Volume om some cd’s

I think you can do it in dBPoweramp, at least if the output is .flac (I guess some others, too. And you can convert flac to flac). Use the batch convert option to do multiple albums if you wish. Select add DSP effects, and you can set replaygain (may need to read up what it does first!). Other options include dynamic range compression (though I guess that in this instance the problem isn’t too wide a dynamic range) and bit depth.

You will get a slight drop in sound quality due to the reprocessing and a layer of dither being applied.

It's a problem with the way that the CDs are being mastered.  The 'reference transfer level' (i.e. average signal level) for CD is specified as -18dB.  However, the marketeers and sound production engineers found that by using severe compression to squash the peaks, and then setting the peak level just at 0dB, they could increase the average transfer level by as much as 10dB; and this type of artificial loudness appealed to the non-audiophile mass market who are using 'bass boom' type ear-buds from their iPhones.  Hence more sales, hence more profit (but at the expense of audio quality).

Claus posted:

 

Is there any way to reduce the level of a ripped cd, or during the ripping, without a drop in SQ ? I guess I could even live with a slight degradation of SQ if the output level was ‘normalized’. 

How do you play back the ripped files?

I wouldn't normalise during ripping - IMO you want a master copy which is "bit perfect" from the CD - but either create a normalised version, or depending how you are playing back* use replay gain or similar to normalise when playing back from playlists.

* for example if you are using Asset UPnP its possible to configure it to use replay gain while converting to WAV.

Thanks for your replies. 

I play ripped cd's from NAS / ND5XS / NDAC with Minimserver. Right now into Nait 2 (unserviced, still sounds way above it's size!), and hopefully soon via 52/Scap/250. 

Eloise has an important point which I didn't consider in the first place. So ideally I should rip to WAV/FLAC, make a copy with reduced level to put in my music folders, and keep the original WAV/FLAC file in a folder that Minimstreamer doesn't use. Should I ever want to use the original file, it's still accessible. 

Depending on the difference in SQ between original and "harmonized" file, I might decide not to keep the original files of all the music (sometimes it's really not that important considering the content of the cd ). 

I understand that dbpoweramp can reduce the level with only a slight drop in SQ. I suppose it also can be done after the ripping. At the moment I use EAC for copying (it's free and it works). Do you have any recommendations for free software that can do the job as well, preferably with as little loss of SQ as possible ? Must be for windows, btw. 

Claus

It is not dBPoweramp itself that is causing the change in sound quality, but the fact that the software is making a change to the file. So dBP is fine for the job.

The queation is, which is worse: the annoying high level having to be turned down, or any audible change in SQ - and it might not be audible. Easily assessed for yourself.

I change my own made up NAS stored files of various WAV tracks (playlists) to be of the same volume using the dBpoweramp DSP feature,  these are normally used for background/dinner/party music,   & re SQ although in these albums its not serious listening,  I haven't noticed anything different.     I don't bother with DSP changes trying to match volume between whole albums,  the app volume control does that.   

This is something that I raised a couple of years ago and got heavily criticised for.

Most library management software - I use MediaMonkey - has the ability of analysing average volume level and applying a tag to each file with a gain setting (replay gain as mentioned above). Then dependent on the software, the volume is raised or lowered at playback. Naim streamers do not have the ability to do this. But with digital volume controls it is possible. MediaMonkey does have the ability - 95% sure because I haven't tried it - of taking a playlist and providing a levelled copy of the files.

What I would like to do is save these files as FLAC (2 or 3) and for each album in a tag set a percentage or a dB (have no idea whatr measure is used) how much I want the volume reduced. From experimenting I'll probably soon find a suitable common level for most of the cd's involved, and then adjust individually if the standard value is to much or too little. 

Then, I could use Minimserver to transcode to WAV during streaming and at the same time reduce volume according the the tag info. 

Is it as simple as that ?

Oh yes, with most of the albums concerned a minor and hardly noticable drop in SQ will be perfectly acceptable, some of them I have even considered storing as MP3/320k files to save some space, finding that with these 320k would be more or less ok. Since disc space is no issue for a foreseeable future I have so far stretched to not saving as WAV but as FLAC 3 or 4. 

Claus

Applying digital volume control to the underlying file doesn't seem a good idea to me. I'd want to keep them bit-perfect. Use the sound level equalising function in your playback software. Good software should use upsampling or some other method to reduce the impact on SQ. Or just use the volume control.

Claus posted:

What I would like to do is save these files as FLAC (2 or 3) and for each album in a tag set a percentage or a dB (have no idea whatr measure is used) how much I want the volume reduced. From experimenting I'll probably soon find a suitable common level for most of the cd's involved, and then adjust individually if the standard value is to much or too little. 

Software already exists to do this automatically... the usual system is called ReplayGain though there is also a new standard EBU R128.  You may find that if you ripped using dbPowerAmp (well regardless of how you ripped) your files already have ReplayGain tags... all you need now is t9 find some way to utilise those tags.

