Hi

Regarding recording levels, it is said it's generally good practice to work with average levels floating around -20 to -15dBFS, and highest peak levels no higher than -6dBFS to allow plenty of headroom for additional processing or mastering.

Will using the above recording levels be acceptable for online streaming platforms? And what will be the new relationship between dBFS and the new LUFS standard for measuring loudness?

Apple Music -16 LUFS
iTunes -16 LUFS
Spotify -14 LUFS

While working on a project and mixing you should keep the levels down. Once you have a mix you are happy with, the mastering stage is where you adjust the levels for whatever format you want to release it on. There have been a few articles in the magazine about mastering for different platforms and there's a video series at

https://www.soundonsound.com/techniques ... -mastering

and something from Hugh at

https://www.soundonsound.com/techniques ... udness-war

Thank you for the links, helpful.

So complying with the recording levels above.. staying with dBFS would be an acceptable volume level for online streaming?

DigitalMusicProduction wrote:Will using the above recording levels be acceptable for online streaming platforms? And what will be the new relationship between dBFS and the new LUFS standard for measuring loudness?

Different tools for different jobs.

The sample-peak meters (dBFS) you see on your DAW channels and outputs are there to help you adjust the signal path gain staging and to ensure you are recording and mixing with a sensible headroom margin. They show the instantaneous peak level of (typically) a small group of digital audio samples.

The Loudness meter (LUFS) is an entirely separate thing intended as a means of quantifying perceived loudness which is something you need to know about when mixing and/or mastering -- the stage after recording! It is weighted -- meaning it prioritises frequencies that are most audible to us -- and is averaged over time (either 300ms, 4s, or the entire track length, depending which format you're using -- the last being the most important for streaming services.

So the two types of meter are used in different stages of the audio recording and mixing/mastering processes.

That said, the LUFS value is loosely related to the average value when gain-staging (the -20 to -15dBFS level), but will tend to be higher on an LUFS meter, especially with a percussive source like piano. So if you track with an average level around -20 the LUFS meter will probably end up around -16LUFS -- although this relationship is highly dependent on the nature of the source and the style of music.

Nevertheless, setting the gain structure as previously recommended will get you into the right ball park for streaming services, but with the emphasis on maintaining a sensible headroom margin to minimise the risk of overloading anything and to leave some space for any post-processing.

Once you have recorded your track (and applied any desirable processing) you can then measure the review with an LUFS meter, and then adjust the overall level of the finished track as a second and separate process to comply with the requirements of different streaming services.

dBFS tells you nothing about the perceived loudness as it measures all frequencies equally at a given point. LUFS is weighted so that the frequencies that our ears are most sensitive to are considered more important than those that we don't hear so well. In addition, LUFS is usually averaged over the whole song.

When mastering you would normally use a limiter to keep your peaks (measured in dBFS) below a certain value (say -1dBFS) while increasing/decreasing the level until you reach your desired level measured in LUFS.

Hugh Robjohns wrote:Nevertheless, setting the gain structure as previously recommended will get you into the right ball park for streaming services, but with the emphasis on maintaining a sensible headroom margin to minimise the risk of overloading anything and to leave some space for any post-processing.

So if the rule of peakIng up to -6dBFS is to leave headroom for Mastering, In my case seeing as its a straight solo piano recording with no intended effects or Mastering to be applied, i could just as well increase the gain and peak up to 3dBFS ?

You could.... But then your performances will probably suffer because you'll be worrying about peak overloads all the time, or your perfect performance will be accidentally ruined by an unintended peak overload.

It's really just not worth compromising headroom during recording. You lose far more than you gain.

Tweaking -- mastering if you will -- your recording to optimise the LUFS figure for your chosen streaming service is something that can be done simply and quickly at the end of your review of your recorded performance(s). And it's best done as a second and separate stage of your recording process.

I’m not sure where you got 3 dBFS from!

If you’re going to release on streaming services, you need to aim for a peak dBFS of no more than -1 on the master channel.

If you just have the one channel, you can of course make it peak at -1 dBFS if you want, though you should record with it peaking at a much lower level and adjust upwards when mixing. Or stick to the standard range for the instrument channel as recommended by Hugh and put some gain onto the master channel. If you’re not using effects on the channel, it doesn’t matter too much, but if you are, it’s generally better to take the second approach.

You need to take an informed decision on what LUFS value you are aiming for. Solo piano can have a very wide dynamic range, which means that without compression or limiting, a peak of -1 dBFS can easily be reached with a LUFS value of -20 or so.

You need to decide this based on the style of the music - classical, pop, relaxation, jazz all might need something different. Compare you track to examples of a similar genre.

DigitalMusicProduction wrote:So if the rule of peakIng up to -6dBFS is to leave headroom for Mastering, In my case seeing as its a straight solo piano recording with no intended effects or Mastering to be applied, i could just as well increase the gain and peak up to 3dBFS ?

It is a little confusing, and you mainly have to go slow and to resist the temptation to jump to conclusions before having understood it all

A mastering engineer may apply a little EQ, make parts of the track mono, look at the stereo width, etc... some this stuff will change the level of some of the frequencies in your mix, possibly increase it a little bit (in the sense that it has some peaks end up higher up in the full scale than before). It can also decrease it (if for example the mastering engineer see fit to make only cuts) or leave it unchanged (if the mix is perfect as it is).

If your mix was already peaking at the top of the scale, there would be literally no space to do these adjustments without overshooting and distorting the signal. Also, most equipment works best when it's not really pushed to the limit of its performance envelope, so it's best to leave a little more headroom for processing than a little less.

So: the at-least-6dB-of-headroom you conventionally leave in a mix is there to give the mastering engineer space to work comfortably, while also keeping the gear in a good working range.

You may reduce it to 3dB - but you will (potentially) make the mastering engineer's job harder and make certain improvements he would like to do impossible. Or, you may force him/her to gain down your track before applying his/her processing, which reduces the signal/noise ratio and therefore the quality. And like doctors, for mastering engineers it's important to "first, do not harm" (to the signal, not people). 6dB is a good rule of thumb compromise, especially when the recording is already (potentially) as high quality as a 24 bits resolution one.

So that's that. Seldom the mastering engineer will need to use all the headroom, but it's safe to leave it (or in the ballpark) just in case.

Now, that convention was born in an environment where the final product was generally a CD, so the assumption was that any remaining headroom would be removed by the mastering engineer by limiting the track as the last step of mastering.

Note that "limiting" can be just enough gain to bring the highest peak to 0dBFS - doesn't necessary need to compress anything (and in practice, a very small amount of compression on a few highest peaks would be unnoticeable anyway). A gentle limiting will simply remove any leftover headroom, which is just fine with CDs, which do peak happily at 0dBFS (so long there's no intersample peaks but that's a side benefit of using limiting as opposite to just the gain knob).

With streaming, the expectation is that the track will be attenuated as necessary to hit the target LUFS. Whether not attenuation happens really depends on the material, since as Hugh explained LUFS is a perceptual measurement, taken over the course of the entire track.

The track may be attenuated, or it may not - regardless if it has peaks near 0dBFS or not.... sure, something that's near 0dBFS all the time will surely need attenuation.. but it's got nothing to do with the headroom you leave for the mastering engineer.

He/she's gonna need headroom to work anyways.

As of LUFS, it's not something your should really think much about. Just do the best mix you can, have it mastered properly and you'll be fine. If you are interested, I wrote more details on why in my post here.

I think the OP means he's going to be doing his own 'mastering' - in this case a bit of levelling really - in which case, once he's done the mixing he can just raise the level to his target.
The reference to -3dBFS probably comes from the advice to keep plenty of headroom left after mastering in case it's going to be compressed to a very lossy codec - like a low bitrate MP3 - as this can introduce artefacts that would cause clipping if it's been mastered louder.
If I remember correctly.

Regardless, even if you're going to 'master' it yourself, treat yourself just like you would a separate mastering engineer - give yourself the headroom to do a proper job of it.

There are good, solid reasons to adopt and maintain a way of working that's broadly in line with the best industry practices, and this is certainly one of those times - especially important when you're starting out. It's worth getting into the habit and sticking to it until it become second nature - don't fudge it for the sake of it.

Afaik Apple, iTunes, have moved to -14LUFS -1dB True Peak

blinddrew wrote:I think the OP means he's going to be doing his own 'mastering' - in this case a bit of levelling really - in which case, once he's done the mixing he can just raise the level to his target.

That's correct, the project being worked on is a solo Piano album intended as a commercial recording for digital distribution and online streaming, no heavy processing or effects involved, maybe some EQ at the most.

To summarise then, floating at around -15dBFS to -20dBFS with no peaks any higher then -6dBFS is best recommended for a clean signal recording and streaming services?

DigitalMusicProduction wrote:To summarise then, floating at around -15dBFS to -20dBFS with no peaks any higher then -6dBFS is best recommended for a clean signal recording...

Correct.

And then, as a separate stage in creating your album, you can 'master' those individual recordings to make them compliant with the specs for the streaming services.

That means using one of the Loudness metering tools to assess the Integrated Loudness of value of each track and the True Peak level. Some tools require playing the track in real time, others can do it as an off-line process by analysing the file.

Once you have the numbers for your raw track you can assess how they compare to the target loudness specified by the streaming service, and raise/lower the track level as necessary. If raising the level you may need to introduce some brickwall limiting so ensure the transient peaks don't exceed the maximum True Peak value allowed by the streaming service.

Alternatively, you could just upload your raw recorded tracks and let the streaming service do any processing required automatically.... However, although all services will turn excessively loud tracks down, not all will raise the level for tracks with an Integrated Loudness value below the target. So it's generally best to correct the tracks yourself so that you know it's as good as it can be.

The Loudness penalty website is a really great site where you can drag and drop your track, and it will analyse and tell you what different streaming services will make of it.

https://www.loudnesspenalty.com/

And for a bit of further reading:
https://www.soundonsound.com/techniques ... udness-war
and
https://www.soundonsound.com/sound-advi ... udness-war