, you still choose to beleive otherwise.

 

OK, that's your prerogative.

 

Edited January 21, 2023 by March Audio