Chaka-ASR public release

Co-authored-by: Abrham Belete <Abrham Belete@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

NOTICE.md

@@ -0,0 +1,7 @@
+# Upstream credits
+
+Chaka-ASR is fine-tuned from badrex/Ethio-ASR-amharic, which declares CC BY 4.0, and builds on facebook/w2v-bert-2.0.
+
+Adaptation used Leyu Amharic speech datasets with WAXAL replay. Consult upstream model cards and dataset terms for attribution and applicable conditions.
+
+This notice provides attribution and does not replace a LICENSE file or grant a new license.

README.md

@@ -0,0 +1,116 @@
+---
+language:
+  - am
+library_name: transformers
+pipeline_tag: automatic-speech-recognition
+base_model: badrex/Ethio-ASR-amharic
+base_model_relation: finetune
+tags:
+  - amharic
+  - speech-recognition
+  - wav2vec2-bert
+  - ctc
+  - leyu
+---
+
+# Chaka-ASR
+
+**Amharic automatic speech recognition**
+
+Chaka-ASR converts Amharic speech into text. The model is fine-tuned from [badrex/Ethio-ASR-amharic](https://huggingface.co/badrex/Ethio-ASR-amharic) using Leyu Amharic speech data, with WAXAL replay during training.
+
+It is intended for transcription, voice interfaces, and speech applications in Ethiopian contexts, including environments where background noise may be present.
+
+## Model details
+
+| Property | Details |
+|---|---|
+| Model ID | `abrilo/Chaka-ASR` |
+| Maintainer | [abrilo](https://huggingface.co/abrilo) |
+| Language | Amharic |
+| Task | Automatic speech recognition |
+| Architecture | Wav2Vec2-BERT with a CTC output head |
+| Parameters | Approximately 606 million |
+| Starting checkpoint | `badrex/Ethio-ASR-amharic` |
+| Input | Mono audio at 16,000 Hz |
+| Output | Amharic text |
+| Weights | Safetensors, float32 |
+
+## Training overview
+
+Chaka-ASR was adapted using Leyu Amharic speech datasets. Training also included sampled WAXAL examples as replay data. The training summary records 8,900 training steps and speaker-separated Leyu training, validation, and test splits.
+
+The starting checkpoint is an Amharic model from the Ethio-ASR project, based on `facebook/w2v-bert-2.0`. Chaka-ASR builds on that work; it is not trained from scratch.
+
+## Evaluation
+
+| Dataset | WER (%) ??? | CER (%) ??? |
+|---|---:|---:|
+| Leyu | 20.48 | 6.02 |
+
+The reported Leyu result was measured on its Addis Ababa dialect test subset. It is a subset result rather than an aggregate score for the complete Leyu dataset. WER is word error rate and CER is character error rate; lower values indicate fewer errors.
+
+These values come from the recorded evaluation of the released checkpoint. They are not a direct comparison against the starting model. The original evaluation's text normalization and decoding settings should be consulted before reproducing or comparing these scores.
+
+## Quick start
+
+Install the example dependencies in a separate Python environment:
+
+```bash
+python -m pip install -r requirements-example.txt
+```
+
+Run the included example on a WAV or FLAC recording:
+
+```bash
+python examples/transcribe.py audio.wav
+```
+
+The example converts stereo audio to mono, resamples it to 16 kHz, and uses greedy CTC decoding. It uses an NVIDIA GPU when CUDA is available and otherwise runs on CPU. Audio is processed as a complete clip; this example does not implement live streaming or long-recording segmentation.
+
+The checkpoint configuration records Transformers `5.16.1`. The dependency file uses that version as its minimum; the example must still be checked in your target runtime before production use. Install a PyTorch build appropriate for your hardware if GPU inference is required.
+
+## Intended uses
+
+- Amharic audio transcription.
+- Voice assistants and conversational applications.
+- Speech interfaces for customer service and accessibility.
+- Research and evaluation of Amharic speech recognition.
+
+## Background noise
+
+Chaka-ASR is intended for applications that may encounter background noise. Actual performance depends on microphone placement, speech clarity, reverberation, and the level and type of noise. The published Leyu result does not establish a separate noise robustness benchmark. Evaluate the model on representative recordings from your deployment environment.
+
+## Limitations
+
+Recognition quality may vary with dialect, recording conditions, speaking style, and overlapping speech. Names, numbers, abbreviations, and mixed-language speech may require additional review. The model produces transcriptions; it does not verify identity, amounts, or factual content.
+
+The reported error rates describe the evaluated subset and should not be treated as guaranteed performance for all Amharic recordings.
+
+## Attribution
+
+Chaka-ASR acknowledges:
+
+- [Ethio-ASR](https://huggingface.co/collections/badrex/ethio-asr), including the Amharic starting checkpoint.
+- [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0), the upstream architecture and pretrained model.
+- Leyu Amharic speech datasets used for adaptation.
+- [WAXAL](https://huggingface.co/datasets/google/WaxalNLP), used as replay data during adaptation.
+
+## License
+
+A release license for Chaka-ASR has not yet been declared in this repository. The upstream Ethio-ASR Amharic model declares CC BY 4.0. Consult the applicable upstream model and dataset terms; public availability alone is not a license grant.
+
+## Feedback
+
+Report transcription issues and technical questions in the [Community tab](https://huggingface.co/abrilo/Chaka-ASR/discussions). Include the model revision and runtime versions. Share recordings only when you have permission and remove personal or confidential information.
+
+## Citation
+
+```bibtex
+@misc{chaka_asr_2026,
+  author = {abrilo},
+  title = {Chaka-ASR: Amharic Automatic Speech Recognition},
+  year = {2026},
+  url = {https://huggingface.co/abrilo/Chaka-ASR}
+}
+```

