Sound differences

[Kingpin]

Hello,

1. We have a discussion on another board about sound differences during transcoding.
Some people hear differences when playing alac or flac as wav to the streamer and between the original wav file?

2. Next question is that they hear differences when transcoding to wav24.

How is this possible?

Because alac/flac are both lossless, when streaming as pcm to the server it shouldn't make a difference to the original wav file?
Next thing is why upsample to wav24? It doesn't make sense to me because you can't add anything what isn't there. The 8 bit difference will be filled with '0'!
Can you explain to me how this is possible?

Does minimserver anything with the files (like upsampling to 96khz(for example)?

[simoncn]

(05-02-2013 12:33)Kingpin Wrote:  Hello,

1. We have a discussion on another board about sound differences during transcoding.
Some people hear differences when playing alac or flac as wav to the streamer and between the original wav file?

2. Next question is that they hear differences when transcoding to wav24.

How is this possible?

Because alac/flac are both lossless, when streaming as pcm to the server it shouldn't make a difference to the original wav file?
Next thing is why upsample to wav24? It doesn't make sense to me because you can't add anything what isn't there. The 8 bit difference will be filled with '0'!
Can you explain to me how this is possible?

Does minimserver anything with the files (like upsampling to 96khz(for example)?

MinimServer doesn't resample. It just fills the low-order bits with zeros.

I agree with you that it seems very surprising that this should make a difference. I didn't believe it until I had heard it myself consistently on a lot of different source material. Some other forum members have also heard this difference.

I don't have a technical explanation for this. My guess is that there is some difference within the Linn DS between how 16-bit inputs and 24-bit inputs are processed. I haven't been able to try it with any other renderer.

Have you tried it yourself? Can you hear a difference?

[Kingpin]

Thanks for your answer.

(05-02-2013 13:14)simoncn Wrote:  MinimServer doesn't resample. It just fills the low-order bits with zeros.

I agree with you that it seems very surprising that this should make a difference. I didn't believe it until I had heard it myself consistently on a lot of different source material. Some other forum members have also heard this difference.

Me as a technician and know how digital data works are very suprised that it makes a difference, because the filled gap (8 bits) of zero data '0' is nothing that could make a difference when decoding.

Quote:I don't have a technical explanation for this. My guess is that there is some difference within the Linn DS between how 16-bit inputs and 24-bit inputs are processed. I haven't been able to try it with any other renderer.

That could be an explanation, but still ... then there should be an difference in the Linn for example when converting wav to pcm.

Quote:Have you tried it yourself? Can you hear a difference?

No, but i'm using another software which can convert different formats straight to pcm on the server.
I do this to export the file straight to the dac without having my streamer decoding the file into pcm.
And there are no hearing differences here, that's why i'm curious how this is possible with Minimserver.

[simoncn]

(05-02-2013 13:37)Kingpin Wrote:  That could be an explanation, but still ... then there should be an difference in the Linn for example when converting wav to pcm.

The Linn DS doesn't support direct PCM input.

Quote:No, but i'm using another software which can convert different formats straight to pcm on the server.
I do this to export the file straight to the dac without having my streamer decoding the file into pcm.
And there are no hearing differences here, that's why i'm curious how this is possible with Minimserver.

Is the DAC attached to the server? If so, what interface (e.g., USB, FireWire, S/PDIF) does it use?

[Kingpin]

(05-02-2013 15:14)simoncn Wrote:  The Linn DS doesn't support direct PCM input.

Yes i know, i meant converting in the Linn.
Into the Linn the received wav file will be converted to pcm which transports the pcm to the internal dac of the Linn for conversion to analog output.

Quote:Is the DAC attached to the server? If so, what interface (e.g., USB, FireWire, S/PDIF) does it use?

No.
My server converts to pcm and streams it to my streamer.
My streamer doesn't do anything with it only graps the meta data for displaying the tag info on the display.
The pcm file goes through the spdif to my external dac and from there the pcm file will be converted to the analog output.

