ian-shepherdIan Shepherd’s first feature on RecordProduction.com

The game has changed – forever.

It’s no longer about sounding great on CD, it’s no longer about sounding great on the radio.

It’s about sounding great on Spotify, and this is fantastic news for anyone who loves great-sounding music, because it means the so-called “Loudness Wars” are about to become an irrelevant footnote in history.

If you live outside Europe, you may not be familiar with Spotify yet, but trust me – you will be.


Spotify enables you to stream almost any music you like to your computer or mobile device, in return for simply listening to a few ads, or for a modest monthly subscription fee. (Required for the mobile version.)

The interface is great – slick, professional and easy-to-use. The music sounds good – better than any mp3 thanks to the use of the Ogg Vorbis codec – and is available in even higher quality for subscribers. The catalogue is huge, and growing every day, and mobile users can even listen offline.

But what does this have to do with the Loudness Wars ?

The answer is - Spotify uses “Volume Normalisation” by default.

It adjusts the playback level of all songs so you don’t have to keep adjusting your volume control.

Which means that a genuine pop classic like “Billy Jean” will play at the same volume as the flat, fuzzy distorted mess that is Cheryl Cole’s new single.

And that anything off Kasabian’s latest album will play at the same volume as anything by Black Grape.

Or that “In Bloom” from Nirvana’s masterpiece “Nevermind” will play back at a similar level to U2′s recent Loudness-War-casualty “Vertigo”.

Guess which ones sound better ? The modern, brickwalled, crushed-to-death clipping victims, or the lower-level, more dynamic, open, punchy, older stuff ?

You guessed it. To take that last example, Nirvana wins – by a mile. The kick kicks, the guitars bite, the whole thing rocks. Vertigo is a limp, mushy lump by comparison.

Don’t believe me ? Try it yourself ! Fire up Spotify and pop those two into a playlist. If you don’t have Spotify, just line them up in any audio editor, and turn the U2 version down by 5dB. Or compare any of those tracks, making sure you use the TT Dynamic range meter to level-match them first.

Listen to the kick drums, the snares, the guitars, the bass.

(And while you’re listening, pause for a moment to think – have you ever heard anyone complain that “Nevermind” was too quiet ?!)

QED.

Every track has it’s playback volume adjusted according to “ReplayGain” values which give a decent estimate of a song’s apparent volume and therefore how dynamic it is. Very dynamic material will be compressed somewhat to boost it’s average level, but almost all recent, heavily compressed and high-level material will simply be turned down a little.

Now some of you will be saying – “ReplayGain ? I’ve heard all of this before – so what’s the big deal about Spotify ?”

Easy. Spotify is going to take over the world.

Think about it – the first evening I paid my £10 subscription and installed Spotify on my iPod, I downloaded over 200 songs. 100% legally. A week later it was over a thousand. Some are old favourites, some are new experiments. None of them are things I would have bought on CD or even mp3. And as long as I pay my subscription, I can listen to them wherever and whenever I want.

Pretty soon, this is how everyone will listen to music. (Except it’ll be with a lossless codec :-D ) Eventually nobody will own any hard copies at all, it’ll all just sit in the cloud waiting for them to call on it.

More importantly, Spotify is where everyone will listen to new music first.

And all those people listening to music for the fist time on Spotify will have volume normalisation enabled by default.

Meaning if you want your music to sound great and leap out of the speakers at you, it doesn’t need massive level, it needs great dynamics.

Game Over.

PS. Metallica aren’t on Spotify yet, but when they are, do you think they’ll still be impressed by the 2dB-dynamic-range sound of “Death Magnetic” ?

Ian Shepherd is a mastering engineer and producer at Sound Recording Technology. He writes the respected Mastering Media blog, and runs the Production Advice website.

  22 Responses to “How Spotify will end the Loudness Wars – Ian Shepherd”

  1. The audiophiles I know still turn to piracy in order to get music first. Spotify will have an edge when it can get new, high quality releases faster than the pirates can.

  2. @ Alex – the 320 kbps Ogg Vorbis stream that Spotify offers subscribers sounds pretty damn good. And, you can listen instantly. When you factor in the effort needed to locate a genuine FLAC torrent, wait for it to download, rip and upload to the player of your choice, versus just typing the name into Spotify and hitting “play”, Spotify still wins, for me.