added_tokens.json

@@ -0,0 +1,4 @@
+{
+  "</s>": 410,
+  "<s>": 409
+}

config.json

@@ -0,0 +1,82 @@
+{
+  "activation_dropout": 0.0,
+  "adapter_act": "relu",
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": true,
+  "apply_spec_augment": false,
+  "architectures": [
+    "Wav2Vec2BertForCTC"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "classifier_proj_size": 768,
+  "codevector_dim": 768,
+  "conformer_conv_dropout": 0.1,
+  "contrastive_logits_temperature": 0.1,
+  "conv_depthwise_kernel_size": 31,
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": true,
+  "diversity_loss_weight": 0.1,
+  "dtype": "float32",
+  "eos_token_id": 2,
+  "feat_proj_dropout": 0.0,
+  "feat_quantizer_dropout": 0.0,
+  "feature_projection_input_dim": 160,
+  "final_dropout": 0.1,
+  "hidden_act": "swish",
+  "hidden_dropout": 0.0,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.0,
+  "left_max_position_embeddings": 64,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.0,
+  "max_source_positions": 5000,
+  "model_type": "wav2vec2-bert",
+  "num_adapter_layers": 1,
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "output_hidden_size": 1024,
+  "pad_token_id": 408,
+  "position_embeddings_type": "relative_key",
+  "proj_codevector_dim": 768,
+  "right_max_position_embeddings": 8,
+  "rotary_embedding_base": 10000,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "transformers_version": "5.16.1",
+  "use_cache": false,
+  "use_intermediate_ffn_before_adapter": false,
+  "use_weighted_layer_sum": false,
+  "vocab_size": 411,
+  "xvector_output_dim": 512
+}

examples/transcribe.py

@@ -0,0 +1,38 @@
+"""Transcribe a short WAV or FLAC recording with Chaka-ASR."""
+import argparse
+from math import gcd
+import numpy as np
+import soundfile as sf
+import torch
+from scipy.signal import resample_poly
+from transformers import AutoProcessor, AutoModelForCTC
+
+
+def main():
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("audio", help="Path to a WAV or FLAC recording")
+    parser.add_argument("--revision", default="main", help="Model commit or branch")
+    args = parser.parse_args()
+    audio, sample_rate = sf.read(args.audio, dtype="float32", always_2d=True)
+    audio = audio.mean(axis=1)
+    if audio.size == 0:
+        raise ValueError("Audio recording is empty")
+    if not np.isfinite(audio).all():
+        raise ValueError("Audio contains non-finite samples")
+    processor = AutoProcessor.from_pretrained("abrilo/Chaka-ASR", revision=args.revision)
+    model = AutoModelForCTC.from_pretrained("abrilo/Chaka-ASR", revision=args.revision)
+    target_rate = processor.feature_extractor.sampling_rate
+    if sample_rate != target_rate:
+        divisor = gcd(sample_rate, target_rate)
+        audio = resample_poly(audio, target_rate // divisor, sample_rate // divisor).astype("float32")
+    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    model.to(device).eval()
+    inputs = processor(audio, sampling_rate=target_rate, return_tensors="pt")
+    inputs = {key: value.to(device) for key, value in inputs.items()}
+    with torch.inference_mode():
+        predicted = model(**inputs).logits.argmax(dim=-1)
+    print(processor.batch_decode(predicted.cpu())[0])
+
+
+if __name__ == "__main__":
+    main()

