updated to wavio

gradio_asr_en_libri100_word_vs_bpe.py

@@ -9,8 +9,7 @@ from difflib import Differ
 from glob import glob
 from loguru import logger
 # import librosa
-from scipy.io import wavfile
-import scipy.signal as sps
+import wavio
 import gradio as gr
 
 from espnet2.bin.asr_inference import Speech2Text
@@ -46,7 +45,10 @@ def predict(wav_file):
     logger.info('wav file loaded')
     # Load audio    
     # speech, rate = librosa.load(wav_file, sr=16000)
-    rate, speech = wavfile.read(wav_file)
+    wav = wavio.read(wav_file)
+    speech = wav.data
+    rate = wav.rate
+    
     # number_of_samples = round(len(speech_raw) * float(16000) / rate)
     # speech = sps.resample(speech_raw, number_of_samples)