Upload README.md with huggingface_hub

README.md

@@ -2,6 +2,7 @@
 license: mit
 language:
   - en
+  - de
 tags:
   - g2p
   - grapheme-to-phoneme
@@ -21,13 +22,15 @@ benchmarks. These are static test fixtures, not runtime model assets.
 | File | Language | Entries | Description |
 |------|----------|---------|--------------|
 | `g2p/en_us/test.tsv` | English (US) | 5000 | Held-out word-IPA pairs for CER benchmarking |
+| `g2p/de_de/test.tsv` | German | 5000 | Held-out word-IPA pairs for CER benchmarking |
 | `g2p/kokoro_vocab.json` | n/a | 114 | Kokoro phoneme vocabulary (char → token ID) used by the normalizer correctness benchmark |
 | `g2p/en_us/kokoro_normalizer_ref.tsv` | English (US) | 71 | Raw IPA → expected (`en_us`, Kokoro) normalized IPA + phoneme IDs, for `IpaNormalizer` regression testing |
+| `g2p/de_de/kokoro_normalizer_ref.tsv` | German | 32 | Raw IPA → expected (`de`, Kokoro) normalized IPA + phoneme IDs, for `IpaNormalizer` regression testing |
 
 ## Format
 
-`g2p/en_us/test.tsv`: tab-separated, two columns, UTF-8, no header row, sorted
-alphabetically:
+`g2p/en_us/test.tsv` and `g2p/de_de/test.tsv`: tab-separated, two columns,
+UTF-8, no header row, sorted alphabetically:
 
 ```
 word<TAB>IPA
@@ -39,8 +42,8 @@ Example:
 abandons	əbˈændənz
 ```
 
-`g2p/en_us/kokoro_normalizer_ref.tsv`: tab-separated, three columns, UTF-8, no
-header row:
+`g2p/en_us/kokoro_normalizer_ref.tsv` and `g2p/de_de/kokoro_normalizer_ref.tsv`:
+tab-separated, three columns, UTF-8, no header row:
 
 ```
 raw_ipa<TAB>expected_normalized_ipa<TAB>expected_phoneme_ids
@@ -66,6 +69,13 @@ Pronouncing Dictionary (https://github.com/cmusphinx/cmudict, BSD-style
 license). These entries are held out and never seen by the G2P engine
 during lexicon construction.
 
+`g2p/de_de/test.tsv` is a 5000-entry random sample (seed 42, no replacement,
+one pronunciation per word) from Moonshine Voice's German IPA lexicon
+(MIT-licensed multilingual IPA dictionary data), deduplicated to unique
+words before sampling since the source lexicon has multiple pronunciation
+rows for some words. These entries are held out and never seen by the G2P
+engine during lexicon construction.
+
 `g2p/kokoro_vocab.json` is copied verbatim from Moonshine Voice's
 `core/moonshine-tts/data/kokoro/config.json` `vocab` object (114 entries;
 the HuggingFace `onnx-community/Kokoro-82M-v1.0-ONNX` export used by Crane's
@@ -85,6 +95,22 @@ pairs and the vocab-filter/whitespace-collapse logic are simple enough that
 this port only needs `str.replace`, `str.strip`, and Python's
 `unicodedata.normalize("NFC", ...)`.
 
+`g2p/de_de/kokoro_normalizer_ref.tsv` combines 6 realistic German lexicon
+entries sampled from `g2p/de_de/test.tsv` (single words plus one 2-word
+combination, chosen to exercise both plain lexicon passthrough and the
+affricate/geminate/`ʏ` replacements together with dropped combining
+diacritics) with 26 hand-crafted edge cases (each
+`DE_EXTRA_KOKORO_REPLACEMENTS` pair in isolation, shared
+affricate/diphthong passthrough under `"de"`, German phonemes with no
+replacement needed, syllabic-consonant and non-syllabic-offglide diacritic
+dropping, unknown-codepoint dropping, and whitespace handling). Expected
+output was hand-derived by applying
+`SHARED_KOKORO_REPLACEMENTS`/`DE_EXTRA_KOKORO_REPLACEMENTS` and the
+vocab-filter/whitespace-collapse algorithm against `g2p/kokoro_vocab.json`
+directly, then cross-checked by running the real `IpaNormalizer` and
+confirming an exact match — not authored by trusting `normalize()`'s own
+output as ground truth.
+
 ## License
 
 Both upstream sources use permissive, MIT-compatible licenses: