DSP - Filter

[DerPicknicker]

Hi @maximmaxim345,

as discussed here is the feature request. It would be great if the DSP would support the following Filters:

• Loudness
• Limiter / Compressor
• High/Low Pass with frequency setting

Best regards & many thanks for your work.

[maximmaxim345]

• Loudness

Can you explain what you exactly expect from such a filter? Could this also be a preset for a Compressor or Limiter, or should this be something separate?

• High/Low Pass with frequency setting

You mentioned this is for use with a subwoofer. How is the sub connected? (I mean, how does it play right now?)

[maximmaxim345]

If you mean sending the same DSP to all grouped members. One idea is to add a DSP option to permanent sync groups.

[vortexrotor]

I could definitely live with that and a step in the right direction. Can we put that into roadmap or maybe into this beta?😎

[vortexrotor]

Also, since we are on the subject of the DSP... suggest making the DSP PEQ scale 1Hz to 20Kh. I am an audiophile and actually have setup AOP presets that have settings at sub 5Hz. I would be nice to see them. 😁

[maximmaxim345]

You can already type in any frequency above 0 and below 100kHz.

[vortexrotor]

👍🏻 Yep. I know but the current scale cuts off the visibilty of it. It's more of an aesthetic thing.

[OzGav]

All of the filters in this featuire request are now implemented with the exception of loudness. If this is still desired then another feature request can always be opened but after research my understanding is that the main technical challenge is that loudness compensation isn't a fixed EQ curve. The amount of bass and treble compensation needs to change continuously with the listener's actual playback level, based on equal-loudness contours such as ISO 226.

That means accurately knowing the acoustic level reaching the listener, which is difficult because the DSP only knows the digital signal level not the actual SPL at the listener's ears. You'd need assumptions or calibration around volume, amplifier gain, DAC output and headphone/speaker sensitivity, and those vary enormously between setups. You then need to calculate and apply a changing EQ curve without introducing audible jumps, pumping, clipping, or excessive bass/treble as the volume changes. There are also different approaches to the target curve, and human hearing varies significantly between people.

So unlike a compressor, PEQ or convolution filter (which we now have!) where the requested processing is explicit and deterministic, a "correct" loudness compensation setting requires information Music Assistant generally doesn't have and involves a lot of subjective choices.