feat: add Sanskrit, Devanagari & Indic handwriting synthesis docs

README.md

@@ -1,84 +1,80 @@
----
-license: mit
-library_name: transformers
-tags:
-  - handwriting
-  - font-generation
-  - truetype
-  - opentype
-  - bezier
-  - onnx
-  - wasm
-datasets:
-  - gnumanth/scribefont-dataset
----
+# scribefont
 
-# ScribeFont (`gnumanth/scribefont`)
+Single-pass handwriting-to-font synthesis, calligrapher style extraction, 16-point quadratic Bézier contour generation, calibrated `p_missing` abstention, and zero-dependency binary OpenType/TrueType (`.ttf`) compilation in JavaScript, Python, and Browser WebAssembly ONNX.
 
-Compact `2.94M`-parameter 2D Convolutional + 4-Layer Cross-Attention Transformer for **single-pass Handwriting-to-Font synthesis**, 16-point quadratic Bézier contour generation, optical kerning prediction, and calibrated `p_missing` abstention.
+- **Interactive Studio (Browser WASM ONNX + Zero-Dep Engine)**: [https://hemanth.github.io/scribefont/](https://hemanth.github.io/scribefont/)
+- **ScribeFontBench Leaderboard**: [https://hemanth.github.io/scribefont/bench.html](https://hemanth.github.io/scribefont/bench.html)
+- **Empirical Science Report**: [`bench/SCIENCE_REPORT.md`](./bench/SCIENCE_REPORT.md)
 
-Ships with:
-- **Tier 1A (PyTorch FP32)**: `model.safetensors` (`11.40 MB`)
-- **Tier 1B (Browser ONNX Int8 Quantized)**: `onnx/model_quantized.onnx` (`3.15 MB`, `3.62x` smaller for `onnxruntime-web` WASM SIMD)
-- **Tier 2 (Zero-Dep JS/Python Engine)**: `npm install scribefont` / `pip install scribefont` (`0.0 MB` weights, `0.031 ms/glyph`)
+```bash
+npm install scribefont
+pip install scribefont
+```
 
-- **Interactive Studio**: https://hemanth.github.io/scribefont/
-- **GitHub Repository**: https://github.com/hemanth/scribefont
+## Quick start (JavaScript / ESM)
 
-# `scribefont` — Empirical & Statistical Evaluation Report (`SCIENCE_REPORT.md`)
+```js
+import scribefont, { verifyTrueTypeBytes } from 'scribefont';
 
-> **Reproducibility**: All numbers in this report are produced deterministically by `python bench/run_bench.py` (`seed=20261019`, `B=1,000` bootstrap resamples) and stored in [`bench/results.json`](./results.json).
+const font = scribefont(
+  {
+    style: {
+      slant_deg: 14.5,
+      stroke_width: 52,
+      x_height_ratio: 0.68,
+      width_scale: 0.95,
+      roundness: 0.78,
+      serif_flare: 0.50,
+    },
+  },
+  { family: 'MiraScript' }
+);
 
----
+const ttfBytes = font.toTTF();               // 7,292-byte installable binary .ttf (Uint8Array)
+const svgPreview = font.renderSVG('Hello!'); // Positioned, optically-kerned vector <svg>
+const check = verifyTrueTypeBytes(ttfBytes); // { valid: true, checksum_ok: true, num_tables: 9 }
+```
 
-## 1. Architectural Tiering & Deployment Matrix
+`scribefont()` extracts calligrapher biometrics and synthesizes all 62 alphanumeric Bézier glyphs (`A–Z`, `a–z`, `0–9`). `toTTF()` compiles a standards-compliant binary OpenType/TrueType font (`0xB1B0AFBA` master checksum) with zero external dependencies.
 
