Magin bakom mastering är inte någon magi alls, det är kunskap, erfarenhet, fantastisk akustisk och högtalare som kan återge allt korrekt in i minsta detalj. Den här videon från Sonic Scoop med Joe Lambert (http://joelambertmastering.com) visar hur man med små välavvägda förändringar kan åstadkomma något som låter väldigt mycket mer än lite EQ och några dB volym. Det här är vad vi gör, det är både det enklaste och det svåraste inom ljudteknik på en gång.
Spotify uses Replay Gain which is one of the earlier (if not the first) technique to even out the volume between both individual tracks and whole albums so that the listener perceives every track equally loud no matter how hot the original master is.
After some user complaints about their playback level, Spotify added a 3dB boost to the Replay Gain specification and it’s those 3dB’s that’s now been removed!
So what’s the big deal?
Here’s the thing with Replay Gain, for most popular music it’s great… Even with a 3dB boost. In 99% of the top 50 hits, the only thing that will happen is that Spotify will turn down the volume with anything between 1-3dB. It won’t change the audio other than turn it down, pretty much as if you lower the volume yourself. But if you play a really dynamic and mostly quiet piece of say classical music or jazz, Replay Gain will add loads of gain to accomplish the same perceived loudness as in the top 50 tracks. By doing so, the very dynamic and mostly quiet recording will clip at it’s loudest section, sometimes by several dB’s and thereby cause distortion.
Spotify has dealt with that problem by adding a limiter post Replay Gain, however, it doesn’t sound that good and would need to be more transparent to be considered a good solution. all that said; With the 3dB boost removed that scenario are less likely to happen and the Spotify listening experience will be much more enjoyable. So, overall, well done Spotify!
Here’s my wish-list to Spotify:
- Improve your Limiter and make it ISP Aware or…
- [better] – Scan every file for their highest peaks and only increase the playback level if the peaks remain below 0dB or…
- [best] – Skip Replay Gain and implement the AES recommendations for streaming music (-16LUFS, -1dBTP… Quiet masters are only pushed up to -1dBTP).
- Add uncompressed playback if the client’s internet speed allows it (I know they’re experimenting with switching codecs to increase quality at low speeds, this would be the next step)
- Spotify actively tries to trash the EU volume regulation of PMP’s which is just stupid and played a big part in the initial 3dB gain to the Replay Gain standard (I don’t know the status of the issue, a quick google says it’s a status quo).
– You can read more about Replay Gain here: http://wiki.hydrogenaud.io/index.php?title=ReplayGain
About once a week I get the question -“What do you think about automated mastering services such as Landr?”. There’s a very short and a profoundly more detailed answer. I tend to go for the short answer since the real answer is a lecture and I’ll probably lose you half way through. The short answer is: Remove mastering from that sentence and I’ll tell you it can be a useful tool to test your mix. Keep mastering in the sentence and I’ll tell you that it sucks monkey balls.
Luckily I stumbled across an article written by Mastering Engineer Justin Perkins (Mystery Room Mastering) that pretty much cover everything there is to be said about Landr and similar services, so from now on I won’t answer, I’ll just give people this link:
It seems like if Spotify silently implemented an album mode to their loudness compensation algorithm!
I was doing some online loudness experiments when I happened to notice that the volume varied on a track in Spotify depending on how I listened to the track. If I browsed my way to the album it was quieter than if I were listening to the same track from a playlist.
Spotify has been criticized for not taking the inherent dynamics within an album into account when compensating for loudness so I was quite happy with my findings. Especially since I just finished master an album for a movie that had huge dynamic differences and I had done some crazy edits to trick Spotify to interpret the loudness in a way so it wouldn’t totally ruin the dynamics of the album and even if it worked it may have sounded better without those edits.
Test it yourself
I made an album with three versions of the same song. The masters are pretty much the same except that they have different loudness targets. The tracks are named with the original files LUFS value, the True Peak value (dBTP) and the Sample Peak value (dBFS). After uploading it to Spotify I created a playlist with just the quietest and the loudest version (if I included all three Spotify handled it as an album play). I played the same part of the track from the Playlist followed by the Album version and measured it all with IzotopeRX and here’s the result.
