Vinyl Studio - Recording Levels

[mkn777er]

A quick question please from a 'novice'.
I'm having great fun using Vinyl Studio to digitise my vinyl collection. I have a Sony USB deck and I record as a FLAC file with a sample rate of 96 KHz and 24 bits. After using Vinyl Studio for track listings, clean up etc I then copy the file as a FLAC to my Astell and Kern KANN device. It works well. But I have a question about volume levels. During the 'Cleanup Audio' process it allows me to select 'normalise', with the loudest being 0db. I still find this too quiet. Recently I did a recording and manually selected the Preamp to +10.0db in both L and R channels. The peak meter was bouncing in and out of the red. But it sounded great to me on playback through headphones. What surprised me though, was that another recording on my KANN of an mp3 file (CD) was even louder, and I had to turn the volume down. So how can I replicate the volume of CD recordings using Vinyl Studio? At the moment, I have to turn up the volume for vinyl recordings, and then turn down the volume for CD recordings - which is a bit annoying when you try to make a playlist! Many thanks.

[Paul Sanders (AlpineSoft)]

Hi,

Probably not.  Many commercial CDs are compressed, making them sound louder than they really are.  VinylStudio doesn't do this, in order to retain the original sound.  It's something we have thought of adding to VS but have not yet got around to.  I'll add it to the request list.

[Lewis]

A quick question please from a 'novice'.

During the 'Cleanup Audio' process it allows me to select 'normalise', with the loudest being 0db. I still find this too quiet. Recently I did a recording and manually selected the Preamp to +10.0db in both L and R channels. The peak meter was bouncing in and out of the red. But it sounded great to me on playback through headphones. What surprised me though, was that another recording on my KANN of an mp3 file (CD) was even louder, and I had to turn the volume down.

Hi. Normally (in a standard way of digitization) you must not exceed or should not even near 0dB in the digital domain.
Look up 'clipping' using your fave search engine.
All you need for your headphones is a typical headphone amplifier.
Virtually, -3dB is quite OK for digitising your LPs. I prefer setting up my level @ -3.5dB. I follow this method not to produce digitizations which seem pretty sullen for my ears but only to have a headroom for normalization. Say, you record your files losslessly and export them into LAME (lossy).

[davelr]

Probably the only way around this problem is to apply R128 Volume Leveling during playback. This is a two-step process where the music files are first analyzed and volume level data (per track, per album) are stored in the file. During playback the software/hardware uses these data to adjust the playback level to roughly "normalize" the perceived volume between files. Since the actual music data is not modified, it's not necessary to "boost" the volume levels during capture as you appear to be doing currently.

I have no idea if your device supports this or not, but most likely it would only be the playback portion and not the analysis phase. I use this feature of JRiver Media Center to level the playback of mixed playlists and it works pretty well. If your device does support R128 playback you could use software like JRiver to analyze your music files and then transfer them to your device as the data are stored in the files themselves.

You can read their brief description of the process at jriver.com support/wiki/volume leveling

[Paul Sanders (AlpineSoft)]

Thanks for this.  Good to know.

I just wanted to chip in that although VinylStudio is not (yet) clever enough to do EBU R 128 levelling - it just looks at the peak level - it can normalise on a per-track basis.  The results are then reflected in the saved tracks and will play correctly on any player.  Also, I don't know what standards exist for recording levelling information in various file formats (it would have to be some kind of tag) so there is a potential issue there.

VinylStudio's mechanism is workable because of the non-destructive nature of the edits it performs.  As long as you keep you original recordings around (which we always recommend, precisely for reasons like this), you can always save your tracks again if you change your mind about something.

So, what we (AlpineSoft) need to do is to look into supporting a smarter levelling algorithm for people who care about perceived loudness.  We will put it on the list.

[Paul Sanders (AlpineSoft)]

Update: I did a little more background reading.  tl;dr; it's more complex than that, we'll see what we can do.

[Steve Crook]

I was intrigued by this (never heard of it before) and some random Googling got me to this site: https://freelcs.sourceforge.net/ which then got me to: https://github.com/jiixyj/libebur128 which is the library used by freelcs, and is on an MIT license. Might be a leg up, or something you could use directly?

HTH

Steve

[Paul Sanders (AlpineSoft)]

Yep, found that one already   Thank you though and I agree, it looks very promising.

[davelr]

I had meant to do this the other day but spaced it, sorry.
Paul had opined that the R128 info must be carried as tags in the files. The attached image shows the playback tags within one of my flac files. There are several tags annotated as R128. I do not know if the tag names are standardized or are specific to JRiver media center's analysis routine, but I would assume that there must be some standardization of tag names otherwise none of this would be particularly useful. Anyway thought this might be of interest.

[Paul Sanders (AlpineSoft)]

Thanks for this.  Yes it is useful.  It's certainly part of the story, but maybe not the whole story - EBU R 128 is quite complex and I don't understand it properly yet.

So I need to learn more about it before I can decide what we should actually do to make sensible use of the various methods that EBU R 128 supports.  In particular, some of them adjust the gain dynamically throughout the recording (and therefore cannot be represented as tags).

I'm not sure that we'll bother with this though - it's probably enough to do loudness-based normalisation of the entire recording to a single level.  This should, hopefully, make all recordings normalised in this way sound roughly the same (I think that's the aim).  I'm not sure if this is what JRiver do or not.  Maybe they'll tell me.  Or you

[balky]

I use an RME ADI-2 PRO FS Black Edition for digitising (some might find it an overkill and it's okay) but I do know that it has its solid advantages using it either as an ADC or a DAC.
From what I know, digitizing with this device is as pristine as it gets for a (crazy) home user...

The device itself supports up to +2db without clipping, so I simply take advantage of that and set recording level up to -0.3db.
No need to alter anything in the digital domain.

The meter on VS bounces to red most of the time with thousands of clip even though the audio level never gets past -0.3db.

During playback, I don't have to crank the main amp volume when playing back the digitized files.
I use a Marantz Network streamer for playback, and the playback volume is just nearly as loud as any commercial CD, but without the irritating compression artifacts you hear on most commercial CDs.