  3. Does it check peak or RMS values and normalise. Most normalise peak which is useless of course, I expect you are talking about RMS normalisation ?

    I don’t expect this will end the loudness wars, you need a digital format that normalises the audio in the file reguardless of the player, until this happens or Spotify conquers the world and destroys all other CD players, online stores and oh yeah Apple, I don’t see it affecting the loudness wars issue at all. Studios, engineers and mastering engineers are the ones who master the songs under the direction of the artists and labels for the consumer.

    The key is the consumer understanding of the volume wars and demanding more dynamic audio.

  4. I love Spotify. It sounds great, and we use it at the studio to compare our mixes to commercial ones when mixing, it’s such a handy reference tool.

    I also love it because it means I don’t have to fill my drive with songs that I probably don’t want to listen to more than once. And I have started to wonder if this is not only my inclination. I think Spotify is radically altering the way that we consume music. Up until now you have had to posess the audio, and whether you’ve paid for it or not, that implies some level of ownership. When you own things, you care about them. Now you don’t own it, I think how you value it changes too. And I think you just care less about volume and finish when you don’t own it. It’s like when you see a band live, you’re not disappointed at a few rough edges.

    I’m not saying it’s leading to an idea rich, but unpolished free for all, but i believe it’s going to have a long term effect on all our conceptions of what quality music or sound is. And for those of us that do the mixing, we have to be aware of this change in listening habits.

  5. @ audioooooyeah – ReplayGain uses a combined RMS and pyschoacoustic model:

    http://replaygain.hydrogenaudio.org/calculating_rg.html

    The results sound quite convincing, to me, anyway. I agree that Spotify alone isn’t the answer, but in order for the public to demand dynamic audio, they need to hear a fair comparison to make that choice. This has never happened before, but I think that Spotify and it’s inevitable competitors will make this happen, and the return of dynamic audio will follow naturally.

    @ julian

    Interesting point ! I certainly agree that Spotify will change people’s listening habits – I wonder if you’re right that it will mean more “in progress” versions out there ? I think currently Spotify only accepts music that has been released on CD, but this may change.

  6. I read somewhere that during the transition from CDs to online stores the record companies saw the opportunity to boost the level of ALL their catalogs to accommodate current standards. If so, Spotify would be getting this “remasters”, and instead of winning the war we would be signing the final surrender. Am I right? I’m from Chile, so i can’t test anything.

  7. Sorry Ian but this article is completely misguided. Almost so right but so wrong.

    The headline seems correct – spotify could contribute to more dynamic music because it’s not being pumped through hardcore radio compression, and I’ve noticed after trawling through a few recent ‘best of’ spotify playlists that a lot of the popular albums are really nicely dynamic. Grizzly Bear is sticking out for me in this respect.

    Thing is, in this article you’ve suggested that a compressor is the end solution to the loudness war. With respect, this is a staggeringly unintuitive conclusion to come to.

    @pure_tone did a post where he tested the dynamic range of the spotify loudness control, and to my memory he concluded it was a pretty hard limiter. I’m looking for the article now and will try to post it here when I find it.

  8. @pure_tone replied to me:

    “That all turned out a little moot really. My only comment was that it kinked the image in places. Ian did the tests”

    ..which surprised me, I was almost certain that he had concluded it was killing the dynamic with a limiter but I guess I was wrong on that point, my apologies.

    Still – isn’t this kind of normalisation yet another reason for engineers to compete with loudness, and therefore a contribution rather than solution to the ‘war’?

  9. Just to set the record straight – Spotify uses normalisation, and a limiter. BUT most modern stuff (anything with DR12 or less, roughly speaking) will be turned DOWN, so no extra compression is applied.

    More dynamic recordings will be boosted, and mildly limited to avoid clipping distortion, but this should be fairly unobtrusive (I haven’t listened carefully to see if this is true in practise). People who don’t like this idea can disable the normalisation, but paradoxically the loudest recordings will then have extra limiting added ! This is because the raw decoded Ogg Vorbis stream results in overlevels, which Spotify currently limits.

    Obviously this isn’t ideal, and I’ve encouraged Spotify’s engineers to pad everything a couple of dBs to avoid it, but as far as I know they haven’t done it yet.

