For a general account of vedic accents and their oral rendering one might refer to the following: vaidika svara-s.
The recitation of the R^igveda, the vulgate atharvaveda and kANva shAkha of the shukla yajur veda follow a generally similar pattern over much of bhArata-varSha. The nambuthiri-s of the chera country, especially while reciting the RV follow a highly derived set of rules that are encountered nowhere else. The widespread mAdhyaMdina-s and the Orissan paippalAda-s have largely lost the original svara-s and have only a pseudo-intoned recitation. The common pattern of the RV, the AV-vulgate and kANVa-SYV (not their shatapatha brAhmaNa) is the elongation of all long vowels and anunAsika consonants with associated svarita-s. Further, all these traditions render short vowel svarita by raising the pitch and exaggeratedly stressing the vowel. A deviation from this latter trend in short vowel svarita-s is only encountered in the R^ivedin-s from the mahArATTa country. In evolutionary terms the general pattern shared by these recitations appears to be the ancestral state of the rendering of the vedic text (contra what some white indologists and their fellow travelers had mistakenly thought it was).
The kR^iShNa yajur vedic recitations that survive (taittirIya and maitrAyaNIya) or survived until recently (kaTha) differed from the above in that they do elongate all svarita-s with long vowel or anunAsika-s. They only elongate long vowel svarita-s if they occur next to a consonant. They also elongate the ‘M’ anunAsika next to a daNDa. But there are some fine points in these rules.
Rule 1: If a consonantal is at the ‘n’ at the end of word, and is preceded by a long vowel bearing a svarita, and is not generated by or engaged in a saMdhi, then it shall merely be rendered in a high pitch with a distinctive stress. It shall NOT result in elongation of the long vowel bearing the svarita.
However, if the consonantal ‘n’ is inside a word or linked in saMdhi then it shall cause the long-vowel bearing a svarita and preceding it to elongate.
The following long vowel svarita-s in bold are not elongated but the ‘n’ after them stressed and raised in pitch:
…pibato janAn | (shata rudrIya last anuvAka)
An interesting case is when it is at the end of a word but not end of a sentence:
However, the following long vowels in bold with svarita-s are elongated:
sa ya eSho.antar hR^idaya AkAshaH | (in TU 1.13; svarita on Sha is due the abhinihita clause in the saMdhi between eSha and antar which in the KYV is rendered like a regular dIrgha svarita rather than as a kaMpa as in the RV or AV)
sendra yoniH | (in TU 1.13; here the consonantal n is inside a word).
In the KYV consonantal ‘n’ preceded by short vowel bearing a svaritA is not subject to nasal elongation as in the RV/AV/kANva-SYV tradition. However, there is such elongation under certain conditions.
Rule 2: If a consonantal ‘n’ occurs as part of a consonant cluster with two successive consonants (usually at n-s = n-th-s junctions) preceded by a short vowel svarita then the ‘n’ is nasally elongated.
The nasally elongated ‘n’ in bold and the svarita in red
tasminth-suparNo (3rd trisuparNa mantra)
tasminth-sahasrashAkhe |(in TU 1.9)