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5a31340cab11ae2cc77c99e417b9e77fa25a5b6d152fc4bd740ae2a908eefdf
+size 2424499668

processor_config.json

@@ -0,0 +1,13 @@
+{
+  "feature_extractor": {
+    "feature_extractor_type": "SeamlessM4TFeatureExtractor",
+    "feature_size": 80,
+    "num_mel_bins": 80,
+    "padding_side": "right",
+    "padding_value": 1,
+    "return_attention_mask": true,
+    "sampling_rate": 16000,
+    "stride": 2
+  },
+  "processor_class": "Wav2Vec2BertProcessor"
+}

requirements-example.txt

@@ -0,0 +1,7 @@
+# Example dependencies; install a hardware-compatible PyTorch build for CUDA.
+transformers>=5.16.1,<6
+torch>=2.6
+safetensors>=0.4
+numpy>=1.26
+scipy>=1.11
+soundfile>=0.12

tokenizer_config.json

@@ -0,0 +1,62 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "|",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "407": {
+      "content": "[UNK]",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "408": {
+      "content": "[PAD]",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "409": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "410": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "backend": "custom",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": false,
+  "eos_token": "</s>",
+  "is_local": false,
+  "local_files_only": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "word_delimiter_token": "|"
+  },
+  "pad_token": "[PAD]",
+  "processor_class": "Wav2Vec2BertProcessor",
+  "replace_word_delimiter_char": " ",
+  "target_lang": null,
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "unk_token": "[UNK]",
+  "word_delimiter_token": "|"
+}

