Add model files

config.json

@@ -0,0 +1,76 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
+  "activation_dropout": 0,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0,
+  "bos_token_id": 1,
+  "conv_bias": true,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.0,
+  "final_dropout": 0.0,
+  "gradient_checkpointing": true,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "mask_feature_length": 10,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "num_attention_heads": 16,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "pad_token_id": 44,
+  "transformers_version": "4.5.0.dev0",
+  "vocab_size": 45
+}

preprocessor_config.json

@@ -0,0 +1,8 @@
+{
+  "do_normalize": true,
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad8b539264b74e769a7115cdc8830ee7024abbfc055befa33747a8a13115ef3a
+size 1262118359

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|"}

utils.py

@@ -0,0 +1,42 @@
+import datasets
+import torchaudio
+import re
+
+
+def get_test_dataset(data_path='ISSAI_KSC_335RS_v1.1'):
+    def read_sentence(idx):
+        with open(f"{data_path}/Transcriptions/{idx}.txt", 'r') as f:
+            text = ' '.join(f.readlines())
+        return text
+
+    def read_text(batch):
+        batch["sentence"] = read_sentence(batch['uttID'])
+        return batch
+
+    chars_to_ignore = ["f", "m"]
+    chars_to_ignore_regex = f'[{"".join(chars_to_ignore)}]'
+
+    def process_text(batch):
+        batch["text"] = re.sub(chars_to_ignore_regex, "", batch["sentence"]).lower() + " "
+        batch['text'] = batch['text'].replace('a', 'а').replace('ə', 'ә').replace('ɵ', 'ө')
+        return batch
+
+    def load_audio(batch):
+        path = f"{data_path}/Audios_flac/{batch['uttID']}.flac"
+        speech_array, sr = torchaudio.load(path)
+        batch["speech"] = speech_array
+        batch["sampling_rate"] = sr
+        return batch
+        
+    test_dataset = datasets.load_dataset(
+        'csv', 
+        data_files=f"{data_path}/Meta/test.csv", 
+        delimiter=' ',
+        split='train'
+    )
+
+    test_dataset = test_dataset.map(read_text)
+    test_dataset = test_dataset.map(process_text)
+    test_dataset = test_dataset.map(load_audio, num_proc=1)
+
+    return test_dataset

vocab.json

@@ -0,0 +1 @@
+{"и": 0, "й": 1, "у": 2, "ь": 3, "ә": 4, "п": 5, "ы": 6, "щ": 7, "ч": 8, "е": 9, "р": 10, "з": 11, "ү": 12, "қ": 13, "х": 14, "г": 15, "б": 16, "я": 17, "с": 18, "ш": 19, "ө": 20, "ғ": 21, "т": 22, "і": 23, "м": 24, "ц": 25, "һ": 26, "ң": 27, "ж": 28, "ю": 29, "в": 30, "а": 32, "д": 33, "ф": 34, "э": 35, "ъ": 36, "л": 37, "ё": 38, "н": 39, "к": 40, "о": 41, "ұ": 42, "|": 31, "[UNK]": 43, "[PAD]": 44}