Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Vocalpoint]

Recently I have been testing FLAC conversion of several SACD iso files - and then comparing the resulting FLAC files with the same album from my own standard CD rip of the same tracks.

In every case -  the converted FLACs are always an order of magnitude quieter - and I do not understand why.

My workflow for FLAC conversion is as follows:

1. Load iso into my Playing Now (Player tab) window by dragging it in via Windows explorer
2. The tracks appear in the windows
3. Highlight all tracks and choose Convert Format
5. In the Convert Format dialog - I do not apply any DSP and the bitdepth is set to Automatic
6. I press Convert and end up with a series of FLAC files @ 352khz/24bit
7. I then highlight the files and apply Audio Analysis
8. Then do a compare with the same album (16/44 usually)

While the mastering of the album from SACD COULD be vastly different - in many cases the DR values I am seeing on the SACD conversions are in line with the same DR on a track that is 16/44 and already in the library. But the actual output is noticeably more quiet all the time. I have yet to play a set of converted FLAC files that are in line volume wise with a standard CD rip.

Am I hearing things or is there something else I am missing?

VP

[thecrow]

This is a well known consequence of SACD here is a good description of why it is so.

https://kb.roonlabs.com/DSD_to_PCM_Settings

Most DSD content is generated from a DXD master. DXD is PCM at 32bits/384kHz or 32bits/352.8kHz. This PCM signal is converted into a DSD signal via a process called Sigma-Delta Modulation.

One of the design tradeoffs in Sigma-Delta modulator design is the management of mathematical instability. Sigma-Delta modulators utilize feedback (or feed-forward) loops that can become unstable if the input signal is too loud.

One of the most common practical measures used to prevent instability is limiting the amplitude of the input signal. In fact, the Scarlett Book (SACD) specifications require that the input signal be limited to -6 dBFS (meaning, 6dB less dynamic range than the original signal).

This has a side effect: if you were to play a CD and SACD side by side on the same transport, assuming the transport does nothing to compensate for the -6dB adjustment that happened during authoring, the SACD would come out 6dB quieter.

[Vocalpoint]

One of the most common practical measures used to prevent instability is limiting the amplitude of the input signal. In fact, the Scarlett Book (SACD) specifications require that the input signal be limited to -6 dBFS (meaning, 6dB less dynamic range than the original signal).

That why I love this forum. That makes perfect sense to me.

So - in terms of MC - looks like a +6 boost is required - when converting to get the resultant FLAC files to level match with an existing set of 16/44 files.

Will try a few test conversions with some processing options to see if I can get around this.

Thanks for the insight.

Cheers!

VP

[Matt]

There is the option to boost by +6 dB when converting DSD files.

Options > Audio > Configure input plug-in > DSD input plug-in > Increase volume of DSD to PCM by +6dB

[Vocalpoint]

There is the option to boost by +6 dB when converting DSD files.

Options > Audio > Configure input plug-in > DSD input plug-in > Increase volume of DSD to PCM by +6dB

Perfect! I figured there would be something out there like that.

VP

[RD James]

I would strongly recommend that you do not enable the 6dB boost.
I checked my DSD library some time ago, and more than 80% of tracks would be pushed to clipping with a blind +6 dB boost.

You can add this as an expression column to check your own library if the tracks are analyzed:

Code: [Select]

If(Compare(Math(RemoveRight(ListItem([Peak Level (R128),0],0),5)+7), >=, 0), <font color="ff0000">WILL CLIP AT +6 dB<//font>,)
Even if it was reduced to +3dB, 15% of my library would still be affected.
Do a proper conversion (no 6dB boost), and enable Volume Leveling on playback.
I would also recommend disabling DSD bitstreaming and using Volume Leveling on playback with those tracks too. Then everything will play at the same level.

[Vocalpoint]

Do a proper conversion and enable Volume Leveling on playback.

I am already doing this - and this is leaving the files noticeably more quiet than they should be.

To be clear - I have no use for the actual ISO file (so no DSD bitstreaming etc) - I am converting all tracks to FLAC for use in MC.

VP

[RD James]

If you are comparing DSD -> PCM playback, to DSD -> PCM conversion, the playback level should be the same.
Are all tracks analyzed? Are you applying DSP on the conversion?

[Vocalpoint]

If you are comparing DSD -> PCM playback, to DSD -> PCM conversion, the playback level should be the same.Are all tracks analyzed? Are you applying DSP on the conversion?

