Reasonable definition of ‘hi-res’ music

That just isn’t going to happen. In 2017 only 32% of music industry revenues were from physical or download purchases, and probably fell further in 2018. Streaming services accounted for 65% of revenues. As the main type of equipment is mobile phones, will you be forced to buy a new phone with MQA decoding?

Qobuz were going to test DSD, but never bothered. Another dead format (for purchasing/streaming at least).

1 Like

@davewantsmoore

Yes, a label could put mp3 inside. Nefariousness happens I suppose. Up/resampling is the norm these days - almost all consumer DAC’s do it in one way or another. I agree with you hi res is used and abused in the effort to re-re-sale catalogs and the like. Still, given the right conditions (which granted is in most cases an “audiophile” with the right hardware/software) hi res is can be an advantage, quite apart from limits of human hearing, of microphones, of the moral reliability of people in the music industry, etc.

No, hi res is not an meta-guarantee of quality “inside”, but then nothing else is either (unless you buy into not only the concept, but the execution of MQA or something like it) - it’s just a piece of the puzzle.

More like a can of worms. There exists at least a decades worth of digitally recorded music that was recorded at bit depths and sample rates below 24bit/96kHz so these recordings will never be high resolution no matter how much they are upsampled.

Up/resampling is the norm these days - almost all consumer DAC’s do it in one way or another.

That has always been the case.

No, hi res is not an meta-guarantee of quality “inside”, but then nothing else is either

Sure, but it is being marketed as such. :slight_smile:

There exists at least a decades worth of digitally recorded music that was recorded at bit depths and sample rates below 24bit/96kHz so these recordings will never be high resolution no matter how much they are upsampled.

This statement isn’t necessarily true.

If I record at 16/44 and 24/96, it is possible I get the same outcome … or it’s possible I get a different, but indistinguishable outcomes.

If I record at 16/44 … it is possible for me to improve the playback performance of that data (in some circumstances) by using a resampler which improves the data (although this is very situationally and equipment dependant/specific).

The fact that there is so much grey area, is the problem … as the marketing department is presenting it to consumers as clear “black and white” things… which is completely understandable - if they explain to consumers that it isn’t “clear”, then nobody would be interested. $$$$$$

I’ve not heard of this, can you point me to a source about it ? I take it you’re referring to something more than upsampling, which I understand as zero stuffing to create bandwidth for gentler filtering.

1 Like

When audio is resampled, it is possible to use different methods (filters) to do this in addition to simple conceptual filters most people think of.

The choice of resampling filter will result in differing digital data, which will result in differing analogue output from your DAC.

Players like HQplayer (etc.) allow the user to choice different resampling filters, for precisely this purpose. Which one is “better” depends on what the DAC does during it’s DA conversion.

Hardware like the Chord M-Scaler also attempt to benefit from this. It resamples incoming digital data to high PCM rates (eg. 768khz). It is not the high rate per se which is the benefit. It is the opportunity to produce digital data which is more accurately converted by the DA converter.

High sampling rates (0r bit depths) are not a proxy for quality… it is what was DONE with them, which matters.

Where we are at now with “high res”, is a bit like saying, how much you will enjoy looking at a painting, depends on which brand or type of paintbrush the artist used.

2 Likes

Okay let me get this straight. My understanding of the way a analog to digital conversion works, and this is a very simplified explanation, is that an electrical signal, and that electrical signal having been created by converting sound into an electrical analog of sound waves, is sampled and for each sample the frequency and amplitude is measured and assigned a numeric/digital value. During the digital to analog conversion this process is reversed, i.e. each digital value is converted back to hopefully the original electrical signal.

So the only way to “improve” the sound of the original analog electrical signal would be to convert it to a digital signal in a different or better manner. Higher sampling frequency or higher bit depth. This is possible if the original electrical signal was the result of an analog recording but if the original electrical signal was the result of a digital recording then the digital data is “fixed” or set at whatever sampling rate and bit depth was in use at the time of the recording.

Please feel free to correct me if I’m wrong but if so then please use facts and not hearsay. Thanks!

So the only way to “improve” the sound of the original analog electrical signal would be to convert it to a digital signal in a different or better manner.

Yes. However another way to view that is “do less damage during the conversion”.

The DAC does not convert the digital data back to the analogue waveform perfectly. If you “rearrange” the digital data, then you can potentially improve that conversion, so what comes out of the DAC is closer to the original.

Resampling the audio can be used to achieve this outcome. We can not say that X or Y rate is “better”, or “needed” … but only that “rearranging” the digital data, offers an opportunity to improve the outcome.

These issues of “resampling” however, is no direct justification for audio to be distributed in high rates or depths … or that audio contained in a high rate or high depth container, is high quality audio.

For example, we could use resampling to 24bit and 384khz (or whatever) to make the output actually worse

You say improved, I say hogwash. If the incoming value is 3476kHz with an amplitude of 76.8 db and the ADC/DAC cycle produces a end result of 3476kHz with an amplitude of 76.8 db how does one “improve” that result?

The same thinking was used years ago when discussing converting a file from wav to flac and back to wav. I ran a file through several hundred conversion cycles and did a check sum on the result, which was exactly the same as the original wav file.

Please leave the magical thinking to the professionals, you know those reviewers in the high end audio magazines.

1 Like

I realize that I was being a little harsh in my last response. I fully realize that the digital to analog conversion is not as perfect as I let on and what you are talking about is a way to improve upon that conversion by making it just a little more perfect. Sorry for my misunderstanding.

