update

examples/wenet/toolbox_infer.py

@@ -42,7 +42,7 @@ def main():
     args = get_args()
 
     # audio convert
-    signal, sample_rate = librosa.load(args.in_filename, sr=16000)
+    signal, sample_rate = librosa.load(args.in_filename, sr=args.sample_rate)
     signal *= 32768.0
     signal = np.array(signal, dtype=np.int16)
 
@@ -95,19 +95,19 @@ def main():
     )
     recognizer = sherpa.OfflineRecognizer(config)
 
-    s = recognizer.create_stream()
-    s.accept_wave_file(
-        temp_file.as_posix()
-    )
-    recognizer.decode_stream(s)
-    text = s.result.text.strip()
-    text = text.lower()
-    print("text: {}".format(text))
-
-    # text = decode.decode_by_recognizer(recognizer=recognizer,
-    #                                    filename=temp_file.as_posix(),
-    #                                    )
+    # s = recognizer.create_stream()
+    # s.accept_wave_file(
+    #     temp_file.as_posix()
+    # )
+    # recognizer.decode_stream(s)
+    # text = s.result.text.strip()
+    # text = text.lower()
     # print("text: {}".format(text))
+
+    text = decode.decode_by_recognizer(recognizer=recognizer,
+                                       filename=temp_file.as_posix(),
+                                       )
+    print("text: {}".format(text))
     return