< Back to IRCAM Forum

Concerning Ambisonic Mix Strategy

Hi Thibaut !

I am currently doing an ambisonic mix on some quite inharmonic and distorted sound sources. I have a hard time in finding a mix strategy where they project well. Currently exploring ambisonics I have so far not been so succesfull : my sources sound muffled and as if remaining inside the speakers. It could of course be the speakers with these particular sound sources. Or the ambisonic panning method.

But to gain knowledge of ambisonic mixing I would like to raise the following questions :

I am searching for a procedure that has good clarity, localisation and projection. I am aware there is a discussion comparing VBAP and HOA on this question, where VBAP is generally considered more “clear and focused”, whereas HOA is considered “ambient and enveloping” (both equally good qualities, of course). However which ambisonic decoder is generally considered the most neutral in frequencies and phases?

When combining many audio layers I am wondering if there would be any difference in adding up many encoded signals as compared to having one central encoder? If there is any difference what could I expect it to be? Any coloration in spectrum or possible phase-issues?

The same consideration concerns decoding : is one central decoder better that adding up many centralised ones? Better meaning not influencing the signal (even if less efficient CPU wise).

Usually I work in 7th order. But sometimes find that I may have to truncate the signals to 5th order. Would there be any difference in signals of a purely 5th order encoded signal as compared to a 7th order encoded signal truncated to 5th order? Yes, I realise that by truncating I loose in precision in localisation, but spectrumwise?

Thanks, kindly Hans Peter

1 Like

Hi Hans Peter,

These are interesting, yet challenging, concerns/questions.
I won’t be able to give you definitive answers, but I hope other users of this forum will offer their feedback.

However which ambisonic decoder is generally considered the most neutral in frequencies and phases?

I cannot answer that. All I can say is that we’re most often using the energy-preserving decoder (with in phase/max-re) because it generally offers the best trade-off between all criteria (global spectral balance vs sweet spot size vs localisation accuracy etc.)

if there would be any difference in adding up many encoded signals as compared to having one central encoder? If there is any difference what could I expect it to be?

Encoding and decoding are linear processes, so having a “central” decoder vs multiple ones would not make any difference. (except for CPU usage)

Yes, I realise that by truncating I loose in precision in localisation, but spectrumwise?

Technically, the decoder is not supposed to alter the spectrum; whether you use 5th or 7th order, the whole frequency range is processed. However, what you perceive (the combination of all loudspeakers signals to your ears) might be different, both in terms of localization (and sweep spot size) and spectrum.
So, spatial quality and spectral quality is somewhat intertwined.

Hope this helps,
T.