Playback volume

[Colin hastie]

Dave C (30.04.21) says I'm sure everyone will have experienced different playback volumes. An explanation as to why this happens doesn't remedy the situation. Daniel Taylor who makes considered and helpful comments via the forum recognises that this is a problem suggesting using Audacity as a solution but admits that it "is probably not a good solution for most or even for me".

I would like to add further comments to my earlier post, 17.04.21, about playback volume levels.

1. When I first started looking at getting a B2 about five years ago the fact that I did not have to set a recording level when ripping a CD as you have to do with a B7 was one of the features that attracted me into buying a B2. This was mentioned by Martin on the Brennan website and I seem to remember a comment to say that there would be some levelling of the playback volume. All you had to do was rip the CD and everything else would be taken care of. Interestingly this does not now appear to be on the website unless I have missed it.

2. Setting a recording level on a B7 is a bit hit and miss. What I used to do was to play a CD and then try to work out a corresponding recording level. I then made a note of the recording level on spreadsheet so that if I had guessed wrong I had a better idea of how to adjust the recording level. This actually sounds a quicker process than using Audacity and both extend the time needed to rip a number of CDs.

3. On comparing the recording levels I had used for my B7 and comparing them with the playback volumes on my B2, most CDs seem to be a quite a broad band in the middle with a number of CDs at either the loud or quiet end of the playback volume spectrum, indicating that the issue that I had assumed the B2 would address has not happened.

4. If you have not noticed this yourself, try turning the volume level up or down 10 levels to get some indication of what it is like when listening to a quiet or loud CD when listening in the random mode which is the problem I, and probably many others, have. Trying doing this when you have company and see what the reaction is.

5. I do not have any MP3 files or downloads and I only use my B2 for listening to my ripped CDs. The random function means I can hear the music differently as the tracks are not in the same sequence that I am used to when listening to the source CDs. This is another feature that attracted me into getting a Brennan in the first place, a juke box of my CD collection. 

6. I think Dave C is right, everyone must be noticing this, so a solution needs to be found, and not one that involves buying more equipment or taking additional time to do what should be a basic function of the B2. Using Audacity seems to be the way forward for ripping vinyl to a B2 but shouldn't be the solution with regard to CD playback volumes.

7. The solution seems to me to be for there to be a new feature in a future upgrade that will level the playback volumes leading to happy B2 owners.

I would appreciate comments from Paul and Martin who I have found very helpful in the past when I have had a problem.

[Daniel Taylor]

I agree that this is a problem, especially when using Random while you have guests.

As I understand it, automatic volume leveling is not a trivial matter.  If it's done in the digital domain, without great care, the audio quality can suffer.  The more the volume is reduced the greater the degradation.  Doing it in the analog domain requires additional circuitry - which I don't anticipate Brennan being interested in adding.

What we have is really the same problem when you play one CD after another on a CD player.  The relative volume will change depending on the level recorded on the disc.  It's just more obvious on the B2 because the next album starts right up, possibly when you're not expecting it.

A short term work-around that I can suggest would be to make a playlist that consists of only albums with similar volume levels.  That would allow playback for guests without any abrupt surprises.

Personally, I just keep the remote control close at hand when playing the B2.

[Rearwing]

I will be very unhappy if playback levels are artificially altered, the levels are dependent on the mastering process and as such should be respected. 

I purchase multiple copies of recordings to encompass the different remastered versions because I enjoy most, but not all of them. A similar difference occurred with different vinyl pressings, with many people hunting out the special versions.

Not all of us want to listen to elevator music, but rather the closest to the original recording/remastering. 

[Mark Fishman]

Let's examine several aspects of the problem:

First of all, a digital recording has a maximum possible amplitude, that is when all the bits are used. Attempts to record higher peaks just clips them, which sounds terrible. But the "apparent" volume" or loudness of music is not due to the maximum peak level achieved, but to the AVERAGE level. In classical music, in acoustic jazz or folk music, this can be 20, 30, or more dB below the maximum peak. In rock or pop music, especially music that has been mastered or released since about 1995, the average to peak ratio can be as little as 3dB, because those releases were trying to be the LOUDEST on the radio or the stream.