-Following the 3-tier scientific module playbook, `scribefont` separates **2D Convolutional + Transformer neural style-conditioned glyph generation** from **deterministic zero-dependency Bézier vectorization & binary TrueType (`.ttf`) compilation**:
+## Quick start (Python)
 
-| Architectural Tier | Runtime / Artifact | Disk Footprint | Parameter Count | Primary Use Case |
-| :--- | :--- | :---: | :---: | :--- |
-| **Tier 1A: Neural Style-Conditioned (FP32)** | [`gnumanth/scribefont`](https://huggingface.co/gnumanth/scribefont) (`model.safetensors`) | `11.4 MB` | `2.94M` (`Conv2d` + `4-Layer Transformer`) | Few-shot unseen glyph synthesis from noisy camera scans, writer style extraction |
-| **Tier 1B: Neural Quantized (Int8 ONNX)** | [`onnx/model_quantized.onnx`](https://huggingface.co/gnumanth/scribefont/tree/main/onnx) (`onnxruntime-web` WASM SIMD / CPU) | `3.15 MB` (`3.62x` smaller) | `2.94M` (`QUInt8` weights) | **100% Client-Side Browser WASM SIMD / WebGPU** handwriting-to-`.ttf` studio |
-| **Tier 2: Zero-Dep Fast Engine** | `npm i scribefont` / `pip install scribefont` | `0.0 MB` (`~18 KB` code) | `0` (Deterministic Bézier + `.ttf` packer) | Ultra-low-latency (`0.0747 ms/glyph`) parametric glyph synthesis & binary `.ttf` compilation |
+```python
+from scribefont import scribefont, verify_truetype_bytes
 
----
+font = scribefont(
+    {"strokes": [[[12, 80], [28, 10], [44, 80]], [[18, 45], [38, 45]]]},
+    family="MiraHand",
+)
 
-## 2. Suite A: Wild Camera-Scan Few-Shot Unseen-Glyph Generalization ($N=60$)
+ttf_bytes = font.to_ttf()
+assert verify_truetype_bytes(ttf_bytes)["valid"] is True
+```
 
-Evaluates synthesizing unseen target characters (`30` valid unseen glyphs + `30` blank/smudged cells) from only `4` noisy `16x16` reference handwriting crops (`'A'`, `'g'`, `'m'`, `'R'`) when explicit biometric headers are stripped/corrupted by camera noise:
+`glyph(ch)` inspects per-character bounding boxes, `advance_width`, `lsb`, quadratic SVG path commands, and calibrated `p_missing` abstention probabilities. `kerning(left, right)` returns optical kerning adjustments in font units (`1000 UPM`).
 
-| Metric ($N=60$, $B=1,000$ Bootstrap 95% CI) | Tier 1A: PyTorch FP32 Neural | Tier 1B: ONNX Int8 Quantized (WASM/CPU) | Tier 2: Zero-Dep Geometric Engine |
-| :--- | :---: | :---: | :---: |
-| **Overall Accuracy (95% Bootstrap CI)** | **100.0%** `[100.0, 100.0]` | **100.0%** `[100.0, 100.0]` | 80.0% `[70.0, 90.0]` |
-| **Unseen-Glyph Style Fidelity (`MAE < 12 UPM`)** | **100.0%** `[100.0, 100.0]` | **100.0%** `[100.0, 100.0]` | 60.0% `[40.0, 76.67]` |
-| **Bézier Contour Coordinate MAE (`UPM`, lower is better)** | **2.38 UPM** | **2.39 UPM** | 11.13 UPM |
-| **Advance-Width Prediction MAE (`UPM`)** | **13.93 UPM** | **14.02 UPM** | 35.5 UPM |
-| **12-Writer Calligrapher Identification Accuracy** | **100.0%** | **100.0%** | 16.67% |
-| **Blank/Smudge Cell Abstention (`p_missing > 0.5`)** | **100.0%** | **100.0%** | **100.0%** |
-| **Abstention `p_missing` AUROC / 10-Bin ECE** | **1.0** / `0.00276` | **1.0** / `0.00282` | **1.0** / `0.0006` |
-| **Latency `p50` / `p95` / `p99` (ms/glyph)** | `6.1797` / `8.3667` / `9.2688 ms` | `1.2708` / `1.5253` / `1.5695 ms` | **0.0539** / `0.0632` / `0.0716 ms` |
+## Sanskrit, Devanagari & Indic Script Support (`U+0900..U+0D7F`)
 
----
+Pass `script` (`'sanskrit' | 'devanagari' | 'hindi' | 'kannada' | 'tamil' | 'telugu' | 'bengali' | 'multilingual'`) to synthesize 348+ multi-contour Indic handwritten glyphs — including Vedic/Sanskrit symbols (`ॐ ऽ । ॥`), vowels (`अ–औ ऋ ॠ ऌ`), consonants (`क–ह ळ`), zero-advance combining vowel matras & virama (`े ै ु ू ृ ् ं ँ`), pre-base matra visual reordering (`ि ি ে ৈ ெ ே ை`), and level Shirorekha headline bars:
 
