[ProAudio] [ProAud] The High-Resolution Challenge

[Richard L. Hess]

 From a friend who I've been discussing this with as well who is not on 
the list:

> I think the law of unintended consequences applies here. Occasionally, 
> when the record companies re-issue the title for HD, they find an 
> earlier generation "master", or they do a better job at re-mastering.
>
So that is one good reason to support HD re-mastering. But it would not 
be necessary if the record companies had done what they should have for 
the first round of CD issues. And this statement does not diminish any 
of the comments below, hence why it's an "unintended consequence."

Cheers,

Richard

On 2020-02-11 3:27 p.m., Richard L. Hess via ProAudio wrote:
>
>
> On 2020-02-11 1:28 p.m., Bob Katz via ProAudio wrote:
>>
>> Come on, guys.... We've been down this road in this reflector several 
>> times. You may recall at least one of the tests I worked on with some 
>> of you many years ago. I've subsequently performed 
>> differently-designed tests designed to try to settle the issue of 
>> "bandwidth", and each time the listening tests lead to the same 
>> conclusions:
>>
>> Thus I am 99.9999999% convinced that the sonic differences between 
>> sample rates are not due to the bandwidth, but rather to the 
>> performance of the converters themselves. Unfortunately, the 
>> bandwidth mafia at HD Tracks keeps maintaining the illusion that what 
>> we can see has anything to do with what we can hear. And I hope that 
>> JJ does not make his application because it will continue to mislead 
>> the public.
>>
> Bob, I thank you for saying that. For some reason, the archival 
> community -- based mostly on the premise that 96 sounded better than 
> 44.1 back in 2001 or so and I told them exactly that it was the 
> converters' filtering and topology that was making the difference, not 
> the actual bandwidth -- decided on the minimum standard of 96/24 for 
> all archival work. When I get in large (for me) collections of oral 
> history on cassettes recorded on the $40 book sized 5-C-cell 
> recorders, I really, really try and talk them out of 96/24 and suggest 
> 48/24.
>
> Part of this trend for the last two decades has been that archivists 
> learn 96/24 in archiving school and they're not audio engineers. 
> Sadly, 96 kHz is too low to capture bias, but it does provide a more 
> gentle rolloff and some tape machines can go out to 35-40 kHz, so for 
> high-quality music I don't argue, but for the aforementioned cassettes 
> I tell people, "you want to use twice the data space for the project 
> for no benefit?" I get one of two responses, "OK, I understand, use 
> 48" or "No, our standard is 96 and we want you to use that."
>
> Goran Finnberg also agrees with this premise and has been saying it 
> for two decades as well.
>>
>> In Bob Stuart's paper, which is open access so you do not have to be 
>> an AES member to download this:
>>
>>   J. R. Stuart and P. G. Craven, “The Gentle Art of Dithering” J. 
>> Audio Eng. Soc., vol. 67, no. 5, pp. 278–299, (2019 May.). DOI: 
>> https://doi.org/10.17743/jaes.2019.0011
>>
>> he points out on page 290 the number of decimators and upsampling 
>> filters that occur in typical chip-based converters. And that these 
>> stages are not dithered. And that there are fewer of these stages 
>> when the converters are used at a higher sample rate. Smoking gun....
>>
>> In a conversation I had with him at the New York AES last year, I 
>> told him that I discovered that audio sounds superior in many current 
>> converters if you upsample it and reproduce it at the higher rate. I 
>> also told him of my experiments showing that if you start with, for 
>> example, a 96 kHz recording, downsample it to 44.1 k and then 
>> reupsample it to 96k, that it sounds identical to the original, but 
>> the 44.1 k intermediate stage sounds worse, smaller and less 
>> resolved. In my book, I point out: How can a second generation in a 
>> chain sound worse than the third generation?
>>
>> Bob Stuart's explanation for this phenomenon is the design of the 
>> converters themselves.
>>
>> Thus my conclusion that the DACs perform better at the higher sample 
>> rate. Stuart explained that when the converters work at a higher 
>> sample rate, the audio goes through fewer stages of either decimation 
>> or upsampling, and that these stages are typically not dithered. The 
>> fewer of these stages, the better the audio sounds. So, folks, it's 
>> not the bandwidth that makes the higher sample rates sound better, 
>> it's the internal design of the converters themselves.
>>
>> Note that in one or more of his MQA papers Stuart describes the 
>> processes' restoration of high frequency information "just in case" 
>> but acknowledges that it may not be necessary. Like chicken soup, 
>> keeping the extra high frequency information couldn't hurt (except 
>> for wasting storage space and processing time). But I am resentful 
>> that many of my great-sounding masters that I have worked so 
>> carefully to make them sound better, including upsampling before 
>> processing --- have to be downsampled in order to be released on HD 
>> tracks because of the high res mafia.
>>
>>
>> BK
>>
>>

-- 
Richard L. Hess                   email: richard at richardhess.com
Aurora, Ontario, Canada           http://www.richardhess.com/
http://www.richardhess.com/tape/contact.htm
Quality tape transfers -- even from hard-to-play tapes.