16/44 vs 24/96. Which sounds better in streaming ?

[audiobomber]

I don't see any problem upsampling/oversampling a16/44.1 recording to 24/88.2, 24/176.4, or DSD, as long as it is done properly, which should be easy to do for anyone involved in production. Upsampling to 24/48 or 24/192, OTOH, requires transcoding, which should be avoided where possible because it requires interpolation.

From my personal testing, hi-rez beats 16/44.1, if it is from the same master and well implemented. Unfortunately, a lot of the higher resolution recordings have been remastered. Remastering can be an improvement, but too frequently it's a change for the worse.

 

[ecwl]

I think in general, if both files/streams are from the same master and the original was in 24/96 and the 16/44 is a downconverted version, the 24/96 should sound better.

However, there are simply too many variables in digital audio (or audio in general). For example, if your streamer is noisy, you could potentially be feeding more noise from the 24/96 track to the DAC compared to the 16/44 track in which case, you're going to get a leaner and more artificial sound with 24/96. It is also possible that a different filter was selected for 16/44 vs 24/96 within the DAC, in which case, you're just hearing that the DAC filter used for 24/96 sounds worse than the DAC filter used for 16/44. I doubt the latter is the case here but you never know. Some DACs store the filter choice based on whether a track is 44kHz vs 96kHz whereas other DACs would always use the same filter you last selected for all tracks.

Ultimately, I think people should just listen to what sounds good to them. Diving into the technical details and reasoning probably don't contribute much to the enjoyment of this hobby anyway.

 

[Ian B]

This interesting, and as others have said, it probably all depends. I will generally prefer hi rez of the mastering is the same, but some upsampling is hit or miss depending on the algorithm, and how much noise it puts into the system. I tend to like DSD DACs.

 

[AdamZ]

Everyone who talks about sound quality here should ask themselves the following questions:
- how old am I?
- when was the last time I went to an ear doctor and/or had a hearing test.
This will bring you a little closer to reality.
-

 

[Ian B]

Everyone who talks about sound quality here should ask themselves the following questions:
- how old am I?
- when was the last time I went to an ear doctor and/or had a hearing test.
This will bring you a little closer to reality.
-

Early 40s, spent half my life as a professional audio engineer, can hear up to around 16khz, and see no value in people who are needlessly insulting, self-absorbed, and pedantic.

 

[AdamZ]

Congratulations.
But if you look at the statistics, that's rather unlikely.
But hey, you and your doctor will have tested it thoroughly. Especially since it's your job.

Yes, I've met people like that too.
And arrogant ones. But I don't necessarily have to talk to them.

 

[ationg]

I am new to streaming (Tidal in Roon). I compared the same titles there with different versions ..and I find the 16/44 usually sound better and more engaging than the hi-res versions (24/96 and higher). The hi-res versions have a wider soundstage but sound a bit lean and artificial to my ears. I don't know the reasons why.
Do most people stream in native resolution or hi-res?

After one year and admittedly still limited experience of streaming, I found in terms of SQ is CD ripped files > 16/44 downloads > hi def downloads > 16/44 Tidal > hi def Tidal. Physical CD playback sounds better than all the above. And vinyl still sounds better than physical CD. Streaming offers unmatched convenience obviously.

 

[Ian B]

After one year and admittedly still limited experience of streaming, I found in terms of SQ is CD ripped files > 16/44 downloads > hi def downloads > 16/44 Tidal > hi def Tidal. Physical CD playback sounds better than all the above. And vinyl still sounds better than physical CD. Streaming offers unmatched convenience obviously.

One tip is using WAV files for downloads. It does narrow the gap for both CD quality and Hi-Res somewhat. CDs use WAV files. There is always going to be the difference between disc transports and servers/streamers.

 

[Kjetil]

My system is digital, running at 96 kHz. That means that all my CD rips are transcoded and I have to trust what the MiniDSP Flex Eight does. Qobus has a lot of «my» music in 48 kHz multiples so that’s fine. When I buy downloads I get 96 or 192 if available.
I just bought my first SACD player, a like new Yamaha CD-S2100. This goes into my MiniDSP ADept analog front end which samples at 24/96. This, in addition to hear the SACD layer of my hybrids for the first time, gives me the possibility to compare my CDs to my ripped CDs. Lovely rabbit hole.

 

[jeromelang]

My system is digital, running at 96 kHz. That means that all my CD rips are transcoded and I have to trust what the MiniDSP Flex Eight does. Qobus has a lot of «my» music in 48 kHz multiples so that’s fine. When I buy downloads I get 96 or 192 if available.
I just bought my first SACD player, a like new Yamaha CD-S2100. This goes into my MiniDSP ADept analog front end which samples at 24/96. This, in addition to hear the SACD layer of my hybrids for the first time, gives me the possibility to compare my CDs to my ripped CDs. Lovely rabbit hole.

