In the audio world "bitperfect" essentially just means the digital end of your dac is getting the file's datastream 1:1 without any resampling or effects etc. Making sure you have bitperfect is usually not an issue and virtually no one would be able to tell the difference between bitperfect vs non-perfect in most circumstances unless the imperfection is really egregious, such as a really lossy transcode, very low sample/bitrate, or an operating system applying it's own effects and EQ on top of a signal which changes the response of what you hear. Also note I said "digital end" of the DAC, if the DAC is made poorly or is busted or can't output a properly reconstructed signal from 20hz-20khz whether or not the signal is bitperfect becomes less relevant anyway. Not something that is an issue with modern dacs usually, especially highly rated ones you'd find on here.
I don't know how well the Android SRC behaves but usually with modern resampling, the typical result even if somewhat "flawed" is still usually "audibly transparent" as far as human hearing is concerned- what you'd encounter even with a less than perfect SRC is maybe some artefacts many decibels (typically over -100 dB) down. So, not anything you'd realistically be able to hear, and definitely not after it's masked by the audio signal itself in music. I don't believe vanilla android does anything strange to the signal other than resample and maybe adding some effects of lowering and raising the volume of playing media when you have an incoming call or notification. There could very well have some sort of "audio enhancements" from certain manufacturer's android builds or device-exclusive software that do stuff like bass boosting or volume limiting, maybe specific EQs if you're using a phone's speakers vs using another audio source like the headphone jack or a USB dac, but I'm sure in some cases audio enhancements could be turned off in settings and ignored completely by using the DAC in exclusive mode with software like UAPP.
Shifts in a response on the other hand is something that is clearly audible in a multitude of circumstances and may be a product of dac, amp, and loudspeaker(+room)/headphone functioning so this is why the the FR takes higher precedence here over the simple virtue of a signal being "digitally untouched". By buying good dacs/amps we eliminate their contribution to FR so now we only have to worry about the speaker or headphone which is what EQ aims to correct. We were not in the minds of the artists during the production process as
@IAtaman touched on, but what we do know is that the artist+producers likely produced/finalised/listened to the file you are listening to in or approximating certain conditions, and the response in said conditions is probably the best representation of what they wanted you to hear, so the most effective way to achieve "as the artist intended" generally speaking is to also approximate those conditions: a flat in-room response generally produced by high-quality studio monitors or other flat loudspeakers in controlled or semi-reflective conditions. Unsurprisingly, research also says this is what humans prefer to listen to. In the case of headphones, the harman research serves as a proxy for these findings so the appropriate target would be something closer to harman-like. If your setup cannot achieve near those responses on their own (few can), EQ is what is used to get closer to enhance the sound which obviously involves losing the "bit perfection". And if you don't like that sound you can always just EQ to another preferred target or use various targets for different conditions.
That being said, "bitperfect" is more or less a technical concern/diagnostic tool to ensure your source-to-dac interaction is perfect, and perfection does not necessarily mean it will sound any different to you, let alone better. As I mentioned already, good modern sample rate conversion is virtually transparent and if I were to bet on it I'm sure the android one functions fine. Bitperfection has more relevance in music production when working with digital files and processes than it does with listening/reproduction.
So to put it simply: choosing bitperfection over EQ will sound worse 99.9% of the time unless there is some other underlying problem with the source.
Depends on the target curves too. Some are too flat for people that are used to listen to ultra colored phones...
