• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Is Amazon Ultra HD up-sampled?

cpt_paranoia said:
Using an upsampling filter. You repeat, or zero-stuff the existing samples, and then apply an interpolation filter to smooth between the new samples. For non-binary interpolation factors, you can upsample and then downsample, with integer rates to give the final upsample rate you need.

You can do this with fixed input and output sample rates. You can even do it when you are not sure of the exact ratio between sample rates (i.e. when source and destination clocks are different); you use an ASRC filter, as provided in a number of SPDIF interface devices.

This is not new; the first Philips CD players used the 14-bit TDA1540 DAC, which required a 4x upsampling filter (the SAA 7030) to achieve 16-bit equivalent performance.
Click to expand...
Are you saying that upsampling can provide an identical (equivalent) result? If so, why wouldn't streaming services simply upsample all of their content to at least 24/96 and market their themselves as "100% super-duper, extra high resolution"?

I think the assumption of most users is that 24/96 and 24/192 provide more information, vs 16/44, and therefore have at least the potential to sound better, all other things being equal. Upsampling certainly results in more bits, but not more information, right? If so, how does one know if a file has just been enlarged via upsampling, or actually does contain more information?
 
Last edited:
evoltap said:
Good to know. I didn't submit the release in question, which was Tunecore
Click to expand...

Interestingly, Tunecore not only accepts 24/192, they recommend it. Maybe they did the upsampling. Or maybe whoever did submit it followed Tunecore's recommendations and upsampled it prior to submission?


In any case, the underlying issue is whether or not some/all of what the streaming services present as 24/96 and 24/192 is actually just upsampled 16/44?
 
Last edited:
KentTeffeteller said:
Your display then is your Provenance! It's telling you the sampling rate. 16/44 is common. Lots of old never remastered content which is 16/44. Not unusual for 1980's and early 1990's digital content which never got a new remastering.
Click to expand...
Provenance normally refers to something's origins, rather than its current state. If an upsampled 16/44 file is displayed as 24/96, then the display is not indicative of its true provenance.
 
spark1 said:
Are you saying that upsampling can provide an identical (equivalent) result?
Click to expand...
Equivalent to what? Given that most modern DACs upsample themselves as part of their conversion process, upsampling from CD quality to 192/24 just to claim 'hires' is pretty pointless, and just uses bandwidth unnecessarily. But it's nice marketing...

Upsampling cannot restore information lost when downsampling from, say, a studio master of 192/24 to 44.1/16, or if the master was only ever recorded at 44.1/16.
 
cpt_paranoia said:
Equivalent to what? Given that most modern DACs upsample themselves as part of their conversion process, upsampling from CD quality to 192/24 just to claim 'hires' is pretty pointless, and just uses bandwidth unnecessarily. But it's nice marketing...

Upsampling cannot restore information lost when downsampling from, say, a studio master of 192/24 to 44.1/16, or if the master was only ever recorded at 44.1/16.
Click to expand...
Equivalent to the same recording released at 24/96. My working assumption, which could well be wrong, is that most digital recordings are captured at higher resolution than 16/44, and then reduced during mastering. If true, the same recording could be released at both the higher resolution and 16/44.

If instead the 16/44 was to be upsampled to 24/96, it's content would still be that of the 16/44 version, right?

And in this case, I would consider it fraud to represent the 24/96 version as anything other than 16/44, even though it technically is indeed a 24/96 file.

Is my thinking right about this?
 
Last edited:
spark1 said:
Equivalent to a higher res
Click to expand...
That's arguable; there are those who insist that the ability to reproduce ultrasonics is important, regardless of what Whittaker, Nyquist & Shannon have to say.

But 44k1/16 upsampled to 192/24 is not equivalent to a genuine 192k/24 recording, unless that used anti-alias sampling filters for 44k1; there will be ultrasonic content that cannot be legitimately present in the upsampled recording.
 
This has been an interesting thread. From what I understand, there was at least one intermediary between the final master and submission to Amazon, so no proof that Amazon upsampled. Was the album submitted to any other streaming services and, if so, how do those sample rates compare?

As I am new to the streaming world, I'm pretty much a rube. It's been about 6 weeks since I got a WiiM Pro and I've been very happy with Amazon HD.
 
Mingo said:
This has been an interesting thread. From what I understand, there was at least one intermediary between the final master and submission to Amazon, so no proof that Amazon upsampled. Was the album submitted to any other streaming services and, if so, how do those sample rates compare?

As I am new to the streaming world, I'm pretty much a rube. It's been about 6 weeks since I got a WiiM Pro and I've been very happy with Amazon HD.
Click to expand...

It would be great if @evoltap could clarify the specifics regarding the handling of the files in question.

As I currently understand it, and in contrast to the initial information provided, @evoltap engineered the album but did not do the submission to Tunecore. I am unclear as to whether or not he did the mastering, and whether or not whoever did the submission might have either produced a true 24/192 version or upsampled a lower bitrate version prior to submission.

As I mentioned earlier, Tunecore certainly accepts 24/192 files (and in fact recommends them). As another link in the "chain of custody", it is also possible that Tunecore upsampled what they were given.

I no longer have Qobuz or Tidal accounts, so cannot check to see what bit depth and sample rate they are showing.

While it may well be that Amazon is misrepresenting at least some of their high-res content, I'd like to get more details before coming to that conclusion. I believer that Amazon does apply some forms of processing to the content they are offering on their streaming platform (if I remember correctly, @cpt_paranoia confirmed this a while back in another thread, and I certainly wish that Amazon was forthcoming about the specifics), but upsampling 16/44 files to present them as high-res (high def and ultra high def, in their parlance) would be egregious, at least in my book.

