what about the speakers telling?
(and try the ghetto-blaster! the second version is a change in the high end but I did not give the time to correct the low end to keep it symmetric, just wanted to show that there is something possible.)
your other song I can't comment too much. seems you found a "local maximum" (this is a phrase in mathematics science) and to change anything makes it worse. you may experiment from start and try very different versions. the total maximum may be much different and even much better.
somehow it reminds me on neneh cherry! you could listen through her early albums and compare what kinds of sounds are there ("manchild" and the like). It is also beautiful but does not sound ready. I did not have any idea what to do with it, compared to the first song. but I am not a professional.
my first advice is to check monitoring. if you need it professional, you might call the guy with the bruel&kjaer room analyzer and install and tune a 30-band eq just for the room and the monitors (one for each pair).
also the headphones are very critical, you might use 2 pairs, one that is very linear, and one like these horrible K55 that the kids listen in the big media shops where you have 30 CD players in one line, these have enormous bass.
in my opinion the most important thing at the moment is to get the bass right and make the voice thicker _and_ brighter with less audible effects.
then you could find an experienced friend to check through all your devices and plugins. some of them may sound very "digital" and can't be used without very good technical knowledge. it depends on the setting, you might identify presets that typically disturb the sound.
is there any 44/48 conversion happening in your system? when you are using 48k samples, you should do the whole thing in 48k and only convert when you burn the CD. if you have mixed samples you may decide to convert and correct the samples with the best software available so all is one sampling type.
so the sample player need not resample twice, i.e. once for the musical note and once for the other system sampling frequency.
depends on how sophisticated the technology in the stuff already is...
what I found is a spectrum that has many ripples, spikes and steps. not the other song, that was better. seems it is in the voice processing! maybe snare also. seems there are effects that sound nice in the first moment but disturb the spectrum.
after compression, there is also an amount of noise to be found. you could identify the track or effect where this comes from.
what I did: some tech stuff that needs knowledge of signal processing, and some bold changes by instinct, which I can't explain. just practising for fun.
- extensive analysing and flattening out the spikes with FFT and free-hand graph EQ.
- re-shaping the whole with 30-band eq which is a matter of style and taste. there I could revive the bass sounds (somehow..) and give the voice more personality
- soft de-esser
- a little short reverb from waves TrueVerb, with very narrow bandwidth (maximum 1k-4k, hipass for early=500hz)
this should glue the whole thing together, but is bad for the kick drum which I must live with, when having only the mix. you can try a second reverb with letting only some instruments out.
- overall old-style 3band eq
- a limiter ("peak master" by steinberg which is very brilliant for beginners and quite useful for many others!)
- second take: I did a process to change high end transients, above 7k. this runs in cool edit and is an algorithm I developed by myself. It can by some extent undo the damage the limiter does to the transients, and enhance poor material, like an exciter, but without being so non-linear. you could try different types of these transient designers, as well.
then a little brightness enhancement with a simple FFT curve.
well thats it
I'm happy you liked it! The best compliment is when people say that's a good song, when they did not realize this enough, before. I am looking forward to the final version.