Note that the original LUFS value is the whole track while the Spotify measurement is made from just the loudest part of the track. I choose to save me some time since what’s matter is that it measures differently and not the values themselves. If I would’ve measured the whole track from the Spotify stream it probably ended up at about -11* (the Spotify target) and not -9LUFS.
*Spotify has since this test was performed lowered their loudness target by -3dB. You can read more about that here: https://www.saintpid.se/spotify-lower-their-loudness-target-by-3db/
We have a list of things to consider before you hand them over to your mastering engineer but my friend and colleague, Sigurdór Guðmundsson, did an extended list with some great additional points. Head over to his site and get inspired!
Vi har justerat våra priser för att göra det mer fördelaktigt att skicka fler spår för mastring, t ex album men även EP’s är nu billigare än tidigare medans enskilda spår kostar marginellt mer. Istället för de olika paketpriserna har vi nu en kalkylator där du enkelt kan få en fingervisning om vad mastring av ditt projekt kommer att kosta.
Vi har fortfarande timdebitering vid större projekt, kontakta oss för offert!
Great news, Tidal joins the club and start to normalize loudness according to the ITU-R BS.1770-4 standard.
I’m not sure what to think about that every streaming service use different standards, it’s not all bad (I know what I think about Spotify’s approach though; Hey Spotify, skip the extra 3dB that you added to the Replay Gain standard). The different standards keep us mastering engineers awake and maybe we do a better job trying to make it sound as it should on every single media plattform out there. Maybe… Just maybe…
To be loud ain’t a competitive advantage, it is and has always been a misconception based on ignorance.
Last week got a little bit crazy, a full mastering slot schedule plus the annual audio fair in Sweden (LLB) where I wanted to attend several seminars plus that I was invited by SSES (Swedish Sound Engineers Society) to be part of a panel debate with Spotify about their loudness practice and target.
The thing with Spotify is that they use ReplayGain, a bit of a DIY loudness normalization… Maybe call it DIY is a bit harsh, it is a proposed standard put together back in 2002 by David Robinson. If you translate ReplayGain to LUFS you end up at about -14LUFS, which isn’t too far away from the AES recommended -16LUFS. So what’s the thing, why stand on a stage and yell at each other?
Let’s make a list:
- ReplayGain does not take True Peak into account when measure loudness so you could argue that the measures are wrong from begin with.
- A former employee decided to add about 3dB’s since he felt that iTunes played some tracks at a higher volume, so the current target level is about -11LUFS.
- To handle the peaks above 0dBFS Spotify added a limiter, but it’s not ISP aware which means that it still might cause distortion in your D/A.
- AES recommend a True Peak of -1dB, this means that Spotify’s loudness target end up 6dB higher than the AES recommendations and even more if you take ISP/TP into account.
- ReplayGain doesn’t change the gain in the same way as EBU R128, it’s basically does more conservative changes to loud tracks, i.e. there’s a risk that their practice fuel the loudness war even further.
- With the current practice; more dynamic mixes such as Jazz and classic recordnings will be boosted and limited by Spotify while loud masters will still be louder. So the practice punish music that are mastered to sound good while music mastered for loudness will be left alone (maybe turned down a dB or two but they will still sound the same).
Spotify are able, without too much of a hassle, to skip the extra 3dB and reach a target of -14LUFS. That’s the middle way and much will be won If they do just that.
To implement a new loudness measurement practice would cost a fortune and they’re not willing to do that at this point.
I personally would appreciate if the AES recommendations would become law and force all streaming services to change their current practice and adapt an uniform standard that would benefit audio quality both now and in the future, all would benefit from this.
To be loud ain’t a competitive advantage, it is and has always been a misconception based on ignorance.
ISP, Inter Sample Peak or TP, True Peak, refers to peaks in the analog domain. That is, how your peaks will look after your waveform has been converted from a digital stream within your computer, phone, iPod or CD to electrical impulses that can be amplified and played back through your speakers.
Aren’t the waveforms the same as they appear on my screen after the D/A conversion?
No, Your D/A will create a waveform that continuously moves from one sample point to the next. In that process the newly created waveform may peak above the 0dBFS ceiling which individual samples adhere to, thereby causing distortion. Just how bad this distortion is depends on the converter used. In some extreme cases, the actual peak can be as much as 3-4dB above the sample peaks detected in the digital domain.
Many of the top mastering business don’t care about ISP so why should I?