As an aside, I have both the original and millenium editions of Michael Jackson's "Bad" -- the millenium edition was remastered to be LOUDER so if you set the average levels of the two CDs the same the new remastering sounds much less "punchy" and much more distorted.

Since making everything LOUDER will lead to lots and lots of audible clipping on the tracks that have anything like the original musical dynamics, to match average levels means making the LOUD tracks quieter.

Replay Gain is an automated attempt to address the problem. It can be applied on an alvbum-wide or track-by-track basis. If you apply it track-by-track, they'll all have the same average level when played randomly, which is what you seem to want -- but if you ever play the whole album they came from, thewy won't have the varying levels the original artist might have wanted. The sad ballad will be as LOUD as the hard rock anthem in the next track.

If you apply it on an album basis, each album will have a standardized average level, but the individual tracks might vary greatly when played randomly.

Furthermore, while replaygain adjustments can be made reversibly for mp3 files, adjusting the gain of wav or flac files means mathematically changing every single sample, and with roundoff errors that process can not be undone.

Think very hard about whether you really want to do this, and what exactly you need.

[Daniel Taylor]

The remastered version of 'Bad' is somewhat famous.  They actually ended up in court.  I think it was the co-producer, Quincy Jones, who sued the record company for screwing up the dynamic range of his production.  Well known audiophile Michael Fremer was called as an expert witness.

[Mark Fishman]

There's a really good discussion of perceived loudness and how streaming services -- specifically Spotify -- achieve some of hwat we've discussed here:

https://www.izotope.com/en/learn/how-will-my-music-sound-on-spotify.html

At the very least, it's interesting, and might offer some ideas about dealing with playlists for folks who care.

[Peter Lowham]

Hi Mark,

Thank you for the link to this article; it is very interesting and informative.  And it does emphasize the problems (and dangers) of taking what seems to be a simple issue and explains how complicated it can really be to provide some sort of solution.

Regards,

Peter.

[Colin hastie]

Thank you to those of you have replied. From reading the comments there doesn't seem to be an easy solution to the problem without some impact on the audio quality. As I see it, the only way around this would be for the B2 to have a recording volume level setting in the same way the B7 does but with a wider range of recording settings. I found that 15 recording volume settings on the B7 was not enough to get more of an equitable playback volume between the quietest and loudest CDs. At least 20 volume settings would hopefully work.

I had hoped that someone from Brennan would have replied to my post as differing playback volumes is something that all B2 users will experience. So Martin and Paul, would a recording volume setting be feasible to be added as part of a software update to address this problem?

[Rearwing]

Im sorry to be a contradictory, but this is not a problem that needs to be mended; differing playback volume is a result of different mastering being applied to different recordings, it is how it should be. Daniels suggestion of using audacity will solve your problem of wanting to have a consistent volume level, but please understand this is not something that everyone finds desirable.

The potential for degradation of the sound quality is too great within the B2, so it might be worth you taking the time to use audacity so that it solves your problem. My problem is spending money on the same pieces of music everytime a re-mastered edition is released!

Vive la difference!

[Jeff. M.]

It's what it is Rearwing You are right, back in the day when you played a cd on your player, you were quite often up a few moments later to turn it up or down, same with the next one. It doesn't need changing and is not a B2 problem

[Mark Fishman]

The easy way out would be for you to apply gain adjustment to your own CDs, if you generally listen to them in randomized fashion, or to make special playlists with gain-adjusted versions of the tracks if you simply need it for parties and the like. I have done this for a bunch of Christmas music that I only play when no one is paying much attention to it anyway, to use as background for Christmas parties, so the music stays at a reasonably consistent level.

I assume you have a backup (export) of your B2 contents? Connect it to a computer capable of running any of a number of programs that can analyze the tracks and apply gain adjustments to them. You seem to want to have every track at about the same average level, so "track" adjustments (rather than "album" adjustments) would be appropriate. Then delete the music from your B2 and import the adjusted backup.

Or change the names so you can have the adjusted copies alongside the originals, in case you ever want to go back -- for FLAC, ALAC, and WAV files the adjustment is NOT reversible as it involves recalculating the value of every sample after analysing the whole track.

In other words, you can get what you want without anyone else having to change what they have.