    So, I stand by my argument – Spotify will even out level differences reasonably well, applying minimal processing. Which means the best way to sound good is to have your music professionally mastered, but keep to the kind of dynamic range that has always sounded good – DR14 to DR12.

    The only gripe I have with Spotify’s audio quality is that currently it doesn’t use dither – but that’s a whole other blog post !

    Ian

    PS. The differences that Tone and I thought we were hearing originally were due to the limiting applied to loud tracks when normalisation is off. Anyone interested can read about it in this post:

    Why I was wrong about Spotify

    and the whole original messy story itself is still online here.

    Ian

  10. Ian skipped my comment…

  11. @VIOZ Oops, sorry – not deliberate, honest !

    I don’t think there is any kind of consistent attempt to re-master music louder for online use – most record companies couldn’t organise a pea soup in a brewery, in my experience !

    More seriously, I think it is true that if music is re-mastered today, it is pretty much guaranteed to be louder – but this isn’t necessarily a bad thing. It’s quite possible (and desirable) to get a CD to a higher average level than a straight transfer from analogue tape, without any detrimental audio effects.

    And, you ca get it quite a bit louder than that and have it sound good. Personally I believe every piece of music has it’s loudness “sweet spot” – where it sounds great. Go much louder and it suffers, stay too quiet and it doesn’t reach it’s potential to sound great on the widest range of equipment and listening environments.

    As far as Spotify goes, they are just using CDs as sources, and if you disable the normalisation feature there’s a wide range of levels present, in all genres.

    Does that help ?

    Ian

  12. Hi Ian,
    I posted a series of articles i wrote about the future of radio / music, including some NYTimes references.
    In them, i predict the rise of computerized internet music services like Pandora & last.fm in the US (and Spotify in Europe) and the eventual demise of satellite radio and traditional human programmed radio.

    Thanks
    http://enigmafon.com/2009/05/07/the-future-of-radio/
    http://enigmafon.com/2009/05/29/the-future-of-radio-part-2/
    http://enigmafon.com/2009/06/08/the-future-of-radio-part-3/

  13. Spotify is not available at my country…

  14. Music is saved!

  15. i like spotify!

  16. I think you have missed the point of what the loudness war is about.
    Yes everything is getting louder and louder. has been since the 80′s and CDs but its the clipping and dynamic range that are considered the main aspects to the loudness war. the fact that you have to turn down your speakers is just an inconvenience.

    With the volumes so high you lose the ability to have a wide dynamic range which obviously in turn causes a track to be at more-or-less the same volume regardless of whether its a quiet or loud section of the song.

    Once a track has been recorded and bounced and is loud to the point of clipping there is NOTHING you can do about reclaiming the sound above the clipping unless you go back to the original recording and remove all of the make up gain and dynamic-range compression to return the track back to a dynamically versatile piece of music that is within the limits of peak volume.

    Normalization and limiting on Spotify is there purely for consumer convenience of not having to play around with their volume control constantly. The quality of the audio is going to remain just as bad because the peaks of the louder sounds are simply not there any more. you cant bring them back with technology.

    I supose my point is even after reading your entire post you havn’t mentioned a single thing that is actually relevant to the Loudness War so don’t title an article something it is not about. Thank you.

    • No, you’ve missed my point. Check out my blogs, you’ll find I have a pretty solid grasp of the “Loudness Wars” :-)

      Let me try again:

      Because making things sound louder removes all the dynamics, when the replay levels are equalised, more dynamic material will sound better. Spotify and others equalise the replay loudness by default.

      So the fact that the damage can’t be undone is the whole *point* – squashed “loudness casualties” will always sound worse, and I believe eventually people will come to realise this fact, and stop crushing things to begin with.

      What’s the point in squashing them when they sound worse – which they always will, when replaygain is used ? QED !

      Hope that helps ?

      Ian

      • This one is going to run and run! Great blog Ian, hope to post more from you soon :-)

  17. see if you can connect spotify to facebook and twitter and then have one company controlling everything everyone does… Surely your hunger for world domination is as bad as sony, emi etc. Everyone will hear all of their new music on spotify first? One company at the heart of everything is not good

Sorry, the comment form is closed at this time.

   
© 2011 All content is copyright RecordProduction.com/Blog No content to be copied or reposted on any other website Suffusion theme by Sayontan Sinha