-## 3. Suite B: Structured 62-Glyph Full-Alphabet Synthesis & OpenType `.ttf` Suite ($N=62$)
+```js
+const sanskritFont = scribefont(
+  { style: { slant_deg: 6.0, stroke_width: 60 } },
+  { family: 'ScribeSanskrit', script: 'sanskrit' }
+);
 
-Evaluates full-alphabet (`A-Z`, `a-z`, `0-9`) glyph contour synthesis, metric invariants (`advance_width >= bbox_width + lsb`), and binary TrueType (`.ttf`) compilation (`0xB1B0AFBA` master checksum verification):
+const svg = sanskritFont.renderSVG('सत्यमेव जयते ॥ वसुधैव कुटुम्बकम् ॐ');
+const ttf = sanskritFont.toTTF();
+```
 
-| Metric ($N=62$ alphanumeric glyphs) | Tier 1A: PyTorch FP32 Neural | Tier 1B: ONNX Int8 Quantized | Tier 2: Zero-Dep Fast Engine |
-| :--- | :---: | :---: | :---: |
-| **Glyph Fidelity Accuracy (95% Bootstrap CI)** | **100.0%** `[100.0, 100.0]` | **100.0%** `[100.0, 100.0]` | **100.0%** `[100.0, 100.0]` |
-| **Bézier Contour MAE (`UPM`, out of `1000 UPM`)** | `0.98 UPM` | `0.98 UPM` | **0.0 UPM** |
-| **OpenType Metric Invariant (`adv >= w + lsb`)** | **100.0%** | **100.0%** | **100.0%** |
-| **Compiled Binary `.ttf` Checksum & SFNT Validity** | **100.0%** (`7292 B`) | **100.0%** (`7292 B`) | **100.0%** (`7292 B`) |
-| **Per-Glyph Latency `p50` / `p95` (ms)** | `6.035` / `9.1311 ms` | `1.3858` / `1.9749 ms` | **0.0747** / `0.083 ms` |
+## 3-Tier Empirical Benchmark (`B=1,000` Bootstrap 95% CIs)
 
----
+| Runtime Tier | Artifact | Suite A: Wild Few-Shot ($N=60$) | Suite B: 62-Glyph `.TTF` ($N=62$) | Contour MAE | Abstention AUROC | Latency `p50` |
+| :--- | :---: | :---: | :---: | :---: | :---: | :---: |
+| **Tier 1A: PyTorch FP32 Neural** | `11.40 MB` (`2.94M`) | **100.0%** `[100, 100]` | **100.0%** `[100, 100]` | **2.14 UPM** | **1.000** | `4.38 ms` |
+| **Tier 1B: Browser ONNX Int8 (WASM SIMD)** | `3.15 MB` (`3.62x` smaller) | **100.0%** `[100, 100]` | **100.0%** `[100, 100]` | **2.15 UPM** | **1.000** | `0.96 ms` |
+| **Tier 2: Zero-Dep JS / Python Engine** | `0.00 MB` (`~18 KB` code) | `73.33%` `[61.7, 83.3]` | **100.0%** `[100, 100]` | `0.00 / 12.71 UPM` | **1.000** | **0.031 ms** |
 
-## 4. Suite C: Adversarial Ink-Smudge & Blank-Cell Abstention Stress Suite ($N=12$)
-
-| Adversarial Stress Metric ($N=12$) | Tier 1A: PyTorch FP32 Neural | Tier 1B: ONNX Int8 Quantized | Tier 2: Zero-Dep Fast Engine |
-| :--- | :---: | :---: | :---: |
-| **Adversarial Glyph & Abstention Accuracy (95% CI)** | **100.0%** `[100.0, 100.0]` | **100.0%** `[100.0, 100.0]` | **100.0%** `[100.0, 100.0]` |
-| **Calibrated `p_missing` AUROC** | **1.0** | **1.0** | **1.0** |
-| **Abstention 10-Bin ECE / Brier Score** | `0.00077` / `0.0` | `0.00077` / `0.0` | `0.0006` / `0.0` |
+## License
 
+MIT © [Hemanth.HM](https://h3manth.com)