vocab.json

@@ -0,0 +1,411 @@
+{
+  "!": 1,
+  "#": 2,
+  "$": 3,
+  "%": 4,
+  "&": 5,
+  "'": 6,
+  "*": 7,
+  "+": 8,
+  ",": 9,
+  "-": 10,
+  ".": 11,
+  "0": 12,
+  "1": 13,
+  "2": 14,
+  "3": 15,
+  "4": 16,
+  "5": 17,
+  "6": 18,
+  "7": 19,
+  "8": 20,
+  "9": 21,
+  "=": 22,
+  "?": 23,
+  "@": 24,
+  "[PAD]": 408,
+  "[UNK]": 407,
+  "a": 25,
+  "b": 26,
+  "c": 27,
+  "d": 28,
+  "e": 29,
+  "f": 30,
+  "g": 31,
+  "h": 32,
+  "i": 33,
+  "j": 34,
+  "k": 35,
+  "l": 36,
+  "m": 37,
+  "n": 38,
+  "o": 39,
+  "p": 40,
+  "q": 41,
+  "r": 42,
+  "s": 43,
+  "t": 44,
+  "u": 45,
+  "v": 46,
+  "w": 47,
+  "x": 48,
+  "y": 49,
+  "z": 50,
+  "|": 0,
+  "ሀ": 51,
+  "ሁ": 52,
+  "ሂ": 53,
+  "ሃ": 54,
+  "ሄ": 55,
+  "ህ": 56,
+  "ሆ": 57,
+  "ሇ": 58,
+  "ለ": 59,
+  "ሉ": 60,
+  "ሊ": 61,
+  "ላ": 62,
+  "ሌ": 63,
+  "ል": 64,
+  "ሎ": 65,
+  "ሏ": 66,
+  "ሐ": 67,
+  "ሑ": 68,
+  "ሒ": 69,
+  "ሓ": 70,
+  "ሔ": 71,
+  "ሕ": 72,
+  "ሖ": 73,
+  "ሗ": 74,
+  "መ": 75,
+  "ሙ": 76,
+  "ሚ": 77,
+  "ማ": 78,
+  "ሜ": 79,
+  "ም": 80,
+  "ሞ": 81,
+  "ሟ": 82,
+  "ሠ": 83,
+  "ሡ": 84,
+  "ሢ": 85,
+  "ሣ": 86,
+  "ሤ": 87,
+  "ሥ": 88,
+  "ሦ": 89,
+  "ሧ": 90,
+  "ረ": 91,
+  "ሩ": 92,
+  "ሪ": 93,
+  "ራ": 94,
+  "ሬ": 95,
+  "ር": 96,
+  "ሮ": 97,
+  "ሯ": 98,
+  "ሰ": 99,
+  "ሱ": 100,
+  "ሲ": 101,
+  "ሳ": 102,
+  "ሴ": 103,
+  "ስ": 104,
+  "ሶ": 105,
+  "ሷ": 106,
+  "ሸ": 107,
+  "ሹ": 108,
+  "ሺ": 109,
+  "ሻ": 110,
+  "ሼ": 111,
+  "ሽ": 112,
+  "ሾ": 113,
+  "ሿ": 114,
+  "ቀ": 115,
+  "ቁ": 116,
+  "ቂ": 117,
+  "ቃ": 118,
+  "ቄ": 119,
+  "ቅ": 120,
+  "ቆ": 121,
+  "ቇ": 122,
+  "ቈ": 123,
+  "ቊ": 124,
+  "ቋ": 125,
+  "ቌ": 126,
+  "ቍ": 127,
+  "ቐ": 128,
+  "ቑ": 129,
+  "ቒ": 130,
+  "ቓ": 131,
+  "ቔ": 132,
+  "ቕ": 133,
+  "ቖ": 134,
+  "ቘ": 135,
+  "ቚ": 136,
+  "ቛ": 137,
+  "ቜ": 138,
+  "ቝ": 139,
+  "በ": 140,
+  "ቡ": 141,
+  "ቢ": 142,
+  "ባ": 143,
+  "ቤ": 144,
+  "ብ": 145,
+  "ቦ": 146,
+  "ቧ": 147,
+  "ቨ": 148,
+  "ቩ": 149,
+  "ቪ": 150,
+  "ቫ": 151,
+  "ቬ": 152,
+  "ቭ": 153,
+  "ቮ": 154,
+  "ቯ": 155,
+  "ተ": 156,
+  "ቱ": 157,
+  "ቲ": 158,
+  "ታ": 159,
+  "ቴ": 160,
+  "ት": 161,
+  "ቶ": 162,
+  "ቷ": 163,
+  "ቸ": 164,
+  "ቹ": 165,
+  "ቺ": 166,
+  "ቻ": 167,
+  "ቼ": 168,
+  "ች": 169,
+  "ቾ": 170,
+  "ቿ": 171,
+  "ኀ": 172,
+  "ኁ": 173,
+  "ኂ": 174,
+  "ኃ": 175,
+  "ኄ": 176,
+  "ኅ": 177,
+  "ኆ": 178,
+  "ኇ": 179,
+  "ኈ": 180,
+  "ኊ": 181,
+  "ኋ": 182,
+  "ኌ": 183,
+  "ኍ": 184,
+  "ነ": 185,
+  "ኑ": 186,
+  "ኒ": 187,
+  "ና": 188,
+  "ኔ": 189,
+  "ን": 190,
+  "ኖ": 191,
+  "ኗ": 192,
+  "ኘ": 193,
+  "ኙ": 194,
+  "ኚ": 195,
+  "ኛ": 196,
+  "ኜ": 197,
+  "ኝ": 198,
+  "ኞ": 199,
+  "ኟ": 200,
+  "አ": 201,
+  "ኡ": 202,
+  "ኢ": 203,
+  "ኣ": 204,
