Update README.md

README.md

@@ -24,7 +24,7 @@ model-index:
     metrics:
        - name: Test WER
          type: wer
-         value: 52.07
+         value: 33.10
 ---
 
 # Wav2Vec2-Large-XLSR-53-Dhivehi
@@ -52,15 +52,15 @@ resampler = torchaudio.transforms.Resample(48_000, 16_000)
 # Preprocessing the datasets.
 # We need to read the aduio files as arrays
 def speech_file_to_array_fn(batch):
-\tspeech_array, sampling_rate = torchaudio.load(batch["path"])
-\tbatch["speech"] = resampler(speech_array).squeeze().numpy()
-\treturn batch
+	speech_array, sampling_rate = torchaudio.load(batch["path"])
+	batch["speech"] = resampler(speech_array).squeeze().numpy()
+	return batch
 
 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)
 
 with torch.no_grad():
-\tlogits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
+	logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
 
 predicted_ids = torch.argmax(logits, dim=-1)
 
@@ -88,37 +88,51 @@ processor = Wav2Vec2Processor.from_pretrained("shahukareem/wav2vec2-large-xlsr-5
 model = Wav2Vec2ForCTC.from_pretrained("shahukareem/wav2vec2-large-xlsr-53-dhivehi")
 model.to("cuda")
 
-chars_to_ignore_regex = '[\\,\\?\\.\\!\\-\\;\\:\\"\\“\\%\\‘\\”\\�\\،\\.\\؟\\!\\'\\"\\–\\’]'
+chars_to_ignore_regex = '[\\\\,\\\\?\\\\.\\\\!\\\\-\\\\;\\\\:\\\\"\\\\“\\\\%\\\\‘\\\\”\\\\�\\\\،\\\\.\\\\؟\\\\!\\\\'\\\\"\\\\–\\\\’]'
 resampler = torchaudio.transforms.Resample(48_000, 16_000)
 
 # Preprocessing the datasets.
 # We need to read the aduio files as arrays
 def speech_file_to_array_fn(batch):
-\tbatch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
-\tspeech_array, sampling_rate = torchaudio.load(batch["path"])
-\tbatch["speech"] = resampler(speech_array).squeeze().numpy()
-\treturn batch
+	batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
+	speech_array, sampling_rate = torchaudio.load(batch["path"])
+	batch["speech"] = resampler(speech_array).squeeze().numpy()
+	return batch
 
 test_dataset = test_dataset.map(speech_file_to_array_fn)
 
 # Preprocessing the datasets.
 # We need to read the aduio files as arrays
 def evaluate(batch):
-\tinputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
+	inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
 
-\twith torch.no_grad():
-\t\tlogits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits
+	with torch.no_grad():
+		logits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits
 
-\tpred_ids = torch.argmax(logits, dim=-1)
-\tbatch["pred_strings"] = processor.batch_decode(pred_ids)
-\treturn batch
+	pred_ids = torch.argmax(logits, dim=-1)
+	batch["pred_strings"] = processor.batch_decode(pred_ids)
+	return 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**: 52.07 % 
+**Test Result**: 33.10% 
 
 ## Training
-The Common Voice `train` and `validation` datasets were used for training.
+The Common Voice `train` and `validation` datasets were used for training.
+
+## Example predictions
+reference: ކަރަންޓް ވައިރުކޮށް ބޮކި ހަރުކުރުން
+predicted: ކަރަންޓް ވައިރުކޮށް ބޮކި ހަރުކުރުން
+--
+reference: ދެން އެކުދިންނާ ދިމާއަށް އަތް ދިށްކޮށްލެވެ
+predicted: ދެން އެކުދިންނާ ދިމާއަށް އަތް ދިއްކޮށްލެވެ ް
+--
+reference: ރަކި ހިނިތުންވުމަކާއެކު އޭނާ އަމިއްލައަށް ތައާރަފްވި
+predicted: ރަކި ހިނިތުންވުމަކާއެކު އޭނާ އަމިއްލައަށް ތައަރަފްވި
+--
+reference: ކޮޓަރީގެ ކުޑަދޮރުން ބޭރު ބަލަހައްޓައިގެން އިން ރޫނާގެ މޫނުމަތިން ފާޅުވަމުން ދިޔައީ ކަންބޮޑުވުމުގެ އަސަރުތައް
+predicted: ކޮޓަރީގެ ކުޑަދޮރުން ބޭރު ބަލަހައްޓައިގެން އިން ރނާގެ މޫނުމަތިން ފާޅުވަމުން ދިޔައީ ކަންބޮޑުވުމުގެ އަސަރުތައް
+--