The a is closed and is similar to the u in but; the vowels ā, ī, and ū are long: for example, ī is pronounced like ee in feet, not like the i in fit; ṛ is a vowel and is pronounced by resting it on a barely discernible i or u. The e is closed as in better; the o is closed as in bother.
The g is always hard: for example gītā is pronounced geeta; the c is always soft: for example, cakra is pronounced chakra. S is always like the s in sound; ś and ṣ are more or less like sh in shun. Ṛṣi is therefore pronounced rishi. The retroflexed ṭ, ḍ, and ṇ are pronounced by bending the tongue back to touch the palate, as in utter, udder, or runner. The aspirate occlusives kh,* gh*,* ch*,* jh*,* th*,* dh*,* th*,* ph*, and bh are single phonemes, and are pronounced with an aspiration after the consonant: for example, ph is pronounced as in top hat, not as in telephone, and th is pronounced as in dirt heap, not as in think or father.
Ñ is like the n in country; h is a sounded aspiration as in inherent.
The accent falls on the last long vowel (for example: Prajāpati is pronounced Prajápati). The vowels e and o are also long. Furthermore, all vowels are long when positioned before consonant groups. If there are no long vowels, the accent is placed on the third or fourth to last syllable (if this is a root syllable). For example: Gáruda, Gótama, śrámaṇa.
A few words that are typically Vedic are stressed with a musical tone called udātta, which required the pitch of the voice to be raised, but this has disappeared in Classical Sanskrit.