You say improved, I say hogwash. If the incoming value is 3476kHz with an amplitude of 76.8 db and the ADC/DAC cycle produces a end result of 3476kHz with an amplitude of 76.8 db how does one “improve” that result?

You say “hogwash”. So why are you asking?

The same thinking was used years ago when discussing converting a file from wav to flac and back to wav

No. This is somewhat quite different. Conversion from (for example) wav to flac, remains digital … it’s a completely different thing.

If the incoming value is 3476kHz with an amplitude of 76.8 db and the ADC/DAC cycle produces a end result of 3476kHz with an amplitude of 76.8 db

As you noted earlier… This is a very “simplified” was to look at it.

In short… “it doesn’t”.

Please leave the magical thinking to the professionals

There is no “magic thinking here”… and you have no idea of what my credentials are (not that it would matter, as “appeal to authority” is a weak argument … there are plenty of idiot “professionals” as you note)

To confirm “no magical thinking” … all you need to do is resample audio with various resampling filters, and see that the analogue output of the DA converter is different.

_<

I realize that I was being a little harsh in my last response

I don’t think “harsh” is quite right. It isn’t smart to call an answer “hogwash”, when you actually have no idea about the topic at hand…

People who actually understand this topic, don’t want to “debate” it with you when you are so misinformed… it drives away people who are actually aware of the answers, and leaves only people like yourselves. Ones which don’t understand the topic, and who want to “debate” it.

If you want to learn, ask more questions… rather than arguing.

I believe this is true. If the content wasn’t captured at a high resolution, no amount of upsampling or resampling can add it, and if it tried it would be guessing.

“Resampling audio” in Wikipedia directs here and appears to be a synonym for upsampling. Different reconstruction filters may alter the analog playback signal, but I wouldn’t regard that process as producing a higher resolution output.

I had to re-read this davewantsmoore - what you mean to say (I think) is that choice of resampling/filter will result in a differing calculation for a waveform, which is then used in the DAC process. It sounds like you are saying that resampling effects the original sampled data.

This simplification, while it can be useful, also leads to an over-simplification. Waveform summation means that sound is actually a “summation” or “collection” (to misuse a word) of all the sound of real audio (say a voice or a band playing), including things such as distortion, background noise, etc. This is true even though an audio waveform (and thus a digital sampling of said waveform) is a relatively “simple” sinusoidal waveform. It’s amazing when you think about it - sound is just a pressure wave through a medium (in our case, the atmosphere), yet this simple waveform can carry all the sound, of all differing frequencies, sources, amplitude, to our ears at the same time.

Well, upsampling/resampling (in relation to filters) is a way to manage (to pick a term) how the calculation of the DAC process is performed, and this is necessary because a DAC process always involves distortion (of the original waveform - the signal). The goal is to perform the calculations in such a way as to minimize distortion in (and out! intermodulation distortion) of the audio band so that the analogue waveform is as close to the digital sampling as possible. Digital sampling has other limits related, such as noise floor, and the like and so how you perform this process can affect this (e.g. noise shaping).

All this is known engineering art so I am somewhat confused about davewantsmore stance. I don’t think he is saying that any of this is irrelevant or is not so, but that it is oversold and/or that the original “quality” of the recording (everything I suppose, the quality of the players, the quality of the performance, the quality of the microphones, the quality of the analogue chain up to the ADC, the quality of the ADC, etc.) is much more important and I agree with him up to a certain point…

I believe this is true. If the content wasn’t captured at a high resolution, no amount of upsampling or resampling can add it, and if it tried it would be guessing.

That’s somewhat correct… depending on your definition of "high resolution… which is exactly my point. Peoples definition of high-resolution is somewhat misguided - as the rate or the depth that the audio is captured with, or played back at (via resampling)… does not tell us anything reliable about the quality in general.

However, resampling CAN be used to improve the way which the DA converter renders the output. Doing this doesn’t “add” anything… instead what it does is prevent the DA converter from rendering the output at a poorer quality than it otherwise would. However this depends on the hardware - we cannot just conclude in general that “resampling = good”. We cannot conclude that if you resample audio to XY rate that it is an improvement … and we cannot conclude that a higher rate is “better” than a lower one (eg. 192 better than 96).

This is also a similar version of this which applies to AD conversion.

Remember. This resampling process happens inside nearly ever DA converter… for exactly the same reason, “to improve the quality” (ie. do less damage, than it otherwise would). All we are talking about here with “resampling” is moving the resampling to outside the DA converter, where we have more control over it.

Requantization

1 Like

Different reconstruction filters may alter the analog playback signal, but I wouldn’t regard that process as producing a higher resolution output.

As mentioned… it depends on how we think about “resolution”?

If I have audio data at 16/44 … and I play it back as is, I get analogue waveform X
If I resample that 16/44 data in a specific way … and I play it back, I get analogue waveform Y

Y is closer to the analogue waveform that is theoretically represented by the original digital data… ie. there was less error in the conversion process.

Another example. I sample an analogue signal into two sets of digital data, 16/44, and 24/176.

Now I play them back through a DA converter… but before I put the 16/44 data into the DA, I resample it to 24/176… using a filter specifically designed to improve the performance of the DA.

Now - if we look at only the sound louder than -96dBFS … and only at the frequencies below 20khz. Which one delivers the higher performance? You cannot say with any certainty.

1 Like