Upsampling

[Nenon]

I've always wondered about this one.

 

Can the people who do upsampling explain to me the benefits of upsampling? I guess it's a DAC dependent thing, and it might work better on some DACs than others. I guess it can also make the sound worse with some DACs. I also guess that the benefit of upsampling is for DAC architectures that upsample everything inside the DAC before converting to analog. And it's only a benefit to upsample from your music server, if the music server is doing it more efficiently (i.e. the DAC having to do less upsampling work results in better overall sound quality). Can someone list DACs that sound better with upsampled music? I would like to study their architecture. 

 

The other way I imagine upsampling would be necessary or work better is for DACs that only do DSD. Since DSD is so much easier to convert to analog than PCM, some manufacturers do pure DSD DACs and stick a chip before the DAC to convert PCM to DSD. I am guessing software like HQPlayer on a proper music server can potentially do a better job of converting PCM to DSD, and if the DAC receives DSD signal and bypasses the PCM to DSD conversion chips, I can see how it can sound better.  

 

But if that's the case, why don't people process their music files and upsample them in advance, so it does not need to be done in realtime while playing? That obviously would not work well for streaming but why not for local music? Has anyone who is upsampling tried to process/upsample the music files in advance, and is that an advantage than doing it in realtime.

 

In a way, I am relating upsampling to realtime vs. bitperfect CD ripping. You can try to extract everything from a CD in realtime, but all those microscratches on the CD can result in some lost bits. Or you can enable bitperfect and let the ripping software take its time to extract every single bit. Wouldn't upsampling files in advance result in better performance? 

 

I would be glad to hear more on the topic from experienced people. 

 

 

[Ralf11]

one benefit is the ability to then use a gentle filter slope

 

many (most?) DACs upsample

[mansr]

https://troll-audio.com/articles/sampling-and-reconstruction/

[rickca]

4 hours ago, Nenon said:

I've always wondered about this one.

@Miska has been answering these kind of questions for years in the HQPlayer forum.  But I don't think you want to crawl through 600 pages of posts!  I've been following Jussi's posts for a long time because he does an excellent job of answering intriguing questions like yours.

 

Did you want to hear from users rather than from the developer?

[Nenon]

7 hours ago, rickca said:

@Miska has been answering these kind of questions for years in the HQPlayer forum.  But I don't think you want to crawl through 600 pages of posts!  I've been following Jussi's posts for a long time because he does an excellent job of answering intriguing questions like yours.

 

Did you want to hear from users rather than from the developer?

 

Thanks. I was never interested in the HQPlayer forum, but I will dig in... 

[Nenon]

9 hours ago, Ralf11 said:

one benefit is the ability to then use a gentle filter slope

 

many (most?) DACs upsample

 

Thank you. I was more interested about why people find that upsampling in software sounds better. And if that's the case, why don't they process/upsample their local library in advance and play the upsampled files instead. But I will look for more info in the HQPlayer forum as @rickca suggested. I use a NOS DAC and don't look to change that but am interested to broaden my knowledge on the topic. 

[rickca]

Another thread that may be of interest to you.  It's old so the technology has evolved but the ideas are still the same.

https://audiophilestyle.com/forums/topic/20409-offline-upsampling/

This only covers your interest in offline upsampling, not all aspects of your question.

[4est]

4 hours ago, Nenon said:

 

Thank you. I was more interested about why people find that upsampling in software sounds better. And if that's the case, why don't they process/upsample their local library in advance and play the upsampled files instead. But I will look for more info in the HQPlayer forum as @rickca suggested. I use a NOS DAC and don't look to change that but am interested to broaden my knowledge on the topic. 

What is your DAC btw?

 

It would seem as if you grasp some of the basics, but might not be connecting all of the dots. To begin with, there is nothing wrong with upsampling beforehand other than the file sizes can be huge depending upon what you upsample to. Beyond that, the only real drawback is that you cannot adjust the algorithm unless you add/switch copies.

 

The programs I use to upsample(on the fly) are Audirvana and HQPlayer. They both have adjustments to the upsampling filters and modulators and can use streams from Tidal and Qobuz as well as your files. HQPlayer in particular is very robust in this regard. In essence, one is using the greater computational power of a computer to run more accurate/intensive upsampling whilst reducing the on chip noise created by that process in a DAC.

 

There are many things in audio that might be considered woo woo. YMMV, but this is demonstrably real. One might prefer what their DAC does btw. There are no perfect filters and it can be taste, recording and equipment dependent.

 

[jabbr]

6 hours ago, Nenon said:

I use a NOS DAC and don't look to change that but am interested to broaden my knowledge on the topic. 

 

NOS DACs are a perfect fit with software upsampling. The reason to do software upsampling on a high powered computer is that the filters are better than those built into a DAC chip. Do you feed CD redbook 16/44.1 into your NOS DAC?

[church_mouse]

I hope Nenon will not think me rude if I raise this adjunct to Nenon's original query.  I have always wondered about the technical merits of the following upsampling scenario (which happens to be my case):

My output to my DAC is limited to 192 (firewire driver, no USB drivers anymore), but I understand that the chip in my DAC (original Mytek Manhattan) upsamples everything to 384 and then downsamples. 

 

If I am correct in my belief of what the chip in my Mytek will do (and I may have misunderstood).  Is there any potential benefit from upsampling in the computer to 192 if it will get further upsampled anyway?

 

[Nenon]

Thank you for all the replies.

 

4 hours ago, 4est said:

What is your DAC btw?

Audio Note DAC 5 Signature with a few mods. 

 

2 hours ago, jabbr said:

Do you feed CD redbook 16/44.1 into your NOS DAC?

Probably more than half of my digital music is bit perfect ripped CDs. 

[jabbr]

I don’t know the details of the Audio Note input stage except that it employs a AD1865 R2R DAC chip ... this chip allows 2x ... 16x Fs input where Fs is 44.1kHz. 

 

I will let @PeterSt discuss his extensive experience using a similar NOS R2R DAC chip (PCM1704) but you really should get a copy of his XXHE and listen for yourself. XXHE is a tremendous bargain!

[Ralf11]

church-mouse, have you contacted Mytek on that question?

[church_mouse]

@Ralf11

 

I tried in the past, but without reply.  My experience of Mytek support has not been impressive - perhaps not helped by my being in Europe.