- Thread Starter
- #21
If science is not an opinion, when you convolve an audio signal with your personal HRTF and reproduce the result at the same point where the HRTF was measured (supposed in the auditory canal) you will perceive that sound coming exactly from azimuth/elevation to which was the source for measuring the HRTF, obviously within the limits of localization capability of the human.
I hope this is not questioned.
Now, if Atmos is not working for one person, the reasons are to be found in the elements of the specific chain.
The possible culprits are essentially 4, from my point of view (assuming we are talking about headphones listening):
1) You are not listening to Atmos binaural mix, but 2.0 downmix (highly probable seen the mess behind the various ways of distributing / playing Atmos)
2) Atmos encoded with HRTF that corresponds little to the personal one. The human physiognomy is highly variable, consequently it is impossible to establish a universally valid HRTF. There are studies for preference that highlight statistically more preferred HRTF, but we are still in the field of statistics so there is inevitably margin for incompatibility.
Furthermore, measuring an HRTF perfectly is only possible with dummy heads that have microphones at the end of the auditory canal, exactly where our biological transducer is located. There are two methods for measuring personal HRTF today, that I know:
- with a microphone in the ear (ex. Knowles), which depends on mic calibration accuracy and does not integrate the effect of the auditory canal;
- estimation with a digital photo + neural network, which in addition to depending on photo processing depends on the training of the network;
Therefore the accuracy of the HRTF will always have a margin of uncertainty that can cause perceptual alterations.
3) The processed signal have to be reproduced with linear amplitude and phase headphones to be faithful to the HRTF, but we know that a perfectly linear reproduction system does not exist, so this too will determine further alteration in perception. We also know how most headphones have deliberately non-linear amplitude tuning to provide a good result with non binaural audio (the vast majority). And even where one linearizes the response based on a calibration, there will always be the uncertainty of the calibration to potentially introduce alteration.
Then, headphones in ear and over ear are not the same for binauralization purposes.
4) The mastering of the Atmos tracks is a complicated task, because it has to be compatible with different reproduction system, therefore space inaccuracy could take over if it is not done specifically for binaural.
There are probably also secondary factors, such as bone conduction, but they should affect perception less.
Aforementioned elements, however, are more than determinants in spatial perception, so it is clear that discourse cannot be generalized.
Atmos, as long as others immersive encoder, works as long as the chain is effective.
How statistically effective it is I don't think we can determine it... but for this very reason it seems absurd to me to say it is a lie and doesn't exist.
Even the earth is spherical but it does not seem to the eyes...
Last edited:
) that has an iPhone and uses it to listen to music in Atmos, by connecting to a DAC&? Thanks!
