update

examples/wenet/toolbox_infer.py

@@ -42,25 +42,25 @@ def main():
     args = get_args()
 
     # audio convert
-    signal, sample_rate = librosa.load(args.in_filename, sr=args.sample_rate)
-    signal *= 32768.0
-    signal = np.array(signal, dtype=np.int16)
-
-    temp_file = temp_directory / "temp.wav"
-    wavfile.write(
-        temp_file.as_posix(),
-        rate=args.sample_rate,
-        data=signal
-    )
+    # signal, sample_rate = librosa.load(args.in_filename, sr=args.sample_rate)
+    # signal *= 32768.0
+    # signal = np.array(signal, dtype=np.int16)
+    #
+    # temp_file = temp_directory / "temp.wav"
+    # wavfile.write(
+    #     temp_file.as_posix(),
+    #     rate=args.sample_rate,
+    #     data=signal
+    # )
 
     # audio convert
-    # in_filename = Path(args.in_filename)
-    # out_filename = Path(tempfile.gettempdir()) / "asr" / in_filename.name
-    # out_filename.parent.mkdir(parents=True, exist_ok=True)
-    #
-    # audio_convert(in_filename=in_filename.as_posix(),
-    #               out_filename=out_filename.as_posix(),
-    #               )
+    in_filename = Path(args.in_filename)
+    out_filename = Path(tempfile.gettempdir()) / "asr" / in_filename.name
+    out_filename.parent.mkdir(parents=True, exist_ok=True)
+
+    audio_convert(in_filename=in_filename.as_posix(),
+                  out_filename=out_filename.as_posix(),
+                  )
 
     # load recognizer
     m_dict = models.model_map["Chinese"][0]
@@ -105,7 +105,7 @@ def main():
     # print("text: {}".format(text))
 
     text = decode.decode_by_recognizer(recognizer=recognizer,
-                                       filename=temp_file.as_posix(),
+                                       filename=out_filename.as_posix(),
                                        )
     print("text: {}".format(text))
     return

toolbox/k2_sherpa/decode.py

@@ -40,7 +40,6 @@ def decode_offline_recognizer(recognizer: sherpa.OfflineRecognizer,
     recognizer.decode_stream(s)
 
     text = s.result.text.strip()
-    print("text: {}".format(text))
     return text.lower()