Update app.py

app.py

@@ -14,17 +14,6 @@ import io
 import librosa
 import ffmpeg
 
-#from torchaudio.io import CodecConfig
-# import numpy
-
-# def my_read_file(audio_path, max_second):
-#     signal, sr, audio_length_second = read_as_single_channel_16k(audio_path, default_sr)
-#     if audio_length_second > max_second:
-#         signal = signal[0:default_sr * max_second]
-#         audio_length_second = max_second
-
-#     return signal, sr, audio_length_second
-
 def create_default_value():
     if "def_value" not in st.session_state:
         def_val_npy = np.random.choice([0, 1], size=32 - len_start_bit)
@@ -98,18 +87,6 @@ def main():
                 file_extension_ori =".mp3"
                 file_extension =".wav"
 
-            #RuntimeError: Could not infer dtype of numpy.float32
-            #wav = torch.tensor(wav3).float() / 32768.0
-            
-            #RuntimeError: Numpy is not available
-            # wav = torch.from_numpy(wav3) #/32768.0
-            # wav = wav.unsqueeze(0).unsqueeze(0)
-            # st.markdown("Before unsqueeze mp3")
-            # st.markdown(wav)
-        
-        #Unsqueeze for line 176
-        # wav= wav.unsqueeze(0)
-
             action = st.selectbox("Select Action", ["Add Watermark", "Detect Watermark"])
 
             if action == "Add Watermark":
@@ -205,179 +182,6 @@ def main():
 