No, you don’t have to treat or be aware of ISP but:
- Digital limiters without oversampling (or other ways of detect ISP) will react to the digital waveform, not the true waveform, and thus won’t react as a limiter should react. This does not mean it will sound bad in any way, just that it don’t react as intended.
- -Tip: If you want to work with limiters that lack oversampling; up-sample your 44.1 or 48KHz projects to a higher sample rate and by that have the limiter react more accurately.
- We’ll most probably have a loudness standard for streaming media within a couple of years and that standard will be ISP aware and set at -1dBTP. This means that if your audio peak at +2dBTP it will be turned down by at least -3dB before reaching the end user. All that extra loudness you gained by letting random equipment create a positive peak (that might distort) will be lost so it will be meaningless.
- Encoding to lossy formats; ISP will make it harder for encoders to do a good job. If you don’t have the tools or knowledge to check how your audio will perform post encoding I would recommend to stay away from positive True Peaks.
Since it’s more or less guess-work, or at least really hard to create a real-time limiter that handles ISP perfectly, we thought we’d put as many limiters as possible to the test. So a handful of mastering colleagues and I (Ian Stewart, Sigurdór Guðmundsson and Johan Eckerblad) went to work.
We used a mastered track that we boosted by 7.5dB (the peaks were at -1dBTP/dBFS in the mastered file so max Gain Reduction would be 6.5dB), we set the limiter ceiling to -1.0dBFS (or -1dBTP if available in the plugin), rendered it as 44.1KHz, 32bfp wav and measured the True Peaks of the resulting waveforms. Limiters with a result as close to -1.0 dBTP as possible have handled ISPs the best, whereas values over -0.8 dBTP (shown with increasing values in white, through yellow, to red when they exceed 0 dBTP) mean that the limiter has failed to handle ISPs to within the margin of error for measurement we observed during testing (more on that in a bit).
Here’s the track that we used:
- “Yes” or “No” in the TRUE PEAK column indicates whether the developer mentions or claims that the plugin handles True Peak or ISP, either in the manual or marketing. A “No” does not necessarily mean that it doesn’t handle or try to handle ISP, just that it’s not explicitly stated by the developer.
- The results published here are measurements taken using Izotope RX. Additionally, we did measure with Sequoia, Wavelab, and Nugen, and even though we observed slightly different results, they were negligible. The biggest differences were in iZotope’s products. Since they probably use the same detection algorithms in all their software, this was expected. The other software measured within a 0.1 to 0.2 dB difference, which still puts iZotope amongst the best performers.
- Where other settings such as Attack and Release were present we used either the default or adjusted it to a value as close to standard as possible
- This test did not take into account sonic qualities at all. That will have to wait for our planned podcast or another post.
- It’s not recommended to use any of the limiters with a positive score in 44.1KHz projects. They should handle ISP better as your sample rate goes up.
- If your favorite Mastering Limiter ain’t here, send us a link so that we can try it out. However, we’re not in the business of buying every limiter out there, so any you’d like us to try must have fully working demo-versions.
Since this was posted in early 2016 Spotify have both implemented an album mode and lowered their loudness target by 3dB.
This is the first of a series of loudness experiments with two goals. First; we wanted to hear for our selves how different online platforms handle different loudness targets. Second; hopefully this will make it easier for anyone to understand loudness and if you produce music you’ll get a better idea of how your desired loudness target will affect your audio on current distribution platforms. This will enable you to make decisions based on knowledge rather than a perception of a loud vs. a dynamic master in a quick A/B comparison (where I would say about 90% choose the louder version whether or not it actually sounds better).
This is a WIP. Measurements and more releases will follow…
Most music retailers such as Spotify and Youtube loudness compensate their content so that you as a listener don’t have to change your volume control every time you change song or video clip, and some don’t (Soundcloud for example). This experiment gives you the opportunity to listen to the same track mastered at different loudness targets on the most common current music platforms and to the raw files that where uploaded. We recommend these steps:
- Download the raw files and extract them: https://drive.google.com/open?id=0B1rylQZfCdreU05PRXdHZFRzZG8
- Load all songs in a player that do not compensate for loudness. We recommend Foobar for Windows and VLC on OSX.
