timster

Replay Gain and Transcoding. Does Katalyst/Akurate make it bollox?

Recommended Posts

Oh dear oh dear. I have been using replay gain (RG) through both Asset and Minimserver for some time now. When I started using it, I had a Majik Exaktbox-I and to be honest I could not detect a difference in SQ, and the volume normalisation is very useful for me (I tend to use Asset's randomiser or my own playlists a lot of the time). I did do A/Bs at the time and like I said, I couldn't detect any discernible deterioration.

I then changed to an Akurate Exaktbox last year, and earlier this year that went Katalyst. Silly me, I didn't bother doing an A/B for replay gain again for either upgrade.

The last few weeks I have been occasionally streaming my purchases from Qobuz (only those cos I'm not a subscriber) rather than from my NAS. I have never bothered before, but it has been more convenient at times lately. But - I have since been sure the Qobuz stream was sounding better than my  local copies with replay gain applied (and transcoded to WAV/24). This can't be so I thought.

So this weekend I have done comparisons of some of my purchased tracks between:

1. Qobuz stream
2. Asset with RG
3. Asset direct source flac
4. Minimserver with RG
5. Minimserver with transcoding to WAV/24
6. Minimserver direct source flac.

Results are that RG is bloody awful! Goddammit!

In order of best to worst of the above:

3 & 6
1
5
2 & 4

Lesson learned! But I wonder if Katalyst (or even just Akurate level  Exakt) is showing this, while the Majik level Exakt didn't?

Share this post


Link to post
Share on other sites

My results would be different to the above. I find that transcoding to WAv24 from standard FLAC (16 bit or Hi-Res) gives resultants consistent with a slight lowering of the noise floor.. This result did not change when I upgraded to Exakt and later to Katalyst. I don't use ReplayGain.

David

Share this post


Link to post
Share on other sites

Can't be Katalyst. I have a KDS/3 with all good specs. I do A/B compare using specially prepared files. Make a copy of an album (flac), edit it with mp3tag and set RG to all 0 (zero) manually. Then load a playlist with songs coming from my Twonky server (no RG) and then from the source with RG = 0 with Minim, transcoding to WAV 24 and RG enabled. Both sound equally loud and I cannot detect any degrade of sound with RG on. Rather the opposite but only very small difference, probably influenced by personal perception. 

However I am the kind of bad listener where louder appears better for me. I can check that when loading the playlist from Twonky on one hand and Minim with an applied RG non-zero. The Twonky songs are louder (approx 6 dB) and in a blind testing I would likely prefer the louder over the quieter. A shme I know but it in the head o.O

Share this post


Link to post
Share on other sites
2 hours ago, DavidHB said:

My results would be different to the above. I find that transcoding to WAv24 from standard FLAC (16 bit or Hi-Res) gives resultants consistent with a slight lowering of the noise floor.. This result did not change when I upgraded to Exakt and later to Katalyst. I don't use ReplayGain.

David

Hmm. My top three/4 are all so close I can't say either way for definite what is top . But for sure RG is terrible. 

1 hour ago, golke53 said:

set RG to all 0 (zero) manually. Then load a playlist with songs coming from my Twonky server (no RG) and then from the source with RG = 0 with Minim, transcoding to WAV 24 and RG enabled. Both sound equally loud and I cannot detect any degrade of sound with RG on. Rather the opposite but only very small difference, probably influenced by personal perception. 

As you aren't applying any gain, I imagine the effect when passed through the transcoder is the same with/without RG enabled. 

Share this post


Link to post
Share on other sites
3 hours ago, DavidHB said:

I find that transcoding to WAv24 from standard FLAC (16 bit or Hi-Res) gives resultants consistent with a slight lowering of the noise floor..

OK, on further analysis and closer listening,  yes I can hear this. So transcode to WAV24 it is. That has always meant the audio quality of the original is hidden from view which slightly bugged me, so I have added minim's #AudioQuality special tag to the album display format :)

Share this post


Link to post
Share on other sites

Where, how do you add this special tag please, it's bugging me too.

Share this post


Link to post
Share on other sites
45 minutes ago, Jail4CEOs2 said:

Where, how do you add this special tag please, it's bugging me too.

As long as you have #AudioQuality in either indexTags or itemTags, then in tagFormat, add: Album.displayFormat={$album^$#AudioQuality^ [^]}

Share this post


Link to post
Share on other sites

Just my two pennies' worth - I am sure people will correct me if I am totally wrong.

I have not tried Replay Gain, although I have tried converting to WAV in various evolutions of the ADS.

I have found an audible difference - the WAV seems to be slightly sweeter, a wider sound stage, but lacks just a tiny bit of punch.  All very subjective.  Not night and day.

But surely whether any of these modifications to the audio file listed on this thread are conducted by the processor, memory and other junk in the NAS.

So I would expect people's experiences to be variable and mainly dependent on what is going on inside the NAS.  For example it is understood in the Melco community that the sound is improved if it is run in "dark" mode - no display and with as few processes running as possible (which is why it is always wise to run only one server software at a time).