     except RuntimeError:
         st.error("Please input audio with one channel (mono-channel)")
-        
-    # if audio_file:
-    #     # 保存文件到本地：
-    #     # tmp_input_audio_file = os.path.join("/tmp/", audio_file.name)
-    #     # st.markdown(tmp_input_audio_file)
-    #     # with open(tmp_input_audio_file, "wb") as f:
-    #     #     f.write(audio_file.getbuffer())
-    #     # st.audio(tmp_input_audio_file, format="mp3/wav")
-
-
-    #     #1st attempt
-    #     #audio_path = " audio_file.name"
-
-    #     # audio, sr = torchaudio.load(audio_file)
-    #     # st.audio(audio_file, format="audio/mpeg")
-    #     # audio= audio.unsqueeze(0)
-
-    #     # st.markdown("SR")
-    #     # st.markdown(sr)
-    #     # st.markdown("after unsqueeze wav or mp3")
-    #     # st.markdown(audio)
-        
-    #     #2nd attempt
-    #     # Save file to local storage
-    #     tmp_input_audio_file = os.path.join("/tmp/", audio_file.name)
-    #     file_extension = os.path.splitext(tmp_input_audio_file)[1].lower()
-    #     #st.markdown(file_extension)
-    #     if file_extension in [".wav", ".flac"]:
-    #         with open("test.wav", "wb") as f:
-    #             f.write(audio_file.getbuffer())
-
-    #         st.audio("test.wav", format="audio/wav")
-
-    #     elif file_extension == ".mp3":
-    #         with open("test.mp3", "wb") as f:
-    #             f.write(audio_file.getbuffer())
-
-    #         st.audio("test.mp3", format="audio/mpeg")
-
-    #     #Load the WAV file using torchaudio
-    #     if file_extension in [".wav", ".flac"]: 
-    #         wav, sample_rate = torchaudio.load("test.wav")
-    #         # st.markdown("Before unsquueze wav")
-    #         # st.markdown(wav)
-    #         file_extension_ori =".wav"
-    #         #Unsqueeze for line 176
-    #         wav= wav.unsqueeze(0)
-                
-    #     elif file_extension == ".mp3":
-    #         # Load an MP3 file
-    #         audio = AudioSegment.from_mp3("test.mp3")
-
-    #         # Export it as a WAV file
-    #         audio.export("test.wav", format="wav")
-    #         wav3, sample_rate = torchaudio.load("test.wav")
-    #         wav= wav3.unsqueeze(0)
-    #         file_extension_ori =".mp3"
-    #         file_extension =".wav"
-
-    #         #RuntimeError: Could not infer dtype of numpy.float32
-    #         #wav = torch.tensor(wav3).float() / 32768.0
-            
-    #         #RuntimeError: Numpy is not available
-    #         # wav = torch.from_numpy(wav3) #/32768.0
-    #         # wav = wav.unsqueeze(0).unsqueeze(0)
-    #         # st.markdown("Before unsqueeze mp3")
-    #         # st.markdown(wav)
-        
-    #     #Unsqueeze for line 176
-    #     # wav= wav.unsqueeze(0)
-
-    #     action = st.selectbox("Select Action", ["Add Watermark", "Detect Watermark"])
-
-    #     if action == "Add Watermark":
-    #         #watermark_text = st.text_input("The watermark (0, 1 list of length-16):", value=st.session_state.def_value)
-    #         add_watermark_button = st.button("Add Watermark", key="add_watermark_btn")
-    #         if add_watermark_button:  # 点击按钮后执行的
-    #             #if audio_file and watermark_text:
-    #             if audio_file:
-    #                 with st.spinner("Adding Watermark..."):
-    #                     #wav = my_read_file(wav,max_second_encode)
-                        
-    #                     #1st attempt
-    #                     watermark = model.get_watermark(wav, default_sr)
-    #                     watermarked_audio = wav + watermark
-    #                     print(watermarked_audio.size())
-    #                     size = watermarked_audio.size()
-    #                     #st.markdown(size)
-
-    #                     print(watermarked_audio.squeeze())
-    #                     squeeze = watermarked_audio.squeeze(1)
-    #                     shape = squeeze.size()
-    #                     #st.markdown(shape)
-
-    #                     #st.markdown(squeeze)
-    #                 if file_extension_ori in [".wav", ".flac"]:
-    #                     torchaudio.save("output.wav", squeeze, default_sr, bits_per_sample=16)
-    #                     watermarked_wav = torchaudio.save("output.wav", squeeze, default_sr, bits_per_sample=16)
-
-    #                     st.audio("output.wav", format="audio/wav")
-
-    #                     with open("output.wav", "rb") as file:
-    #                         #file.read()
-    #                         #file.write(watermarked_wav.getbuffer())
-    #                         binary_data = file.read()
-    #                         btn = st.download_button(
-    #                             label="Download watermarked audio",
-    #                             data=binary_data,
-    #                             file_name="output.wav",
-    #                             mime="audio/wav",
-    #                         )
-
-
-    #                 elif file_extension_ori == ".mp3":
-    #                     torchaudio.save("output.wav", squeeze, default_sr)
-    #                     watermarked_mp3 = torchaudio.save("output.wav", squeeze, default_sr)
-    #                     audio = AudioSegment.from_wav("output.wav")
-
-    #                     # Export as MP3
-    #                     audio.export("output.mp3", format="mp3")
-    #                     st.audio("output.mp3", format="audio/mpeg")
-                        
-    #                     with open("output.mp3", "rb") as file:
-    #                         #file.write(watermarked_wav.getbuffer())
-    #                         binary_data = file.read()
-    #                         st.download_button(
-    #                             label="Download watermarked audio",
-    #                             data=binary_data,
-    #                             file_name="output.mp3",
-    #                             mime="audio/mpeg",
-    #                         )
-    #                 # except RuntimeError:
-    #                 #     st.error("Please input audio with one channel (mono-channel)")
-                                 
-    #     elif action == "Detect Watermark":
-    #         detect_watermark_button = st.button("Detect Watermark", key="detect_watermark_btn")
-
-    #         # if audio_file:
-    #         #     #1st attempt
-    #         #     watermark = model.get_watermark(wav, default_sr)
-    #         #     watermarked_audio = wav + watermark
-    #         #     print(watermarked_audio.size())
-    #         #     size = watermarked_audio.size()
-    #         #     #st.markdown(size)
-
-                
-            # if detect_watermark_button:
-            #     with st.spinner("Detecting..."):
-            #         # result, message = detector.detect_watermark(watermarked_audio, sample_rate=default_sr, message_threshold=0.5)
-            #         # st.markdown("Probability of audio being watermarked: ")
-            #         # st.markdown(result)
-            #         # st.markdown("This is likely a watermarked audio!")
-            #         # print(f"\nThis is likely a watermarked audio: {result}")
-
-            #         #Run on an unwatermarked audio
-
-            #         if file_extension in [".wav", ".flac"]: 
-            #             wav, sample_rate = torchaudio.load("test.wav")
-            #             wav= wav.unsqueeze(0)
-
-            #         elif file_extension == ".mp3":
-            #             # Load an MP3 file
-            #             audio = AudioSegment.from_mp3("test.mp3")
-            #             # Export it as a WAV file
-            #             audio.export("test.wav", format="wav")
-            #             wav, sample_rate = torchaudio.load("test.wav")
-            #             wav= wav.unsqueeze(0)
-                        
-            #         result2, message2 = detector.detect_watermark(wav, sample_rate=default_sr, message_threshold=0.5)
-            #         print(f"This is likely an unwatermarked audio: {result2}")
-            #         st.markdown("Probability of audio being watermarked: ")
-            #         st.markdown(result2)
-            #         st.markdown("This is likely an unwatermarked audio!")
 
 
 if __name__ == "__main__":
@@ -389,9 +193,4 @@ if __name__ == "__main__":
     # model = wavmark.load_model().to(device)
     model = AudioSeal.load_generator("audioseal_wm_16bits")
     detector = AudioSeal.load_detector(("audioseal_detector_16bits"))
-    main()
-
-    # audio_path = "/Users/my/Library/Mobile Documents/com~apple~CloudDocs/CODE/PycharmProjects/4_语音水印/419_huggingface水印/WavMark/example.wav"
-
-    # decoded_watermark, decode_cost = decode_watermark(audio_path)
-    # print(decoded_watermark)
+    main()