+  "ኤ": 205,
+  "እ": 206,
+  "ኦ": 207,
+  "ኧ": 208,
+  "ከ": 209,
+  "ኩ": 210,
+  "ኪ": 211,
+  "ካ": 212,
+  "ኬ": 213,
+  "ክ": 214,
+  "ኮ": 215,
+  "ኯ": 216,
+  "ኰ": 217,
+  "ኲ": 218,
+  "ኳ": 219,
+  "ኴ": 220,
+  "ኵ": 221,
+  "ኸ": 222,
+  "ኹ": 223,
+  "ኺ": 224,
+  "ኻ": 225,
+  "ኼ": 226,
+  "ኽ": 227,
+  "ኾ": 228,
+  "ዀ": 229,
+  "ዂ": 230,
+  "ዃ": 231,
+  "ዄ": 232,
+  "ዅ": 233,
+  "ወ": 234,
+  "ዉ": 235,
+  "ዊ": 236,
+  "ዋ": 237,
+  "ዌ": 238,
+  "ው": 239,
+  "ዎ": 240,
+  "ዏ": 241,
+  "ዐ": 242,
+  "ዑ": 243,
+  "ዒ": 244,
+  "ዓ": 245,
+  "ዔ": 246,
+  "ዕ": 247,
+  "ዖ": 248,
+  "ዘ": 249,
+  "ዙ": 250,
+  "ዚ": 251,
+  "ዛ": 252,
+  "ዜ": 253,
+  "ዝ": 254,
+  "ዞ": 255,
+  "ዟ": 256,
+  "ዠ": 257,
+  "ዡ": 258,
+  "ዢ": 259,
+  "ዣ": 260,
+  "ዤ": 261,
+  "ዥ": 262,
+  "ዦ": 263,
+  "ዧ": 264,
+  "የ": 265,
+  "ዩ": 266,
+  "ዪ": 267,
+  "ያ": 268,
+  "ዬ": 269,
+  "ይ": 270,
+  "ዮ": 271,
+  "ዯ": 272,
+  "ደ": 273,
+  "ዱ": 274,
+  "ዲ": 275,
+  "ዳ": 276,
+  "ዴ": 277,
+  "ድ": 278,
+  "ዶ": 279,
+  "ዷ": 280,
+  "ዸ": 281,
+  "ዹ": 282,
+  "ዺ": 283,
+  "ዻ": 284,
+  "ዼ": 285,
+  "ዽ": 286,
+  "ዾ": 287,
+  "ዿ": 288,
+  "ጀ": 289,
+  "ጁ": 290,
+  "ጂ": 291,
+  "ጃ": 292,
+  "ጄ": 293,
+  "ጅ": 294,
+  "ጆ": 295,
+  "ጇ": 296,
+  "ገ": 297,
+  "ጉ": 298,
+  "ጊ": 299,
+  "ጋ": 300,
+  "ጌ": 301,
+  "ግ": 302,
+  "ጎ": 303,
+  "ጏ": 304,
+  "ጐ": 305,
+  "ጒ": 306,
+  "ጓ": 307,
+  "ጔ": 308,
+  "ጕ": 309,
+  "ጘ": 310,
+  "ጙ": 311,
+  "ጚ": 312,
+  "ጛ": 313,
+  "ጜ": 314,
+  "ጝ": 315,
+  "ጞ": 316,
+  "ጟ": 317,
+  "ጠ": 318,
+  "ጡ": 319,
+  "ጢ": 320,
+  "ጣ": 321,
+  "ጤ": 322,
+  "ጥ": 323,
+  "ጦ": 324,
+  "ጧ": 325,
+  "ጨ": 326,
+  "ጩ": 327,
+  "ጪ": 328,
+  "ጫ": 329,
+  "ጬ": 330,
+  "ጭ": 331,
+  "ጮ": 332,
+  "ጯ": 333,
+  "ጰ": 334,
+  "ጱ": 335,
+  "ጲ": 336,
+  "ጳ": 337,
+  "ጴ": 338,
+  "ጵ": 339,
+  "ጶ": 340,
+  "ጷ": 341,
+  "ጸ": 342,
+  "ጹ": 343,
+  "ጺ": 344,
+  "ጻ": 345,
+  "ጼ": 346,
+  "ጽ": 347,
+  "ጾ": 348,
+  "ጿ": 349,
+  "ፀ": 350,
+  "ፁ": 351,
+  "ፂ": 352,
+  "ፃ": 353,
+  "ፄ": 354,
+  "ፅ": 355,
+  "ፆ": 356,
+  "ፇ": 357,
+  "ፈ": 358,
+  "ፉ": 359,
+  "ፊ": 360,
+  "ፋ": 361,
+  "ፌ": 362,
+  "ፍ": 363,
+  "ፎ": 364,
+  "ፏ": 365,
+  "ፐ": 366,
+  "ፑ": 367,
+  "ፒ": 368,
+  "ፓ": 369,
+  "ፔ": 370,
+  "ፕ": 371,
+  "ፖ": 372,
+  "ፗ": 373,
+  "ፘ": 374,
+  "ፙ": 375,
+  "ፚ": 376,
+  "፠": 377,
+  "፡": 378,
+  "።": 379,
+  "፣": 380,
+  "፤": 381,
+  "፥": 382,
+  "፦": 383,
+  "፧": 384,
+  "፨": 385,
+  "፩": 386,
+  "፪": 387,
+  "፫": 388,
+  "፬": 389,
+  "፭": 390,
+  "፮": 391,
+  "፯": 392,
+  "፰": 393,
+  "፱": 394,
+  "፲": 395,
+  "፳": 396,
+  "፴": 397,
+  "፵": 398,
+  "፶": 399,
+  "፷": 400,
+  "፸": 401,
+  "፹": 402,
+  "፺": 403,
+  "፻": 404,
+  "፼": 405,
+  "€": 406
+}