Sanskrit Studio

Sanskrit Studio is at project-vyasa.github.io/vyasa-apps/sanskrit. Locally: cd apps/sanskrit-studio && bun run dev → http://localhost:5374/

No account, no upload. The engine is WebAssembly from vyutils. If the engine is missing, a banner says so — rebuild WASM and restart the dev server.

It is not a reader of packed publications. Use the Viewer for that.

The Guide control in the Studio header opens this page (a sibling app on GitHub Pages; locally it is the docs server on port 5375).

The window

Region	What it is
Header	Logo, Experimental badge, activity title, Guide, Labels, theme and density
Activity bar (left)	Three tools: Lipi, Pāṭha, Varṇa
Main pane	The tool you picked

Labels change how Sanskrit words in the chrome are written (menus, titles, pattern names). Canonical source is IAST. English glosses stay English. This is separate from output script inside Lipi or Pāṭha.

Lipi — transliteration

Rewrite the same Sanskrit between 11 scripts, keeping Vedic pitch marks.

Open Lipi.
Choose from and to scripts.
Type or paste in the left box (or tap a Vedic mark from the palette).
Read the right box. The lossless badge means a round-trip matches the source script.

The seed is ṚV 1.1.1, Śrī Rudram (TS 4.5.1.1), and BG 2.47, blank-line separated. Default target script is Malayalam.

Pāṭha — recitation patterns

Start from Pada-pāṭha (words already split, usually with | or daṇḍas). The studio can emit Krama, Jaṭā, and Ghana. The other eight traditional patterns are in the Patterns overlay.

Open Pāṭha.
Put pada text in the full-width editor.
Pick Krama, Jaṭā, or Ghana.
Read the interlinear (Krama: one chant line; Jaṭā/Ghana: isolated padas over the chant). Space moves the highlight; Shift-Space goes back. Ghana units that cover three padas are labelled 1-2-3.
Set output script for the generated text (independent of Labels).

The editor is padas. Pasting continuous saṃhitā will not split it into words.

Varṇa — phonetics

Inspect one string against a tradition. The seed is the first word of ṚV 1.1.1 (same opening as Lipi).

Open Varṇa.
Paste a short string, or pick an example (arkaḥ, kanyā).
Click a sound in the akṣara strip (each block is a syllable: mātrā, pitch, visarga).
Switch Pāṇini / Ṛgveda-Prātiśākhya / Taittirīya — the same selection is re-described.
Read the Engine / Label only / Not in engine list for that tradition. Citations there are the honest coverage of this build.

Tradition	What you get
Pāṇini	Which pratyāhāras contain this sound (1.1.71), which Śiva-sūtra it sits on, Śikṣā inspector
Ṛgveda-Prātiśākhya	Inventory groups (Śaiśirīya order, samānākṣara, nāmin as a label). Nati (s → ṣ, n → ṇ) is not applied
Taittirīya	Karaṇa in the inspector; worked svarita junctures (class from a named context, not from raw verse); dvitva on the selected consonant with TPr 14.1 / 14.4 / 14.8

Sandhi rewrites with a rule trace are not in Varṇa. Forward join already happens inside Pāṭha; a dedicated Sandhi activity is later work.

What the engine implements

Live means WASM evaluates it. Label only means a classification or inventory with no rewrite. Not in engine is a known traditional rule this build does not run.

Tradition	Citation	Status
Pāṇini	Śiva-sūtras 1–14; Aṣṭ. 1.1.71; Śikṣā inspect; akṣara split	Live
Pāṇini	Named sandhi sūtras (6.1.77 ff.) as a trace	Not in engine
RPr	1.1–2 inventories; Śaiśirīya order; 1.65 nāmin	Label only
RPr	Nati (retroflexion)	Not in engine
TPr	Ch. 2 karaṇa/sthāna; 14.1 / 14.4 / 14.8 dvitva; Ch. 20 class from a named juncture	Live
TPr	Svarita from a recited string; Raṅga (Ch. 17)	Not in engine

What this app is not

Not a dictionary, morphology analyser, or meter scanner.
Not a full Vedic recitation trainer for every traditional pattern.
Not the place to open a packed Ṛgveda or Gītā — that is the Viewer.
Not a critical edition of the Prātiśākhyas.

If a script shows tofu (empty boxes), the webfont is not loaded yet. Conversion still runs; the glyph may be missing.