Talk:High-resolution audio

External links modified

Hello fellow Wikipedians,

I have just modified 2 external links on High-resolution audio. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added archive https://web.archive.org/web/20140318122826/http://hometoys.com/emagazine.php?url=%2Fezine%2F08.12%2Frusso%2Findex.htm to http://hometoys.com/emagazine.php?url=%2Fezine%2F08.12%2Frusso%2Findex.htm
• Corrected formatting/usage for http://www.mariinskylabel.com/page/faqs

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 03:35, 2 April 2017 (UTC)

How reliable is Reiss 2016?

After reading the Reiss 2016 paper, I wonder how reliable it is. It is using studies with known methodological problems (e.g. Jackson 2014 where the “CD quality” audio was deliberately sampled down from high resolution using the poor rectangular dithering method) in its meta-analysis when concluding that it is possible to hear the difference between 16/44.1 and higher resolution formats. The study seems to have a few mentions when it came out, but no real discussion in reliable sources since then.

I believe the study suffers from garbage in-garbage out, since a meta analysis of multiple unreliable “blind” tests will not be any more reliable than the individual tests, and getting a blind test right (same levels, same setup, truly blind test, ideally using a commercial product to convert the high resolution audio in to red book audio) is non trivial, so I guess this flawed study will remain in this article until someone comes along and replicates Meyer/Moran 2007 yet again. Here is a well articulated critique (from, yes, a non-reliable source) of Reiss 2016. Samboy (talk) 20:11, 28 November 2019 (UTC)

I have made some minor adjustments so that the article at least has a more balanced summary of Reiss. ~Kvng (talk) 17:14, 1 December 2019 (UTC)

Looks like Reiss was removed but another paper which references Reiss has replaced it. Here is what I added to the Super Audio CD article which references Reiss: Hiroshi Nittono pointed out that the results in Reiss's paper showed that the ability to distinguish hi resolution audio from CD quality audio "was only slightly better than chance". Reference: Hiroshi Nittono. "High-frequency sound components of high-resolution audio are not detected in auditory sensory memory". Nature. {{cite journal}}: Cite journal requires |journal= (help) Samboy (talk) 06:46, 16 May 2021 (UTC)

I have retained that but some of your other edits to that paragraph were unwarranted. ~Kvng (talk) 12:37, 20 May 2021 (UTC)

The fundamental problem with the Reiss meta-study is that it’s a meta-study containing at least one study with questionable methodologies. The Meridian study which Reiss uses, for example, used inferior dither which is not used in modern audio production, and even here, listeners could barely hear the difference. The Xiph.org post (which is now an archive-only link) has this to say about those kinds of studies:

If anything, the number of ambiguous, inconclusive, and outright invalid experimental results available through Google highlights how tricky it is to construct an accurate, objective test. The differences researchers look for are minute; they require rigorous statistical analysis to spot subconscious choices that escape test subjects' awareness. [...]

The human ear can consciously discriminate amplitude differences of about 1dB, and experiments show subconscious awareness of amplitude differences under .2dB. Humans almost universally consider louder audio to sound better, and .2dB is enough to establish this preference. Any comparison that fails to carefully amplitude-match the choices will see the louder choice preferred, even if the amplitude difference is too small to consciously notice. [...]

The danger of clipping is especially pernicious in tests that create, resample, or otherwise manipulate digital signals on the fly. [...] It is necessary to either monitor for clipping (and discard clipped audio) or avoid clipping via some other means such as attenuation. [...]

Inadvertant audible cues are almost inescapable in older analog and hybrid digital/analog testing setups. Purely digital testing setups can completely eliminate the problem in some forms of testing, but also multiply the potential of complex software bugs. Such limitations and bugs have a long history of causing false-positive results in testing

There’s a lot of ways people can “hear” a difference between a high-resolution sample and the same sample resampled at 16/44.1 (with a reasonable level set when recording at 16/44.1), most of which have nothing to do with there being some way humans can perceive high resolution audio, e.g. the levels were not carefully matched, the audio was not carefully downsampled, or audio cues were accidentally added to the signal chain.

If we take a meta-study of a bunch of studies with methodological issues, we’ll still get flawed data in the meta-study. What the various studies claiming there is a human perceptual difference with high resolution audio haven’t been able to do is explain how a human is supposed to be able to perceive this difference.

(As an aside, if there were a difference, instead of going from DSD64 to DSD128, we should had gone from 1-bit DSD64 to 2-bit DSD64. Indeed, the low cost but excellent sounding AK5381 ADC uses, for its pre-deicmated input, 2-bit DSD64, which is then deicmated down to 44.1k or 48k audio, depending on what clock rate drives the converter.)

Point being, we’re giving this Reiss meta-study too much WP:WEIGHT and its section is far too long in the Super Audio CD article. Samboy (talk) 14:23, 20 May 2021 (UTC)

I would support putting the Xiph ref as a counterpoint to Reiss. Melchior, Vicki R. (May 3, 2019). "High-Resolution Audio: A History and Perspective". Audio Engineering Society. {{cite journal}}: Cite journal requires |journal= (help) may also be useful. I think it is valuable to mention Reiss since it is central to the subject and has now been cited 34 times. ~Kvng (talk) 13:20, 23 May 2021 (UTC)

I think some of that should be there. I do want to avoid WP:SYNTH and WP:OR, but my general sense is that it’s unlikely we’re getting anything which affects how humans experience music and sound by going over 44.1k (if we were making music for my cat, then 96k would make sense). There’s some tenuous studies with some pretty high p values which perhaps suggest otherwise, but there are a lot of ways of getting these studies wrong too. Perhaps there is something there, but we haven’t found anything really solid yet, much less a cohesive scientific theory of ultrasonic perception, and there are multiple studies which flat out conclude nothing is there which we can perceive at all.