EDIT: Given @cpt_paranoia's wizardry in these matters, I wonder if he might be willing and able to do an FFT analysis of the data in question to determine if it's really 24/192 or just upsampled?

And maybe a Qobuz or Tidal user could report in with info about how those services are presenting it.
 
Last edited:
KentTeffeteller said:
As earlier stated, the high bit rates used in tracking and mixing, enable engineers in tracking, mixing, and mastering to use their plug-ins and DAW effects without audible artifacts. This is sensible and standard practice in the industry.
Click to expand...
Yes, that was me that said that earlier-- but about sample rate, not bit depth. If we were to high fidelity audio into two things, it is the battle against unwanted noise, and getting full and flat human hearing frequency response. Bit depth gives more dynamic range over the noise floor, and sample rate gives varying frequency response. Back in the 16 bit days in the studio, you were always taught to record "hot", because there was less bit depth before the noise floor. Now that we all record 24 bit or 32 bit float, it's not an issue to record something at say -20dbfs and need to boost it 15db in a mix. It's sample rate that has an effect of processing, with higher rates being better for not throwing artifacts back down into the audible range.
 
Mingo said:
This has been an interesting thread. From what I understand, there was at least one intermediary between the final master and submission to Amazon, so no proof that Amazon upsampled. Was the album submitted to any other streaming services and, if so, how do those sample rates compare?

As I am new to the streaming world, I'm pretty much a rube. It's been about 6 weeks since I got a WiiM Pro and I've been very happy with Amazon HD.
Click to expand...
The album was submitted to Tunecore at 16-44. I can only comment on Apple Music and Spotify. Apple lists it as Lossless with their little logo, Spotify doesn't say ****. I supposed I could monitor the bitrate while streaming it to see....
 
spark1 said:
It would be great if @evoltap could clarify the specifics regarding the handling of the files in question.

As I currently understand it, and in contrast to the initial information provided, @evoltap engineered the album but did not do the submission to Tunecore. I am unclear as to whether or not he did the mastering, and whether or not whoever did the submission might have either produced a true 24/192 version or upsampled a lower bitrate version prior to submission.

As I mentioned earlier, Tunecore certainly accepts 24/192 files (and in fact recommends them). As another link in the "chain of custody", it is also possible that Tunecore upsampled what they were given.

I no longer have Qobuz or Tidal accounts, so cannot check to see what bit depth and sample rate they are showing.

While it may well be that Amazon is misrepresenting at least some of their high-res content, I'd like to get more details before coming to that conclusion. I believer that Amazon does apply some forms of processing to the content they are offering on their streaming platform (if I remember correctly, @cpt_paranoia confirmed this a while back in another thread, and I certainly wish that Amazon was forthcoming about the specifics), but upsampling 16/44 files to present them as high-res (high def and ultra high def, in their parlance) would be egregious, at least in my book.

EDIT: Given @cpt_paranoia's wizardry in these matters, I wonder if he might be willing and able to do an FFT analysis of the data in question to determine if it's really 24/192 or just upsampled?
Click to expand...
Sorry, I missed this. Chain of custody was this: I tracked and mixed it at 24-48. Sent it to Golden Mastering at 24-48, they returned the masters at 16-44 and 24-48 mFIT. I sent the 16-44 to a manager, who then uploaded to tunecore. He would not have touched them (as far as upsampling). I'm kicking myself for not checking tunecore's standards, but it wasn't that long ago that I checked-- and everybody was still at 16-44. My consolation is that I think 16-44 PCM is still very much hifi.
 
Also in regard to upsampling, my M.O. is usually that is has no negative effect-- but I've never checked, it just usually sounds fine and no audible change. I've even seen a well known mastering engineer say that he upsamples up to 96k for processing purposes. However, it's probably does matter what algo is used. I will do test and upsample a track from 48 to 96 in Logic, and then perform a null test. If it nulls perfectly, that answers that.
 
evoltap said:
Sorry, I missed this. Chain of custody was this: I tracked and mixed it at 24-48. Sent it to Golden Mastering at 24-48, they returned the masters at 16-44 and 24-48 mFIT. I sent the 16-44 to a manager, who then uploaded to tunecore. He would not have touched them (as far as upsampling). I'm kicking myself for not checking tunecore's standards, but it wasn't that long ago that I checked-- and everybody was still at 16-44. My consolation is that I think 16-44 PCM is still very much hifi.
Click to expand...

I don't think this is about whether or not 16/44 is "hi-fi" (which it most definitely is!).

The issue is whether or not Amazon upsampled your files so as to present them as UHD (24/192).

You've clarified that you handed off 16/44 and 24/48 masters to a third party, who then handed them (or perhaps just the 16/44?) to Tunecore, which in turn provided it/them to Amazon HD (and assumedly other services as well).

So, there are three possibilities:

  1. The manager upsampled the file(s)
  2. Tunecore upsampled the file(s)
  3. Amazon upsampled the file(s)

Hopefully someone with a Qobuz or Tidal account will take a look to see how they show up on those services, as that would be a good clue as to how this happened.
 
Last edited:
evoltap said:
Also in regard to upsampling, my M.O. is usually that is has no negative effect-- but I've never checked, it just usually sounds fine and no audible change. I've even seen a well known mastering engineer say that he upsamples up to 96k for processing purposes. However, it's probably does matter what algo is used. I will do test and upsample a track from 48 to 96 in Logic, and then perform a null test. If it nulls perfectly, that answers that.
Click to expand...

That would be interesting, but it's not really an issue of whether or not upsampling somehow degrades the quality.

I'm just really curious as to whether or not Amazon is "guilty as charged", and would like to get to the bottom of it!
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom