With the amount of high-quality video content being produced solely for online use now, it’s important for post-sound mixers to have an understanding of how a mix will translate to the designated platform for end use.
And while music streaming platforms such as Spotify and Apple Music generally have clear guidelines for loudness targets and file delivery, the specifics for video streaming platforms such as Vimeo or Instagram are still somewhat of a grey area. I therefore thought it would be a useful exercise to do some online digging and see what I can uncover.
YouTube
So what audio specs are actually out there? The short answer is not that many (at the time of writing at least...) The only mainstream platform I’ve consistently found the same (or at least similar) information for is YouTube. These specs are:
Loudness -13 to -15LUFS
Peak -1dBTP
These figures come courtesy of the Mastering the Mix site, which I find to be very reliable. This is also thought to be correct by RTW as set out in this guide on Worldwide Loudness Delivery Standards.
Table courtesy of Mastering The Mix
The A Word
Assuming that these YouTube specs are correct, we now have something to shoot for, but how are they regulated by the platforms? The short answer is that they are all governed by algorithms which have the ability to normalize any uploaded audio.
The algorithms can essentially do one of two things to the loudness of uploaded video content. It can turn the volume down – either by reducing the volume of a whole mix in dBs, or by hard-limiting to prevent a ceiling being exceeded. Alternatively, it can turn the whole mix up in dBs.
A Quick Thought on Algorithms
While the word ‘algorithm’ doesn’t normally incite particular enthusiasm, my feelings are that, as mixers, we should not be trying to ‘beat’ the algorithm, but we should respect it and try to align sonically with what it is trying to achieve. Ultimately, the algorithm, like QC, is not there to screw up your mix, but is in fact there for the benefit of the end user by ensuring that every viewer gets a predictable and consistent experience. The algorithm therefore fulfils the role of an automated QC… of sorts.
Loudness Penalty
What makes things slightly more confusing is that the algorithms operate in a fairly non-transparent way, so we don’t always know what’s going on behind the scenes. Again, I haven’t been able to find specifics on the majority of platforms, but let’s again take YouTube as an example.
According to White Sea Studio’s video on the subject, YouTube will turn down the volume if the loudness is too high. However, it will not normalize it (turn it up) if it’s too quiet. Therefore, if a film has been mixed in line with EBU R128 (-23LUFS), it will be playing considerably lower than the optimum level in online playback. Consequently, making your painstakingly mixed and sound designed documentary sound really quiet next to other uploaded content (that’s possibly been really badly mixed!) or adverts.
A useful article from YouLean also confirms this.
Taking a Punt
My general research would suggest that most online platforms are normalising to around -13 to -16 LUFS. And I think we can largely agree on a maximum True Peak of -1dBTP. Aiming for a LUFS reading of around -15 therefore feels sensible to me. Much hotter than that and the dialogue can become over-compressed and limited which will reduce dynamics and smoothness.
Other Considerations
A final thought is that it’s worth remembering that, in addition to hard limiting or normalisation, the algorithm may apply data compression. Therefore, don’t be surprised if your nice juicy .wav is chiselled down to a meagre 320mp3 or even lower.
Comments