[Fred Waltman]

What I was thinking about was some way to a +/- factor to an album. Say your volume is set to 28 when you start a playlist. If you played a track from an album with a +2 factor it would play it at 30 (then reset to 28). Maybe the next track is really loud, so you had a -2 factor assigned to it, then the volume would drop to 26 for that track.

Maybe it should be a percentage, +10 means up the volume 10%, -10 reduce it by 10%

The idea is you wouldn't have to set it to every album, just those that deviate from some normal. And you are not modifying the audio data at all so it should place a load on the B2. The volume is already controllable thru software (Web UI) so there shouldn't be major changes there.

If you were to get really fancy, you could have a "learning mode" where it remembers that you turn the volume up/down on this track/album and set that as the factor.

[fred.w....@gmail.com]

There are all sorts of things one could do with/on that little Raspberry Pi inside the Brennan units, but there is only so much computing power available and even as things stand we know that doing two things or more at a time can lead to problems, so I would be rather concerned it yet more CPU cycles had to become dedicated to "learning" user preferences. Also the more code the more chances for bugs and the longer the testing cycles need to be.

Fred

[PMB]

Hi All,

It got me thinking along the same lines as Fwal i.e. a manually set volume offset but this would mean a lot of manual adjusting to set up. 

Paul

Brennan Support.

[Fred Waltman]

Paul,

I don't think there will be all that many for most people. I've got 2000 albums loaded and there are probably less than 20 that I would adjust this way. If it bugs me enough I will eventually run the tracks thru some software to adjust them. I'll set up two playlists "TooLoud" and "TooSoft" to remind me which ones need adjusting...

Newbie Fred

[PMB]

Hi New Fred,

I have added it to the Wishlist.

Paul

Brennan Support.

[JFBUK]

Hi Fred & Paul,

FWIW a lot of software that attempts to manipulate playback volume store a target volume in a metadata tag rather than permanently alter the source track audio data.

Take a look at 

https://xiph.org/flac/documentation_tools_metaflac.html

The process needs an analysis pass first to calculate and store values.

I know that tags in general are already on the wishlist for the B2 and this is another example of how they open up functionality.

John

[Mark Fishman]

As you can see from the page you linked, even if the player you use would honor the replaygain tag, using metaflac you would have to process the files that need volume adjustment one at a time, in order to make them similar in volume. If you process them as a bunch, they would be treated as if they were part of one "album", and the differences in relative volume would be preserved (for what should be obvious artistic reasons), which is not what random playback seems to call for.

It could be scripted, of course, but it's not as simple as throwing a directory at a tool. There are other tools that can be told explicitly to do per-track adjustments on bunches of files, e.g., foobar2000.

A good discussion is at https://wiki.hydrogenaud.io/index.php?title=Replaygain

[Colin hastie]

Thank you to everyone who has replied with possible solutions which I am not sure that I totally understand but will have a look at.

No one has commented on the idea of having a recording level setting when ripping a CD, a feature of the B7 which I acknowledge is time consuming but it looks as if all the other alternatives are time consuming as well. Setting a recording level was what we all used to do when making mix tape cassettes as well as something owners owners of a B7 do.

I only rip CDs to my B2, no downloads or MP3 files and I have estimated that about 20%, probably more, of my CDs either play back as either too quiet or too loud which is possibly why I am using my CD player more to listen to music than my B2.

[Mark Fishman]

Colin, your post is not the first time I've seen mention of "setting a recording level" on a JB7. I do not have a JB7, so I hope someone who has one will comment, but I did download the JB7 user manual and looked through it. The only explanation of setting a recording level I could find applies ONLY to recording through the AUX input, i.e., recording analog audio from an external source such as a tape player or turntable. I did not see anything about setting levels while ripping CDs. 

Setting recording levels when recording analog signals, whether the destination is an analog medium (e.g., a tape cassette) or a digital one (e.g., a digitized file on a hard disk or a standalone audio-CD recorder), is very different from ripping or copying a digital bitstream and altering the intended (relative) playback level. You can do the same kind of analog-to-AUX-input recording on a B2 as was done on the JB7, and I think you can set recording levels for that process on a B2 also.

