Update README.md

README.md

@@ -23,7 +23,7 @@ model-index:
     metrics:
        - name: Test WER
          type: wer
-         value: 30.23
+         value: 22.82
 ---
 
 # Wav2Vec2-Large-XLSR-53-lg
@@ -79,6 +79,7 @@ import torchaudio
 from datasets import load_dataset, load_metric
 from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
 import re
+import unidecode
 
 test_dataset = load_dataset("common_voice", "lg", split="test")
 wer = load_metric("wer")
@@ -87,21 +88,29 @@ processor = Wav2Vec2Processor.from_pretrained("lucio/wav2vec2-large-xlsr-luganda
 model = Wav2Vec2ForCTC.from_pretrained("lucio/wav2vec2-large-xlsr-luganda")
 model.to("cuda")
 
-chars_to_ignore_regex = '[,?.!;:"“-]'
+chars_to_ignore_regex = '[\[\],?.!;:%"“”(){}‟ˮʺ″«»/…‽�–]'
 resampler = torchaudio.transforms.Resample(48_000, 16_000)
 
 # Preprocessing the datasets.
 # We need to read the audio files as arrays
 def speech_file_to_array_fn(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
 
+def remove_special_characters(batch):
+    # word-internal apostrophes are marking contractions
+    batch["norm_text"] = re.sub(r'[‘’´`]', r"'", batch["sentence"])
+    # most other punctuation is ignored
+    batch["norm_text"] = re.sub(chars_to_ignore_regex, "", batch["norm_text"]).lower().strip()
+    batch["norm_text"] = re.sub(r"(-|' | '|  +)", " ", batch["norm_text"])
+    # remove accents from a few characters (from loanwords, not tones)
+    batch["norm_text"] = unidecode.unidecode(batch["norm_text"])
+    return batch
+
 test_dataset = test_dataset.map(speech_file_to_array_fn)
+test_dataset = test_dataset.map(remove_special_characters)
 
-# Preprocessing the datasets.
-# 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)
 
@@ -114,13 +123,13 @@ 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"])))
+print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["norm_text"])))
 ```
 
-**Test Result**: 30.23 % 
+**Test Result**: 22.82 % 
 
 ## Training
 
-The Common Voice `train`, `validation` and `other` datasets were used for training, with the additional filter applied to remove `other` data that did not have more up votes than down votes (train+validation+other[upvotes > downvotes]). 
+The Common Voice `train`, `validation` and `other` datasets were used for training, augmented to twice the original size with added noise and manipulated pitch, phase and intensity.
 
 The script used for training was just the `run_finetuning.py` script provided in OVHcloud's databuzzword/hf-wav2vec image.