Then, I could use Minimserver to transcode to WAV during streaming and at the same time reduce volume according the the tag info. 

As far as I’m aware (and can find searching) MimiServer / MinimStreamer doesn’t utilise ReplayGain or similar tags.  AssetUPnP server does (if you tell it to).

Is it as simple as that ?

with the right server software and ripper yes.

Bananahead posted:

Naim streamers do not have the ability to do this. But with digital volume controls it is possible. MediaMonkey does have the ability - 95% sure because I haven't tried it - of taking a playlist and providing a levelled copy of the files.

Naim Streamers (even assuming you mean preamp streamers and Uniti devices) don’t use digital volume controls.  The preamps and Uniti devices do however use a digitally controlled analogue volume (which is very different though could probably still be utilised to implement ReplayGainor similar).

Hi Eloise, EBUR128 is a compression standard to control the dynamic range within media... kind of like the like the loudness wars but to a calibrated level... it’s not specifically about overall loudness or replay gain, as it’s results are about the ratio between the softest and loudest parts of the media. Such techniques are used in DAB,TV and FM radio to help with intelligibility assuming that most are not listening on Hi-Fi equipment, but car radios, personal stereos, phones etc.

Simon-in-Suffolk posted:

Hi Eloise, EBUR128 is a compression standard to control the dynamic range within media... kind of like the like the loudness wars but to a calibrated level... 

Thanks Simon ... I was unaware what it was but found mention in reference to ReplayGain and the suggestion was it was an alternative to that, but sounds like it’s not really.

winkyincanada posted:

Applying digital volume control to the underlying file doesn't seem a good idea to me. I'd want to keep them bit-perfect. Use the sound level equalising function in your playback software. Good software should use upsampling or some other method to reduce the impact on SQ. Or just use the volume control.

Absolutely correct.. unless very careful resampling occurs, which in itself is lossy and may add digital distortion, then aliasing and other sample value errors will be introduced... all in all the chances are you will destroy the bit perfect nature of the media and introduce non harmonic distortion which the brain will have to work harder to decode... it will sound ultimately less natural and easy flowing on a high quality replay system... clearly lesser systems are unlikely to be resolving enough.

Asset has the option to up the bitdepth from 16 bit to 24 bit whilst applying replaygain, this should in most instances remove the SQ impact.

Well, I finally had some time to experiment with this.

Using Foobar I have converted a few cd's from WAV to FLAC (2) and at the same time added Replaygain minus 6 to 9 dB.  Minimserver is converting to WAV on the fly, so ND5XS sees the files as Wav.

This surely does the trick and reduces volume on the albums which is absolutely nice. 

In the proces Foobar gives a warning that when adding replaygain the music files are altered and can't be converted back to original. 
I suppose this is the same as saying that the sound quality will permanently be reduced ?

I have decided not to erase the Wav files, but to move them to a folder on the NAS outside my music library. That way I can easily retrieve them, should I wish so. 

Questions:
How much degradation in SQ should I expect with replaygain set to say -6 or -9 dB? 
Is it possible to do the adjustment in a way that doesn't include any audible degradation, with the use of Exact Audio Copy, Win Media Player, Foobar and Minimserver/streamer ?
I don't have Dbpoweramp or Asset and would prefer to continue using the present software, which otherwise works fine for me.

Claus

-6dB is halving the signal, so you are going from 16 bit to 15 bit, -9dB is  just over 14 bit.

Note that there is a difference between applying a replay gain - which permanently changes the data.

and tagging a file with a replay gain - which doesn't change the data.

The first is a lossy operation - the information lost can never be regained.

In the second case, the media server can be told to first upscale the data to 24bit on the fly and then apply the replay gain; then there's almost no loss of data as the wider data word has more resolution so that when the data values are reduced the least significant data can still be accommodated by the higher resolution.

DBPoweramp's converter has both options, but I don't know about FooBar.

I don’t think it is possible to quantify any SQ degradation - or even to be sure that your ears (or anyone else’s) will necessarily detect it. Best just compare for yourself on a selection of types of music, comparing playing the original version with reduced volume control setting against the modified version at normal volume control setting. 

It may be that different software works differently in changing levels, so one may have less effect on sound than another - but unless anyone else has directly compared, that is likely to be down to you to do that.

Although I suggested replaygain, another alternative comes to mind: use DSP effects in dBPoweramp  to convert from 16 to 24 bits and using volume normalise to decrease the volume. The increase in bit depth might serve to reduce SQ effects arising from the volume change, but this is not something I know enough about to be sure.

Incidentally, dBPoweramp is offerered on a triaL basis for free.

Edit : I got distracted before finishing my post, and in meantime Huge has posted a related response. N.B. Replaygain on the fly requires a compatible renderer.

Thanks, I wasn't aware that adding Replaygain would reduce the bit number, or in fact that the bit number was related to the output level. I thought the bit number had to do with resolution only. 
Anyway, what's important for me to understand is that adding replaygain I permantly reduce the quality as well. 

