The Upanishads, Part 1 (SBE01), by Max Müller, , at sacred-texts.com
1. The udgîtha, of which a poet said, I choose the deep sounding note of the Sâman as good for cattle, belongs to Agni; the indefinite note belongs to Pragâpati, the definite note to Soma, the soft and smooth note to Vâyu, the smooth and strong note to Indra, the heron-like note to Brihaspati, the dull note to Varuna. Let a man cultivate all of these, avoiding, however, that of Varuna.
2. Let a man sing 3, wishing to obtain by his song immortality for the Devas. 'May I obtain by my song oblations (svadhâ) for the fathers, hope for men, fodder and water for animals, heaven for the sacrificer, food for myself,' thus reflecting on these in his mind, let a man (Udgâtri priest) sing praises, without making mistakes in pronunciation, &c.
3. All vowels (svara) belong to Indra, all sibilants (ûshman) to Pragâpati, all consonants (sparsa) to Mrityu (death). If somebody should reprove him for his vowels, let him say, 'I went to Indra as my refuge (when pronouncing my vowels): he will answer thee.'
4. And if somebody should reprove him for his sibilants, let him say, 'I went to Pragâpati as my refuge: he will smash thee.' And if somebody should reprove him for his consonants, let him say, 'I went to Mrityu as my refuge: he will reduce thee to ashes.'
5. All vowels are to be pronounced with voice (ghosha) and strength (bala), so that the Udgâtri may give strength to Indra. All sibilants are to be pronounced, neither as if swallowed (agrasta) 1, nor as if thrown out (nirasta) 2, but well opened 3 (vivrita), so that the Udgâtri may give himself to Pragâpati. All consonants are to be pronounced slowly, and without crowding them together 4, so that the Udgâtri may withdraw himself from Mrityu.
33:2 These are lucubrations on the different tones employed in singing the Sâman hymns, and their names, such as vinardi, anirukta, nirukta, mridu slakshna, slakshna balavad, krauñka, apadhvânta.
33:3 It would be better if the first ity âgâyet could be left out. The commentator ignores these words.
34:1 Grâsa, according to the Rig-veda-prâtisâkhya 766, is the stiffening of the root of the tongue in pronunciation.
34:2 Nirâsa, according to the Rig-veda-prâtisâkhya 760, is the withdrawing of the active from the passive organ in pronunciation.
34:3 The opening, vivrita, may mean two things, either the opening of the vocal chords (kha), which imparts to the ûshmans their surd character (Rig. Prât. 709), or the opening of the organs of pronunciation (karana), which for the ûshmans is asprishtam sthitam (Rig. Prât. 719), or vivrita (Ath. Prât. I, 31; Taitt. Prât. 5).
34:4 Anabhinihita, for thus the commentaries give the reading, is explained by anabhinikshipta. On the real abhinidhâna, see Rig. Prât. 393. The translation does not follow the commentary. The genitive pragâpateh is governed by paridadâni.