- Compare the raw files with the audio from online retailers
- Bonus: There’s level matched versions in the LUFS.rar archive so that you easy can hear how the loudness targets affected the sound.
Links to the most common platforms:
- iTunes – Apple seem to have some problems with the release, we’ve contacted their support.
We distributed the album to lots of common online stores but choose to link to the most interesting ones when it comes to loudness. If you’re interested you easily find the album by searching for Saintpid or LUFS.
Details about the experiment
We took one track (Truck Driver by Noisebud) and mastered it to 10 different loudness levels, all with the limiter set to -1dBFS. We named the tracks according to their dynamic range (DR) and loudness values (LUFS), and released them as one album with the 10 versions as is first, then with all versions level matched to make it easier to compare audio quality. The track in itself is pretty dirty (Noisebud call their music electronic punk) so this is no audiophile experience. That’s why we plan for this to be a series of releases, with styles ranging from pop to metal, singer songwriter to techno, and other electronic genres.
We’re searching for material so if you want to contribute please contact us at info(at)saintpid.se. You get your song mastered for free but that’s the only compensation we can offer at this point.
Here’s a Key-value report from the software Toscanalyser.
The interesting values are the RMS average, RMS max, Peak max, Peak average, Sub Bass (SB%) and the K-system values (K12-20).
The chart show some notable facts.
- The amount of low-end has almost been cut in half in the loudest version vs the most dynamic one (1.1% vs 2.0%).
- The last version that show an OK K12 value is the DR5 -8.6LUFS version. We should’ve stopped here if not before.
- The last version that gets a safe K12 value is the DR8 -12LUFS version. This is the version the artist (Noisebud) prefer.
Here’s the DR report from the offline Dynamic Range Meter:
Thanks to Sigurdór Guðmundsson (http://skonrokkstudios.com/) and Ian Stewart (http://ianstewartmusic.us/) for ideas, consultation and trying to push the track to DR1 (which Sigurdór finally did but it’s not on the album).
This is the first more thorough measurement. We did Spotify first because we needed the data for a meeting this Monday where SSES (Swedish Sound Engineers Society http://www.sses.org) and Spotify having a sit down to discuss loudness normalization. Or rather, we’re going to make a plan on how to discuss the issue in public during the LLB audio fair in Stockholm next month. Hopefully, we’ll manage to influence Spotify to bring down their loudness target to the AES recommended -16LUFS (http://www.aes.org/technical/documents/AESTD1004_1_15_10.pdf).
Since this article were posted early 2016 Spotify has lowered their loudness target by 3dB and implemented an album mode. We leave the measurements here for historical reasons but strike everything that no longer apply.
We measured the tracks in both “Album mode” and in “Playlist mode”. We did that because it makes sense to loudness normalize tracks from different releases in a playlist but if you play an album you would prefer the internal gain differences between tracks to be intact as the artist, producer or mastering engineer intended. To be fair, there are situations when you might want to loudness compensate within an album. For example in a noisy environment or listening on tiny tiny speakers, but the way to go would be an extra user setting specific for those situations.
Sadly, Spotify currently adjusts the loudness independently even in album mode.
The Album as a whole (Integrated BS.1770-2/3)
With Normalization turned on:
Peaks -2.11dBTP (-2.13dBFS)
Measures from the album playback (we only measured the same three as for the playlist mode)
“DR2 -4.2LUFS” measures -10.2LUFS with peaks at -4.36dBTP
“DR5 -8.6LUFS” measures -10.8LUFS with peaks at -1.33dBTP
“DR12 -16LUFS” measures -13.2LUFS with peaks at +0.92dBTP The recorded waveforms of “DR2 -4.2LUFS”, “DR5 -8.6LUFS” and “DR12 -16LUFS” in both Album and playlist mode measures the same.(scroll down for the screenshots of quick measurements for Youtube and iTunes)
We will do more thorough measurements of the following streaming services as soon we have the time. Meanwhile, Sigurdór recorded the whole playlist from the most common places and you can see the results here:
Youtube perform their level normalization within a day or so after the file being uploaded. My guess is that they leave it as is at first so when people check the upload they don’t get confused when the level dropped by half. These screen captures are taken after the level normalization kicked in. It’s the -16LUFS to the left and the -4.2LUFS to the far right. We could’ve used 8 bits for the loudest version and saved some bandwidth. =)