acn_editor (Customer) asked a question.
How can you control the stereo downmix loudness within an ADM file?
Apologies if this has been answered on the forums already, but I couldn't find a specific solution.
We are currently mixing our podcasts in Atmos but of course still forced to distribute 2.0 files until the platforms accept the full files. We do this by offline bouncing the re-renders within Pro Tools alongside export of the ADM file. I get that the 2.0 mixes (whether stereo or binaural) reach different loudness measurements, and the best course of action is to make adjustments to the re-rendered files (which we’ve been doing using RX’s Loudness Control). For example, we have a mix that comes out to -19.4 LKFS / -4.2 dBTP for the Atmos measurement, but -18.5 LKFS / -0.5 dBTP for the stereo re-render.
Now let’s fast-forward to whenever a platform like Apple Podcasts or Spotify allow Atmos distribution of their podcasts. Does this mean that if eventually someone was to play/reference the stereo version of the ADM master file, the stereo downmix would be louder and exceed -1 dBTP? In other words, how can we adjust/ensure that the stereo or binaural downmix of a file won’t be too loud, without simply turning down the entire Atmos mix?
Note that I am NOT referring to multi-channel deliverables, where one would deliver separate 7.1.2, 5.1, 2.0, etc. files of the same mix; I get how you can easily adjust that. I’m specifically referencing the downmix encoding done by the ADM file. For example, if someone was to play an MP4 of the mix with Atmos/Spatial Audio “on” and then turn it “off” to reference the stereo version.
We are currently mixing our podcasts in Atmos but of course still forced to distribute 2.0 files until the platforms accept the full files. We do this by offline bouncing the re-renders within Pro Tools alongside export of the ADM file. I get that the 2.0 mixes (whether stereo or binaural) reach different loudness measurements, and the best course of action is to make adjustments to the re-rendered files (which we’ve been doing using RX’s Loudness Control). For example, we have a mix that comes out to -19.4 LKFS / -4.2 dBTP for the Atmos measurement, but -18.5 LKFS / -0.5 dBTP for the stereo re-render.
Now let’s fast-forward to whenever a platform like Apple Podcasts or Spotify allow Atmos distribution of their podcasts. Does this mean that if eventually someone was to play/reference the stereo version of the ADM master file, the stereo downmix would be louder and exceed -1 dBTP? In other words, how can we adjust/ensure that the stereo or binaural downmix of a file won’t be too loud, without simply turning down the entire Atmos mix?
Note that I am NOT referring to multi-channel deliverables, where one would deliver separate 7.1.2, 5.1, 2.0, etc. files of the same mix; I get how you can easily adjust that. I’m specifically referencing the downmix encoding done by the ADM file. For example, if someone was to play an MP4 of the mix with Atmos/Spatial Audio “on” and then turn it “off” to reference the stereo version.
Apologies for the massive delay. This wound up in the wrong queue and was just forwarded to me today. The process of creating a re-render from a Dolby Atmos master whether from Pro Tools or the Dolby Atmos Renderer with not clip as an output limiter is applied to prevent this. If the streaming service does not request separate stereo or multichannel deliverables they would derive them from the Atmos master in the same manner. In general these would be encoded as HE-AAC and the client device would pick the appropriate stream from the manifest. There is no stereo downmix within the ADM BWF itself. Atmos is encoded as Dolby Digital Plus JOC. In the case of Spatial Audio the Dolby Digital Plus JOC is decoded to 7.1.4 and then virtualized as Spatial Audio. If it is turned off while playing an MP4 file it is downmixed to stereo from the 5.1 core (DDPLUS JOC is encoded as 5.1) using the downmix settings applied when the ADM BWF is created. The dialnorm metadata in the bitstream is used to ensure the downmix is not louder than other presentations of the same content. In Apple devices there is may be further normalization applied.
I hope this helps and again apologies for the delay.
Best,
Adam