Per my workflow at the top - I simply mount the iso in MC (Now Playing) - convert the listed files to FLAC and run audio analysis. No dsp. No plugins. No nothing

When I play the converted FLACS against the same copy of a 16/44 album I already have in the collection - the SACD converts are always markedly lower in volume - even after analysis.

The +6 plugin seems to be the missing piece here as these converts seem to be about that much quieter.

VP

[RD James]

When I play the converted FLACS against the same copy of a 16/44 album I already have in the collection - the SACD converts are always markedly lower in volume - even after analysis.

Oh I see, you're comparing two different versions of the same album, not DSD vs PCM or SACD playback vs FLAC conversion.

Due to the way that Volume Leveling works, make sure that both albums are named differently if you are trying to compare the playback volume. If they have the same name, one adjustment will be applied across both.
Since an average level for the album is used, make sure that you have the same number of tracks and are not including bonus tracks for one album but not another.
 
Alternatively, set the Media Sub Type to 'Podcast' temporarily, since that forces track-based leveling rather than album-based leveling.

Keep in mind that [Dynamic Range (DR)] is also a factor. The lower the dynamic range, the louder a track may sound.
Volume Leveling tries to keep the average loudness the same, but they're never going to sound identical.

[Vocalpoint]

Due to the way that Volume Leveling works, make sure that both albums are named differently if you are trying to compare the playback volume. If they have the same name, one adjustment will be applied across both.
Since an average level for the album is used, make sure that you have the same number of tracks and are not including bonus tracks for one album but not another.

I do not understand this bit. When I convert the iso to FLAC tracks - they are converted to my local drive and displayed in my "tag" view - which only shows files from a specific folder on my local drive. Then I run Analyze Audio on them and then right click on the album and choose "Add to Playing Now)

Then I go find another version of the SAME album (that is already in MC but out on our Music share) and add that version to Playing Now. Then I slap on the phones and start comparing track 1 from SACD to Track 1 from CD and so on.

I am not understanding what you mean by "If they have the same name, one adjustment will be applied across both". I have done this for a pile of converted tracks in the last week or so and all of them have their own specific audio analysis values. (and they playback with unique values as noted in the dialog that appears when I click the "blue" (bitperfect) button in the upper right corner of MC during playback.

The names are identical and so are track counts. I can confirm that Track one from a converted SACD album has completely different audio analysis characteristics than the same track contained with my main FLAC library (ripped from a CD like 2 years ago). And it shows in the playback - the SACD versions are always noticeably more quiet - even with the identical DR (in some cases)

Please elaborate on your statement a bit more.

VP

[RD James]

If the album tag is the same, and they are played next to each other in the playlist, Volume Leveling will treat them as though they are from the same album - so an average volume level is applied to those tracks.
If the album tag is different, each album will be adjusted to a separate level.
 
Changing the album name to be different, or temporarily setting the media sub type tag to 'Podcast' will force them to be leveled independently.


Here's an example:
I typically use the [Description] tag to separate different album versions, but that doesn't play nicely with Volume Leveling.
With [Album] being the same for both tracks, they are reduced by -11.4 dB (the average level) and the "Remastered" CD track sounds much louder.


Moving the description into the [Album] tag makes them unique, so Volume Leveling treats each track differently.
Now both tracks play at approximately the same loudness - though the SACD sounds better as it has much more dynamic range.

[Vocalpoint]

If the album tag is the same, and they are played next to each other in the playlist, Volume Leveling will treat them as though they are from the same album - so an average volume level is applied to those tracks.
If the album tag is different, each album will be adjusted to a separate level.

Wow. Awesome. I will test this tonight. In my case - it simply a case of ensuring the Album title is different for my converted files. Easy enough to do as I can put anything in there while the tracks are in an untagged (freshly converted) state.

VP

[RD James]

Wow. Awesome. I will test this tonight. In my case - it simply a case of ensuring the Album title is different for my converted files. Easy enough to do as I can put anything in there while the tracks are in an untagged (freshly converted) state.

VP

If you're only trying to determine which one is best and making these changes as a temporary measure, setting the Media Sub Type to 'Podcast' rather than 'Music' may work best, since that levels on a per-track basis rather than a per-album basis, which should ensure that leveling is working correctly for the comparison. Just remember to change it back when you're done.
 
If you are potentially going to be keeping both albums and mixing them in playlists, then I'd recommend giving them unique album names.