Fix the "'" character

README.md

@@ -30,6 +30,7 @@ import torch
 import torchaudio
 from datasets import load_dataset
 from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
+
 lang = "br"
 test_dataset = load_dataset("common_voice", lang, split="test[:2%]") 
 
@@ -38,15 +39,23 @@ model = Wav2Vec2ForCTC.from_pretrained("Marxav/wav2vec2-large-xlsr-53-breton")
 
 resampler = torchaudio.transforms.Resample(48_000, 16_000)
 
+chars_to_ignore_regex = '[\\\\,\\,\\?\\.\\!\\;\\:\\"\\“\\%\\”\\�\\(\\)\\/\\«\\»\\½\\…]'
+
 # Preprocessing the datasets.
-# We need to read the aduio files as arrays
+# We need to read the audio files as arrays
 def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = torchaudio.load(batch["path"])
    batch["speech"] = resampler(speech_array).squeeze().numpy()
+   batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower() + " "
+   batch["sentence"] = re.sub("ʼ", "'", batch["sentence"])
+   batch["sentence"] = re.sub("’", "'", batch["sentence"])
+   batch["sentence"] = re.sub('‘', "'", batch["sentence"])
    return batch
 
+nb_samples = 2
 test_dataset = test_dataset.map(speech_file_to_array_fn)
-inputs = processor(test_dataset["speech"][:2], sampling_rate=16_000, return_tensors="pt", padding=True)
+
+inputs = processor(test_dataset["speech"][:nb_samples], sampling_rate=16_000, return_tensors="pt", padding=True)
 
 with torch.no_grad():
    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
@@ -54,11 +63,11 @@ with torch.no_grad():
 predicted_ids = torch.argmax(logits, dim=-1)
 
 print("Prediction:", processor.batch_decode(predicted_ids))
-print("Reference:", test_dataset["sentence"][:2])
+print("Reference:", test_dataset["sentence"][:nb_samples])
 ```
 The above code leads to the following prediction for the first two samples:
 * Prediction: ["nel ler ket dont abenn eus netra la vez ser mirc'hid evel sij", 'an eil hag egile']
-* Reference: ['"N\\'haller ket dont a-benn eus netra pa vezer nec\\'het evel-se."', 'An eil hag egile.']
+* Reference: ['"N\\\\'haller ket dont a-benn eus netra pa vezer nec\\\\'het evel-se."', 'An eil hag egile.']
 
 The model can be evaluated as follows on the {language} test data of Common Voice.
 ```python
@@ -68,22 +77,15 @@ from datasets import load_dataset, load_metric
 from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
 import re
 
-test_dataset = load_dataset("common_voice", "br", split="test")
+lang = 'br'
+test_dataset = load_dataset("common_voice", lang, split="test")
 wer = load_metric("wer")
 
-processor = Wav2Vec2Processor.from_pretrained('Marxav/wav2vec2-large-xlsr-53-breton')
-model = Wav2Vec2ForCTC.from_pretrained('Marxav/wav2vec2-large-xlsr-53-breton')
+processor = Wav2Vec2Processor.from_pretrained('Marxav/wav2vec2-large-xlsr-53-breton2')
+model = Wav2Vec2ForCTC.from_pretrained('Marxav/wav2vec2-large-xlsr-53-breton2')
 model.to("cuda")
 
-
-chars_to_ignore_regex = """[\\\\,\\\\?\\\\.\\\\!\\\\-\\\\;\\\\:\\\\"\\\\“\\\\%\\\\‘\\\\”\\\\�\\\\'\\\\(\\\\)\\\\/\\\\«\\\\»\\\\½\\\\…]"""
-
-def remove_special_characters(batch):
-     sentence = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower() + " "
-     sentence = re.sub("ʼ","'", sentence)
-     sentence = re.sub("’","'", sentence)
-     batch["sentence"] = sentence
-     return batch
+chars_to_ignore_regex = '[\\\\,\\,\\?\\.\\!\\;\\:\\"\\“\\%\\”\\�\\(\\)\\/\\«\\»\\½\\…]'
 
 resampler = torchaudio.transforms.Resample(48_000, 16_000)
 
@@ -91,6 +93,10 @@ resampler = torchaudio.transforms.Resample(48_000, 16_000)
 # We need to read the aduio files as arrays
 def speech_file_to_array_fn(batch):
   batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
+  batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower() + " "
+  batch["sentence"] = re.sub("ʼ", "'", batch["sentence"])
+  batch["sentence"] = re.sub("’", "'", batch["sentence"])
+  batch["sentence"] = re.sub('‘', "'", batch["sentence"])
   speech_array, sampling_rate = torchaudio.load(batch["path"])
   batch["speech"] = resampler(speech_array).squeeze().numpy()
   return batch
@@ -100,7 +106,7 @@ test_dataset = test_dataset.map(remove_special_characters)
 test_dataset = test_dataset.map(speech_file_to_array_fn)
 
 # Preprocessing the datasets.
-# We need to read the aduio files as arrays
+# We need to read the audio files as arrays
 def evaluate(batch):
   inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
 
@@ -113,9 +119,4 @@ def evaluate(batch):
 
 result = test_dataset.map(evaluate, batched=True, batch_size=8)
 
-print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))
-```
-
-**Test Result**: 44.34%
-## Training
-The Common Voice `train`, `validation` datasets were used for training.
+print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))