[simoncn]

(05-02-2013 15:44)Kingpin Wrote:  Yes i know, i meant converting in the Linn.
Into the Linn the received wav file will be converted to pcm which transports the pcm to the internal dac of the Linn for conversion to analog output.

It appears the internal conversion in the Linn DS has sound differences (audible to some people) between the following combinations:
FLAC (16-bit) => PCM
WAV (16-bit) => PCM
WAV (24-bit, zero padded) => PCM

I don't have a technical explanation for this.

Quote:No.
My server converts to pcm and streams it to my streamer.
My streamer doesn't do anything with it only graps the meta data for displaying the tag info on the display.
The pcm file goes through the spdif to my external dac and from there the pcm file will be converted to the analog output.

OK, I understand. This means that the combination of S/PDIF interface and DAC internal decoder in your setup is treating 16-bit data and zero-padded 24-bit data the same, as would be expected.

[Alekz]

Even more interesting. Different servers (the software that is) sound differently. (See http://www.psaudio.com/vanilla/discussio...an-psa-emm)

So far (almost?) everybody who tried both PS Audio eLyrics Music Manager (EMM) and MinimServer found that Minim sounds brighter than EMM. It's neither bad not good, but the difference is there and can be clearly heard.

Also no technical explanation.

The "usual suspects" are: buffering, flow control and conversion methods. For example, the sound quality depends on the server load, that might affect the flow. There is even a new term for that: "software jitter".

[Alekz]

About conversion methods. It's interesting to compare checksums. BTW, they might differ, because of different tagging methods/conversion.

[simoncn]

(05-02-2013 16:42)Alekz Wrote:  Even more interesting. Different servers (the software that is) sound differently. (See http://www.psaudio.com/vanilla/discussio...an-psa-emm)

So far (almost?) everybody who tried both PS Audio eLyrics Music Manager (EMM) and MinimServer found that Minim sounds brighter than EMM. It's neither bad not good, but the difference is there and can be clearly heard.

Also no technical explanation.

The "usual suspects" are: buffering, flow control and conversion methods. For example, the sound quality depends on the server load, that might affect the flow. There is even a new term for that: "software jitter".

In the MinimServer vs. EMM comparisons where MinimServer sounded "brighter", was MinimServer doing wav24 transcoding? Was EMM also doing 24-bit WAV transcoding?

Does the same "brightness" difference apply when comparing MinimServer streaming native FLAC and EMM streaming native FLAC?

[Kingpin]

(05-02-2013 16:42)Alekz Wrote:  Even more interesting. Different servers (the software that is) sound differently. (See http://www.psaudio.com/vanilla/discussio...an-psa-emm)

So far (almost?) everybody who tried both PS Audio eLyrics Music Manager (EMM) and MinimServer found that Minim sounds brighter than EMM. It's neither bad not good, but the difference is there and can be clearly heard.

Also no technical explanation.

The "usual suspects" are: buffering, flow control and conversion methods. For example, the sound quality depends on the server load, that might affect the flow. There is even a new term for that: "software jitter".

That's scary!
I can realy go in detail but technically it shouldn't be different, because the files itself have CRC, when converting they should be identical, you can do the test yourself and compare both formats.

Software jitter on digital streams ...... your stream must be that distorted that even the streamer can't rearrange the software packages internally.
It's already digital, a distorted square wave is already recognised as a logical '1' if a small top of the square is 2.5V.
Like a zip file downloaded from around the world would be received bitperfect on my computer, because if for some reason the packages are distorted it's over and out, the file won't work anymore because the checksums are different.
So i don't believe that software jitter will that have much impact.

Like simonsc said, it could be the Linn's internally.

Server load shouldn't be an issue.
See my simple Qnap TS-119 NAS converting a 24/96 flac file to pcm and sending to my streamer is hardly using processor power or ram use.

CPU


RAM


RESOURCES