It could be that in a lot of systems it is swings and roundabouts.  It could be argued that any decent streamer should be up to the task of stripping down a FLAC file without any trade off in performance.  The advantages of spoon-feeding it a WAV file could be slight.  Especially if there is a corresponding detrimental burden on the circuitry of the NAS.  And restructuring a file at a different volume level could be onerous.

Just thoughts...

Donuk beautiful downtown York

Share this post


Link to post
Share on other sites
22 hours ago, Donuk said:

I have found an audible difference - the WAV seems to be slightly sweeter, a wider sound stage, but lacks just a tiny bit of punch.  All very subjective.  Not night and day.

I can't say that I have detected any lack of punch,; otherwise my experience (subjective, I agree) is similar. And that is the kind of improvement I would expect if the noise floor had been lowered.

22 hours ago, Donuk said:

But surely whether any of these modifications to the audio file listed on this thread are conducted by the processor, memory and other junk in the NAS.

Not so. The file on the NAS is unchanged. What matters is how that file is rendered in the player. It seems that, though the digital audio signal in one lossless format is identical to the same signal in another format, players can render one format with a bit less noise (or whatever) than another. So that difference is a function of how players work, rather than of the format per se; it is perfectly possible to imagine a player in the future that did not have this rendering difference. If that is correct, timster's question as to whether and how Exakt and/or Katalyst make a difference is a good one.

My experience is that the difference (small as it is) remains in Exakt Katalyst systems. My suggestion (i.e. guess) is that the difference arises upstream of the Exakt engine or DAC, and that the extra processing required to convert a FLAC format file to a stream does generate a (very) little more noise. Likewise, padding out the 16-bit audio data with zeros to make it 24 bit (though the resolution of the data itself does not change) reduces quantisation noise slightly. The combination of those two effects is the difference that I hear. But, fair warning, that's all guesswork.

David

Edited by DavidHB

Share this post


Link to post
Share on other sites

During my Squeezebox period there was some talks in the forum about having either the endpoint or the server unpack the FLAC would make a difference. Maybe due to more network traffic or to more processing power used where the file is unpacked. After some  blind tests I could not hear any differences at all. Maybe there is a difference this time but to be certain I would suggest blind testing. I would not be surprised if results differs since people are using different storage, different players, different networks etc etc. So maybe the differences that might occurs could be traced to something else than just the difference between the formats. And even if it is found to be just the format, maybe it will not remain in the next firmware but might be back with a later one again.

Personally I prefer FLAC since it has standard tagging which WAV does not.

/Harry

Share this post


Link to post
Share on other sites
6 hours ago, Harryup said:

Personally I prefer FLAC since it has standard tagging which WAV does not.

Perfectly true. All my music is flac (other than the odd mp3/m4a), but transcoding to WAV24 (and I think it's the 24-bit that really makes the difference) is beneficial it seems.

  • Like 1

Share this post


Link to post
Share on other sites
7 hours ago, Harryup said:

Personally I prefer FLAC since it has standard tagging which WAV does not.

Me too. And it's Vorbis tagging, which is the most flexible and future-proof form. Given that there is no difference, in terms of the digital audio signal, between a 'native' WAV file and a FLAC file transcoded to WAV, there just isn't much of a case for storing one's library in WAV format.

David

Share this post


Link to post
Share on other sites
On 21/10/2019 at 18:34, DavidHB said:

Me too. And it's Vorbis tagging, which is the most flexible and future-proof form. Given that there is no difference, in terms of the digital audio signal, between a 'native' WAV file and a FLAC file transcoded to WAV, there just isn't much of a case for storing one's library in WAV format.

David

Well... I have reformatted all my music to WAV - from sources such as AIFF and FLAC - because it DOES make a difference... at least it does to my ears. And my wife’s ears.An my friend’s ears who I had listen to the difference between various formats...

This was a tip from Ian at Billy Vee in London. We had a listening session and we - my wife and I - not only could tell the difference but als liked WAV better than other formats. And we still do.

I’m not sure how to explain the difference. Everything just sounds more relaxed, more natural. Ian’s explanation was that WAV doesn’t need to be processed as much as AIFF and FLAC. This has the effect of the DS having to work significantly less and this results in better performance and better sound. I’m sure Ian explains it far better than I do,  but this is the core of his message.

Willem

Edited by DutchWillem

Share this post


Link to post
Share on other sites
33 minutes ago, DutchWillem said:

Well... I have reformatted all my music to WAV - from sources such as AIFF and FLAC - because it DOES make a difference... at least it does to my ears. And my wife’s ears.An my friend’s ears who I had listen to the difference between various formats...

I agree that there is a difference between FLAC and WAV, as rendered by our players, and have speculated on the reasons for this difference in previous posts. But I contend that there is no audible difference between a FLAC file transcoded to WAV (and seen by the player and control point and reported as WAV) and the corresponding file stored in WAV format. That was my point.

Naim always used to say that the difference between losslessly compressed formats (including FLAC) and native WAV arose because WAV could be rendered with a lower noise floor than the other formats. In the absence of other information, I think that this is a reasonable assumption to make.

David

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.