Even more EQ bands are necessary for many users

[saratoga]

One has to wonder with such complex hearing if perhaps a dedicated audiophile music player wouldn't be a better option than any rockboxed unit?

The software is pretty awful on most so that's not a great option.

Anyway I don't see a convincing argument in favor of more bands so I think we will probably keep it at 10 bands for newer devices and perhaps reduce older devices to fewer bands since that seems to be a common problem for users.

[ZoSo]

since that seems to be a common problem for users.

How is having too many bands available for use a problem for anyone? Are people really so dumb as to enable too many EQ bands on very old hardware, find their music lagging/skipping, then can't figure out that it's the EQ causing it, so they come on these forums to complain that your software sucks? Honest question.

Anyway I don't see a convincing argument in favor of more bands

I don't see a logical argument against having more bands available for use. Those who don't need the extra bands simply won't use them. Those who do need the extra bands will be glad to have them. Those who find that enabling too many bands causes their music to lag or skip can simply disable bands until the music no longer lags or skips. I fail to see how making your software more powerful and versatile would be a bad thing unless you feel there's more dumb people who would use too many bands, cause their music to lag, and not realize it's the EQ doing it than there are competent people interested in precisely shaping their desired sound.

My argument for more bands would be as follows: more bands would make your software more powerful and versatile while not adversely affecting the competent user in any way. I bought a Sansa Clip+ specifically because it can be Rockboxed and specifically because I wanted a portable with a parametric equalizer. I was disappointed to discover only 8 peak bands were available and I'm sure I'm not the only one. In fact, I'm quite sure that "X-band parametric equalizer" (if you were to raise it to X) would be an attractive selling point to many consumers, causing them to purchase Rockbox-compatible hardware specifically for it. Rockbox is already very attractive software to audiophiles for its parametric EQ and crossfeed, both of which are rare on portables, and making its EQ more powerful would make your software even more attractive to audiophiles. I've been a member of the Internet audiophile community for quite a long time now. Word about good products spreads very rapidly and virally on the Internet. Making Rockbox more powerful and attractive to the audiophile would certainly lead to word-of-mouth about Rockbox spreading quite virally across various audio communities on the Internet.

[bluebrother]

How is having too many bands available for use a problem for anyone? Are people really so dumb as to enable too many EQ bands on very old hardware, find their music lagging/skipping, then can't figure out that it's the EQ causing it, so they come on these forums to complain that your software sucks? Honest question.

Yes.

Search the forums / IRC log if you want to. We had pretty much 0 questions about this until the number of EQ bands was raised.

I don't see a logical argument against having more bands available for use. Those who don't need the extra bands simply won't use them.

They still require resources, and that's on a system with rather limited resources. A DAP isn't a PC. On the one hand people put a great deal of work into making playback as efficient as possible, and on the other hand we should put in more workload for the CPU to figure if it needs to do some more EQ calculation? That doesn't make sense.

My argument for more bands would be as follows:

As already has been said: Rockbox is open source. You can recompile it for yourself with as many EQ bands as you like. If there's really that much interest in the audiophile community someone can do that and provide builds for others. We have unsupported builds since years. But increasing flexibility for a minority of people causing the majority to get confused and increasing our support workload is simply a really bad idea.

Can we now stop this argument? It's getting really annoying, as well as the topic title is (I could start a topic saying "less EQ bands are necessary for most users", which would be -- from support experience -- much closer to reality than your "more bands" for "many" users. How many are "many" anyway?)

[ZoSo]

Yes.

Search the forums / IRC log if you want to. We had pretty much 0 questions about this until the number of EQ bands was raised.

But increasing flexibility for a minority of people causing the majority to get confused and increasing our support workload is simply a really bad idea.

Fair enough. I understand why you don't want to add bands now. Trying your best to please everyone must be a hard job. Thanks for making your great software and discussing this with me like gentlemen. My workout headphones are still sounding pretty great right now with only 8 peak bands and crossfeed and I'm thankful for that.

[ulmutul]

I just had a short look a the source. The setting you may want to change is in this file:
rockbox/lib/rbcodec/dsp/eq.h
You can try out to set EQ_NUM_BANDS to whatever you like and recompile.
I didn't try it for myself, so I can't confirm this works. Maybe you have to do other changes as well.

[phr]

I had the impression at least at the start of this thread, that the OP didn't understand the difference between parametric eq (a few bands with variable center frequency and variable Q) and traditional eq (lots of fixed bands with fixed Q).  But I can't really tell.

It does seem to me that increasing bands would increase CPU load, maybe beyond what the CPU in a little player like the Sansa Clip is capable of, and also mess up the phase response at normal sample rates.  Maybe it could work better with 96 khz sampling if the source material and cpu power was there?  I do like the idea of making crappy earphones sound like good ones, through the magic of customized equalization.

[saratoga]

I had the impression at least at the start of this thread, that the OP didn't understand the difference between parametric eq (a few bands with variable center frequency and variable Q) and traditional eq (lots of fixed bands with fixed Q).  But I can't really tell.

I think part of the problem is that our interface doesn't make it easy to figure out what you are doing.  We just sort of assume the user understands how to use an EQ correctly.

It does seem to me that increasing bands would increase CPU load, maybe beyond what the CPU in a little player like the Sansa Clip is capable of, and also mess up the phase response at normal sample rates.  Maybe it could work better with 96 khz sampling if the source material and cpu power was there?  I do like the idea of making crappy earphones sound like good ones, through the magic of customized equalization.

I don't know if the current implementation would even run at 96k, but in theory doubling the sampling rate just halves the resolution in the audible part of the spectrum while doubling the CPU requirement.