On the other hand, people selling expensive stereo systems and high resolution recordings would love to have people believe there really is something there which affects how we hear music. There’s a lot of money to be made selling $10,000 stereo systems and $900 two track recorders (the DA-3000, with 24/192 and DSD128) instead of $300 four track recorders (the SD-20M, but even this one supports 24/96), so I do not believe a lot of the research we see is done in good faith.

I will, when writing in the article space, try to be neutral, of course. For the record, since it may influence my edits, I’m recording using four stereo AK5381 analog to digital converters (8 channels total) at 44.1k, but they can also record at 48k (internal sampling is a 2-bit DSD64 variant but, unlike the PCM4222, one can only access the decimated PCM form), which has a total component cost of under $10, so that could be the axe I am grinding here (I have used 96k, but only to reduce converter latency). Samboy (talk) 19:25, 23 May 2021 (UTC)

It is clear that trained listers can distinguish different kinds of anti-aliasing filters. Higher sample rates give designers more options in how they do anti-aliasing. It is hard to control for this in listening tests. There seems to be a growing consensus among audio scientists that anti-alsing filter, not sample rate itself, is what makes the higher sample rate systems sound different or better. ~Kvng (talk) 12:41, 26 May 2021 (UTC)

Controversies

Why are we citing articles on cnet and Gizmodo about the opinions of Hi-Res audio? The least we could do is site acadamic research or experts... — Preceding unsigned comment added by 98.223.229.101 (talk • contribs)

Wikipedia prefers WP:SECONDARY sources. ~Kvng (talk) 14:48, 29 June 2020 (UTC)

Feel free to cite academic research papers. I don't know what "experts" you refer to. Binksternet (talk) 15:46, 29 June 2020 (UTC)

Anti-aliasing filter

Anti-aliasing contriversal only outside of acedemia Ohgddfp (talk) 05:22, 14 June 2023 (UTC)

About the last sentence in the Controversy section: "Some technical explanations for sonic superiority cite the improved time domain impulse response of the anti-aliasing filter allowed by higher sample rates. This reduces the energy spread in time from transient signals such as plucking a string or striking a cymbal." Ohgddfp (talk) 06:33, 14 June 2023 (UTC)

I would like to see a definition of "improved" in this context. If it's true that the human cochlea has cells with hairs and other features that are tuned to various frequencies of physical vibration, and such physical vibration is needed for humans to hear anything, then adding or subtracting a frequency for which no "hair cell" exists that can resonant with that frequency, would not be detectable by humans, making such a frequency "ultrasonic". Since the entire set of hair cells are vibrating exactly the same with or without a change in the ultrasonic frequencies, a "better" impulse response, which changes only the ultrasonic frequencies, therefore cannot change how we perceive sound. What is not controversial in acedemia is the fact that a "better" time domain impulse response can only add ultrasounic frequencies. So in short, the sound frequencies would have to change for humans to detect a change in the sound. Ohgddfp (talk) 06:33, 14 June 2023 (UTC)

In a related filter issue, the time delay of any linear system must be the same for all frequencies in order to avoid a change in the phase response, manifesting as a change of the shape of the time domain response--a bad impulse response, for example. It is said that a brickwall filter causes this kind of phase distortion, and a higher sample rate makes the design of an anti-aliasing filter easier. So oversampling solves this. But what is left out of the conversation is that post-sampling digital processing by a second digital filter produces this brickwall response easily without phase distorion. And with this brickwall response, it is easy to down sample (for CD sample rate streaming release) without incuring any phase distortion. (See the "Oversampling" section of the "Anti-aliasing filter" wiki article.) Ohgddfp (talk) 06:33, 14 June 2023 (UTC)

very old (2012-14) citations (imo no reliable sources)

citation:

...and with other opinions ranging from skeptical to highly critical:

• "If [the music business] cared about sound quality in the first place, they would make all of the releases sound great in every format they sell: MP3, FLAC, CD, iTunes, or LP."—cnet 404 dead)

• "Impractical overkill that nobody can afford"—Gizmodo (June 2013)

• "A solution to a problem that doesn't exist, a business model based on willful ignorance and scamming people."—Xiph.org

Business magazine Bloomberg Businessweek suggests that caution is in order with regard to high-resolution audio: "There is reason to be wary, given consumer electronics companies' history of pushing advancements whose main virtue is to require everyone to buy new gadgets."

Did Business magazine author Joshua Brustein in 2014 really have High End experience - and equipment that could show the sound surplus of High Res ?

In the last decade, equipment has become much better than it was in 2014.

High Res is 'cheap today (I'm just listening to a Qobuz Abo for 150 € per year = 12,50 € per month = 0,41 € € per day) - Gizmodo (2013) is outdated since years.

Let's remove citations 2-4 or the whole section. 88.153.240.29 (talk) 12:34, 5 March 2024 (UTC)

References

No, we are not removing sources from around ten years ago. Their conclusions are still valid. Binksternet (talk) 13:16, 5 March 2024 (UTC)