app

util.py

@@ -1,10 +1,8 @@
 import random
 from umsc import UgMultiScriptConverter
-import torchaudio
 import string
 import epitran
 from difflib import SequenceMatcher
-import numpy as np
 
 # Lists of Uyghur short and long texts
 short_texts = [
@@ -36,23 +34,23 @@ def generate_long_text(script_choice):
     return text
 
 # ASR Utils
-def load_and_resample_audio(audio_data, target_rate):
-    """Load audio and resample based on target sample rate"""
-    if isinstance(audio_data, tuple):
-        # microphone
-        sampling_rate, audio_input = audio_data
-        audio_input = (audio_input / 32768.0).astype(np.float32)
-    elif isinstance(audio_data, str):
-        # file upload
-        audio_input, sampling_rate = torchaudio.load(audio_data)
-    else: 
-        return "<<ERROR: Invalid Audio Input Instance: {}>>".format(type(audio_data))
-    # Resample if needed
-    if sampling_rate != target_rate:
-        resampler = torchaudio.transforms.Resample(sampling_rate, target_rate)
-        audio_input = resampler(audio_input)
-
-    return audio_input, target_rate
+# def load_and_resample_audio(audio_data, target_rate):
+#     """Load audio and resample based on target sample rate"""
+#     if isinstance(audio_data, tuple):
+#         # microphone
+#         sampling_rate, audio_input = audio_data
+#         audio_input = (audio_input / 32768.0).astype(np.float32)
+#     elif isinstance(audio_data, str):
+#         # file upload
+#         audio_input, sampling_rate = torchaudio.load(audio_data)
+#     else: 
+#         return "<<ERROR: Invalid Audio Input Instance: {}>>".format(type(audio_data))
+#     # Resample if needed
+#     if sampling_rate != target_rate:
+#         resampler = torchaudio.transforms.Resample(sampling_rate, target_rate)
+#         audio_input = resampler(audio_input)
+
+#     return audio_input, target_rate
 
 def calculate_pronunciation_accuracy(reference_text, output_text, language_code='uig-Arab'):
     """