FS#7204 - Greater EQ Customization

Attached to Project: Rockbox
Opened by Matt M (Chesteta) - Wednesday, 23 May 2007, 12:35 GMT
Task Type Feature Requests
Category Operating System/Drivers
Status Closed
Assigned To No-one
Operating System All players
Severity Low
Priority Normal
Reported Version Daily build (which?)
Due in Version Undecided
Due Date Undecided
Percent Complete 0%
Votes 0
Private No


I am using an Ipod nano however I could see anyone wanting this functionality: the idea is to be able to change the type of filter implimented in the EQ; we already have the ability to change gain, Q and frequency, why can't we change the type of filter (personally I would use all peaking EQ filters rather than a lowshelf, highshelf and 3 peaking EQ's) Also, if there is a possibility of adding a few more filters it would be cool too... granted this would be dependent on available cpu... any ideas or patches are welcome. Thanks!
This task depends upon

Closed by  Bj√∂rn Stenberg (zagor)

Reason for closing:  Fixed
Additional comments about closing:  Closing all feature requests.
Comment by max (suriaj) - Friday, 13 July 2007, 04:50 GMT
i agree

and FIR option would be great to, for the most powerfull players
Comment by Ecki (Oppaunke) - Monday, 11 February 2008, 20:34 GMT
I also agree. It wouldn't have to be too fancy, just switchable characteristics (shelving filters changeable to peaking filters).

But most important: I beg you for a LowCut (i.e. HighPass) Filter with the following specifications: 24dB/Octave (4th Order) and a tuneable cutoff frequency (between let's say 20 and 150 Hz). This would decrease (analogue) distortion at higher levels while most headphones are incapable of producing low frequencies - so why amplify them? Furthermore, with a well tuned cutoff frequency one would be able to boost low frequencies (even with a shelving filter!) without "mumble" the low end and/or stressing the Headphone(-amplifier) too much within the "useable" range.

This is a modification crucial to a sound-guy (/-fetishist) as myself and I would very much prefer it to an update in some game or plugin... ;-)
Comment by Dan Everton (safetydan) - Monday, 11 February 2008, 22:20 GMT
Changing filter types might be possible but I doubt you're going to get any more filter bands. The CPU is pretty limited on these things and, as far as I know, the EQ routines are as optimised as they can be. Maybe the Gigabeats might be able to handle more bands though.
Comment by Thom Johansen (preglow) - Monday, 11 February 2008, 22:33 GMT
The lowcut filter will be doable, but have you actually listened to see if this highpassing of high-level low frequency components is audible in the form of less distortion? Don't know if Ipods and Sansas will run very well with all five EQ bands and this lowcut filter at the same time too, since the lowcut filter will pretty much be comparable to another two EQ bands in CPU utilization.
Comment by Ecki (Oppaunke) - Tuesday, 12 February 2008, 01:01 GMT
Yes, it's audible. I tested the issue with preprocessed files (I'm a sound engineer). Besides, it will also slightly enhance battery lifetime (assuming same cpu load), especially with low impedance headphones.

And, in my opinion it wouldn't hurt too much to spare one of the EQ-Bands when the LowCut is activated (possibly by the means of a 3-way mode switch in the low band: LP24/LoShelf/Peak).
Comment by Thom Johansen (preglow) - Tuesday, 12 February 2008, 18:43 GMT
To original poster: what kind of extra filters do you have in mind? Ordinary lowpass, highpass, bandpass, etc? Is there really much use for these in an equalizer?
Comment by Matt M (Chesteta) - Tuesday, 12 February 2008, 19:00 GMT
I mainly wanted to use 5 peaking EQ filters (actually 7 would be nice but i doubt the cpu can do that) a "3-way" mode switch as described earlier would suit my needs very well.

Something else to consider: there is an EQ package I have been using for winamp called "AIXcoustic - Electri-Q" which has an option to generate an IIR from input EQ parameters. This way, however many filters are used, once the coefficents are computed, there is minimal CPU used. I am not sure if that is possible on rockbox. If it is, you could have an IIR generated with given eq parameters (may take 3-6 sec to compute, then route audio signal through that; in the end less cpu could be used. again, I have no idea how the audio routing works in rockbox, if IIR would take too much cpu or anything else... just another idea.
Comment by Thom Johansen (preglow) - Tuesday, 12 February 2008, 19:30 GMT
Any kind of sophisticated filter design is best done on a computer for two reasons: 1. it's heavy computationally, and will take a long time on a target with a slow CPU, 2. it's usually a bitch to convert things like this to use fixed point calculations. Simple parametric filters like we currently use are easy and fast to calculate, in addition to pretty much being the standard way to do EQs. Also, Rockbox currently uses IIR filters for the EQ, it's FIR (or at least longer ones) filters we'll have trouble doing efficiently.
Also, it seems like the product you mention seems to use pretty much the exact same filters we do... (as seen here