How does a human ear differentiate syllables? For example, I can say the word "for" in two different tones. From a physics point of view, the waveforms of the sound signals will be different, but the ear will decode both of them as "for". SO I guess there are some components of both signals that are similar. What are those components?
 A: This may take us a bit away from physics: in linguistics one distinguishes the disciplines of phonetics and phonology. The former deals with physical characteristics of sounds: frequency spectrum, length, amplitude, how it is formed/articulated, etc. The latter deals with the sound systems specific to different languages.
From the phonetic, i.e. physical point of view, one can form an infinite number of sounds: the number of consonants is countable, but very large or possibly infinite, whereas vowels form a continuous distribution (in terms of the manner of articulation, i.e. the position of the speech organs), i.e. vowels are not even countable.
However from the phonological point of view, every language has a finite number of sounds (phonemes) that are distinct in terms of the meaning that they convey. In other words, the space of possible sounds is cut in big chunks that are associated with a specific sound. Going back to the example in the question, there are many ways to pronounce [o], which however an English speaker would always identify with the same phoneme. This may be however not true for a speaker of a tonal language, such as Chinese, which distinguishes five different tones potentially conveying unrelated meanings.
In terms of "components of signals": all possible vowels can be classified (at least for non-tonal languages) by their manner of articulation (which determines their acoustic spectra)

*

*the position of the tongue (front/back and open/closed)

*the roundness (rounded/unrounded)

*the "nasalness" (normal/nasal)

*length (long/short)

Out of these possibilities Arabic comes up with 3 vowels (o, a, u and their long/short varieties), Russian has 6 (no nasals, no length distinction), French has 17 (nasals and more finely grained basic vowels, but no length distinction), English has more than 20 (although nasals are not contrastive).
Update
Getting closer to physics: perhaps somebody could comment on the resolution of the human ear in terms of frequency, spectral shape and sound length.
A: The processing that allows an English speaker to recognise a syllable like “for” when it is spoken in different tones, pitches and accents (and distinguish it from a similar syllable like “fur”) takes place in the brain, not in the ear. Waveforms that are very different as far as the ear and impulses in the cochlear nerve are concerned can still be perceived by a fluent speaker as the same syllable. And the identification of syllables is context dependent - it depends on the surrounding syllables.
On the other hand, someone who is unfamiliar with a language may struggle to even locate boundaries between phonemes and syllables when they hear that language.
In other words, this is really a question about neurology and linguistics, not about physics.
A: One may plot the frequency content of a signal against time. Consonants are found to be produced by rapidly changing frequency content, with recognisable patterns for particular consonants. The pattern in the way the frequency content changes is decoded to produce the consonant which one actually here. A particular pattern could be subject to a change in pitch, or tone, but this will leave the basic pattern and one will hear the same consonant.