Yamaha CD-S2100 invert absolute polarity

 

[Kjetil]

Indeed, so I swapped pin 2 and 3 in one end when I made the cables for it. Very nerdy.

 

[jeromelang]

Indeed, so I swapped pin 2 and 3 in one end when I made the cables for it. Very nerdy.

Only if it is at the source end.

 

[Kjetil]

What do you mean by that?

 

[Ian B]

My system is digital, running at 96 kHz. That means that all my CD rips are transcoded and I have to trust what the MiniDSP Flex Eight does. Qobus has a lot of «my» music in 48 kHz multiples so that’s fine. When I buy downloads I get 96 or 192 if available.
I just bought my first SACD player, a like new Yamaha CD-S2100. This goes into my MiniDSP ADept analog front end which samples at 24/96. This, in addition to hear the SACD layer of my hybrids for the first time, gives me the possibility to compare my CDs to my ripped CDs. Lovely rabbit hole.

If you can, play the SACDs natively because 96khz is actually a significant step down from the original DSD.

 

[Kjetil]

I cannot*. My system runs on 96 kHz. Last time i checked DSD64 and 24/96 is approximately the same resolution.

*I could of course feed the single ended output of the SACD directly to a headphone amp.

 

[Ian B]

I cannot*. My system runs on 96 kHz. Last time i checked DSD64 and 24/96 is approximately the same resolution.

*I could of course feed the single ended output of the SACD directly to a headphone amp.

The size of the files is the same, but the formats aren't really comparable. Even converting DSD64 to 24/352khz (which was supposed to be an equivalent production format) loses something. It's worth listening to the analog output of the SACD player when you can. It will be different.

 

[mtemur]

CDs use WAV files. There is always going to be the difference between disc transports and servers/streamers.

A CD doesn’t contain WAV files. It uses the CDDA (Compact Disc Digital Audio) format, which is an encoded digital audio format — neither WAV nor AIFF is a direct equivalent. When a CD is ripped to an uncompressed format like WAV or AIFF, there’s always some level of decoding, and depending on the software, possibly even re-encoding involved. So while the resulting files may be uncompressed, they are not exact copies of what’s on the CD.

Even converting DSD64 to 24/352khz (which was supposed to be an equivalent production format) loses something.

That doesn’t mean DSD is superior to 24/352 kHz PCM. Even processing 16/44.1 to the same resolution — or upsampling it — causes some loss in sound quality. That’s because digital processing always leaves its mark on the sound.

To make a fair comparison between 1x DSD and 24/96 PCM, an analog source must be independently encoded into both formats. Only then can the two files be compared side by side in a meaningful way.

 

[Ian B]

A CD doesn’t contain WAV files. It uses the CDDA (Compact Disc Digital Audio) format, which is an encoded digital audio format — neither WAV nor AIFF is a direct equivalent. When a CD is ripped to an uncompressed format like WAV or AIFF, there’s always some level of decoding, and depending on the software, possibly even re-encoding involved. So while the resulting files may be uncompressed, they are not exact copies of what’s on the CD.


That doesn’t mean DSD is superior to 24/352 kHz PCM. Even processing 16/44.1 to the same resolution — or upsampling it — causes some loss in sound quality. That’s because digital processing always leaves its mark on the sound.

To make a fair comparison between 1x DSD and 24/96 PCM, an analog source must be independently encoded into both formats. Only then can the two files be compared side by side in a meaningful way.

I stand correct on CDDA. Bottom line though, uncompressed without added metadata (like AIFF has). Sounds closest to CD that I've heard.

I have done a lot of direct comparisons with DSD64 and 24/96, including stuff I recorded myself from analog. Single rate DSD is not without flaws (double rate should have been the standard), but it is capable of things PCM can't really touch in terms of speed, bass, low-level detail, and 3D imaging, and with a really good DAC it's leaps ahead. PCM may sound taller and more defined in the treble, but also stiffer, thinner, and more flat. They may have the same data rate, but the sound isn't equivalent.

That's why I say, best to listen to SACDs natively because the value is in the direct DSD to analog.

 

[Kjetil]

That’s what I do except that afterwards the analog signal is sampled at 24/96. If not I cannot listen to it though my speaker system.

 

[Balle Clorin]

To make a fair comparison between 1x DSD and 24/96 PCM, an analog source must be independently encoded into both formats. Only then can the two files be compared side by side in a meaningful way.

My streamer presents DSD and 16/41k or whatever at a different output level. Any comparison becomes invalid without level matching. Do peOple consider the possibility of unequal level when comparing?