As an example, If I reduce with 6 dB, coming from FLAC, would the result be on par with a 320k MP3 or better? And if better, then closer to FLAC or to 320k?

I can tag the files with replay gain in Foobar as well, it looks so at least. But it does't seem to have effect on the replay level. The info seems to be lost somewhere between 1) Foobar adding the tag, 2) Minimserver converting flac to wav and sending file to Naim, 3) Naim receiving and processing/playing the file. 

If the Naim player can read and respond to the tag, then it's "just" a matter of having it set in the file. If not then I guess it must be part of the transcoding done by the server software...

Huge, I didn't quite catch the upscaling to 24bit and then adding replay gain after that. Does Dbpoweramp do it as a file conversion, or is it done during streaming, like when I convert FLAC to WAV on the fly but now as a 2-step conversion, and if so then by with streamer software ? 

Thanks for you patience, I'm still on the beginning of the leaning curve when it comes to streaming music

I.B., just saw your reply as well. 

I take it that your comment re replay gain on the fly and a compatible renderer means that I can't do that with a ND5.

I might try dbpoweramp out to see if it gives me better options on this than what I have now. 

And surely, if I myself don't perceive a drop in SQ, or find it too small to be relevant, then it's really just a matter of adding to replay gain to the files and be happy with that. And of course keep a not reduced FLAC copy should I ever want to listen to the tracks in their original quality.

Thanks all,

Claus

Claus posted:

I.B., just saw your reply as well. 

I take it that your comment re replay gain on the fly and a compatible renderer means that I can't do that with a ND5.

I might try dbpoweramp out to see if it gives me better options on this than what I have now. 

And surely, if I myself don't perceive a drop in SQ, or find it too small to be relevant, then it's really just a matter of adding to replay gain to the files and be happy with that. And of course keep a not reduced FLAC copy should I ever want to listen to the tracks in their original quality.

Thanks all,

Claus

I don’t know if NDS is replaygain compatible, or if it , tpo, would just ignore - best check with Naim or search the detailed specs, or ask as a separate thread on here.

And the answer to your “and surely “ question is that is precisely it!  Whether that would remain the case if you change or upgrade other parts of your system or listening room is of course uncertain - but if you keep the original flacs somewhere you would be able to revert to them if they then sound better.

I would have thought that it would be the UPnP server software which reads the replaygain metadata tags and serves the attenuated audio file data to the Naim renderer which would play the file as it would any other. As Mr Spoon indicated, Asset has a specific config. settling to accomplish this. You certainly wouldn’t want the renderer duplicating this process.  

The ND5 XS does not process the <Replay Gain> tag.

At one time - after I'd optimised my network for the my ND5, I tried processing some files using Audacity, and I could hear almost any reprocessing as a loss of quality, the effect varied from subtle to very noticeable, but i was always there to some degree.  I've not tried FooBar, but i'd be surprised if it was that much better.  I suspect that if I'd done testing this before optimising everything, I may not have been able to hear the more subtle changes.

The loss of resolution is relative to the full scale...

The reference transfer level for CD is -18dB with respect to the 16 bit full scale, this is to allow for extreme transients.  However most music doesn't have such dramatic transients (particularly more recently mastered material that's heavily compressed for the younger "MP3 generation").  So to make it appear louder (and more dramatic without them turning the volume control up) the data are encoded using a higher level.  This is just a matter of numbers.

To illustrate:
If for instance you were to take a system that could use values from -128 to +127, with no signal you'd have a steady value of 0 (i.e. the mid point).
One piece of music could have an average of +16 to +16, but an extreme transient may reach +127.
Another piece of music could have an average of +16 to +16, but the most extreme transient may only get to +63.
The recording engineer could record this second piece using a +6dB gain, and the second piece of music will then have an average level of +32 to +32, but the transient will still be +126 and so still be in range; however the average level would will be 6dB greater than the first piece.
In both cases the smallest sound* that can be encoded is going between -1 and 0.  So if you set a -6dB replay gain for the second file, the -1 to 0 becomes -0.5 to 0, and this will be rounded down to 0 to 0; i.e. the signal disappears*!

If you increase the mathematical resolution (i.e. convert 16 bit to 24 bit) before doing the replay gain, this problem goes away.

This type of loss of quality can't be compared to the difference between WAVE and MP3, as the way in which the quality is lost is completely different.

Upscaling to 24bit (to avoid loss of quality can be done by Minimserver while it's doing the FLAC to LPCM (i.e. WAVE) conversion, I believe it can also apply the Replay Gain.

It is also possible to apply upscaling to the FLACfiles before you apply the replay gain.  This will be better than leaving them as 16bit and losing the resolution.

* The use of dither actually makes it possible to record smaller sounds in some circumstances.

Guys, I have looked into this to some extent... I don’t believe the loudness of CDs is often down to the overall gain, but more to do with compression used in the master...  a subtly compressed master can sound quite a bit louder than a non compressed master...