If you are ripping a CD, though, you aren't "playing" the CD, and there's no volume control or analog signal step involved in the process. You are copying the bits that are on the CD, which have been set to a relative level by the engineer who mastered the CD, usually under the musical supervision of the producer or artist. (I say "relative level" because obviously you can turn the volume control up or down when listening, but the level of the music is still the same relative to the maximum possible value of the digital samples. Maximum would be all 1s: attempting to increase the signal level further just clips off the signal instead of making it louder.) To alter the relative playback level of digital audio, you either have to recompute the value of all the digital samples (making them higher or lower values), or you have to store instructions inside the file that tell the player software to do that at playback time.

Apart from the fact that not all software honors such instructions, either method is not as simple as altering the gain of an analog stage, and depending on how it is done it can permanently alter the file, possibly with audible distortion.

Basically that is the reason that almost all radio stations still use signal processors in the analog signal path to both compress and limit dynamics. It is the easiest and most flexible method to force every signal source to sound approximately the same loudness. I have an old dbx118 compressor/limiter that was sold as a consumer product to do a similar job, and if I ever decide I want background music I'll connect it between the B2 and my amp.

Cheers -- m.

[Mark Fishman]

I've been thinking about this problem and how to address it without adding too much of a burden to either Martin Brennan's software development, or to the B2's CPU, and I think the answer lies in making sure that when the B2 plays a file it honors any ReplayGain metadata that is stored in it.

I'm not suggesting that the B2 do any analysis of the file to determine its loudness or dynamics -- that's a pretty time consuming and computation intensive process, and there's already lots of Windows, Mac, and Linux software that can do that. It's a once-per-file computation, anyway, so if it has to be done outside the B2 (either by NAS or a copy on an external disk) it's relatively painless for the people who want it. Those of us who don't want it don't have to worry about it burdening the B2.

Once ReplayGain tags are added to file(s), though, what is necessary is that the B2 honor -- instead of ignoring -- the gain tag, during playback. There are software players that do (e.g., foobar2000) and software players that don't. A player that honors the tag adjusts the playback volume according to the tag, as it encounters each file.

If the B2 would honor the tag, it would be helpful to be able to switch that behavior on or off, as appropriate to whether one is playing a randomized playlist or playing an album straight through, for example.

I repeat that it would not be necessary for the B2 to analyze the files, or add the tags into the files. But if they are there, and the feature is "on", it should read the tag and adjust the existing software-controlled output level accordingly.

I think this might be straightforward in use and in implementation. ReplayGain metadata is a standard that is well documented on the web, and if one is only implementing the playback part of the spec I hope it wouldn't be too hard to accomplish. I'm not a good enough programmer to say that it's "simple", but it has to be easier than implementing the analysis and tagging as well. No added hardware would be needed, certainly.

Paul B., would you bring this to Martin Brennan's attention, please?

[Colin hastie]

Thanks for your prompt and clear detailed response yesterday Mark. I had posted a response earlier today but it didn't send. Your post has clarified the issue for me and I now have a greater understanding of the issues involved.

I had a look at my B7 User Guide and it does say that the recording level setting in the Recording menu is for when using an AUX input. I obviously hadn't read it carefully enough. It's a shame no one from Brennan had been able to clarify this as I am sure such clarification would have been helpful to the forum users as this has featured in some posts.

I looked out an old email I had received from Martin Brennan in September 2015 in which he was announcing the new B2. In the attached link I have a vague memory that Martin was saying that he had wanted to have some form of recording level / playback level or something similar but he was persuaded otherwise by the engineers. This had been a feature that had attracted me into upgraded to a B2. Unfortunately the link is no longer available so I haven't been able to clarify this so I may be wrong.

Your final comment in your post today is very helpful, and hopefully a solution can be found.

Cheers, Colin

[J Rathbone]

A quick Google of ReplayGain provides many resources on the adoption of this de facto technical “standard” across media players / software etc eg

https://wiki.hydrogenaud.io/index.php/ReplayGain

[Colin hastie]

Hi Paul,

In a post on this thread on 28.06.21 Mark Fishman asked you if you could bring ReplayGain to Martin's attention as an option with regard to playback volume.

Have you been able to get a response back yet from Martin?

Cheers, Colin

[PMB]

Hi Colin,

Sorry no response from Martin yet but it is on the list.

Paul

Brennan Support.

[Colin hastie]

Hi Paul,

Thanks for the update.

Cheers ,

Colin