Mātṛkā Mantra Coder / Decoder

The mantroddhāra cipher made executable — name a syllable by its grid coordinates, or decode one back. Seeded with attested examples from the Brahma-yāmala and the Bījanighaṇṭu, plus the 159 named-mantra recipes of the Śāradātilaka.

Compose →
← Decode
Mātṛkā grid
Attested examples
Śāradātilaka recipes

Compose a syllable from coordinates

praṇava (oṃ) prefix bindu (anusvāra ṃ) visarga (ḥ) bare consonant (no vowel)

Decode a syllable, bīja, or code-word

Type something above.

The mātṛkā grid

16 vowels + 33 consonants. Click any consonant to load it into Compose. BY ch. II names consonants both by varga+position and by absolute ordinal.

Brahma-yāmala ch. II — Mantroddhāra-paṭalaḥ

The six Yoginī mantras extracted by coordinates (vv. 6–10), each click-loadable.

SyllableCoordinate recipeEncodesSource

Deity-on-letter map (vv. 15–17)

ElementGoddessSource

The six Yogeśīs by ending-particle (vv. 11–13)

EndingYogeśīSource

Bījanighaṇṭu — named bījas

The complementary code-word naming system (krodhīśa = ka, etc.). Decode tab resolves any code-word.

BījaEpithetsSource

Śāradātilaka — named-mantra recipes

The classic mantraśāstra digest of Lakṣmaṇadeśikendra (with Rāghavabhaṭṭa's Padārthādarśa). 159 named mantras across 25 paṭalas, each with its deity, ṛṣi/chandas, construction, and ŚT citation. This is the same overlay the bot consults — mantraśāstra, not a doctrinal source. Cite as ŚT C.V–W.

🔊 Where a recipe's realized mantra has been curated, it renders sound by sound — tap any akṣara to hear it on its own, ▶ syllables to hear each sound in turn, or ▶ word for a slow, connected chant of the whole mantra. Audio uses a browser speech voice (the nearest available to Sanskrit): treat it as an approximate pronunciation guide, not authentic recitation. Recipes still being curated show their construction prose only.

Chrome tip: if you hear nothing, pick a voice marked ✓ (or a “Google” one) and press Test. Drag Syllable length right to draw out each akṣara — at the slow end one sound can be held for several seconds. Exact length depends on the voice (the Web Speech API sets speed, not a fixed duration), so tune it by ear.