Update app.py

app.py

@@ -8,6 +8,7 @@ from datetime import datetime
 import gradio as gr
 import torch
 from fairseq import checkpoint_utils
+import librosa
 
 from infer_pack.models import SynthesizerTrnMs256NSFsid, SynthesizerTrnMs256NSFsid_nono
 from vc_infer_pipeline import VC
@@ -17,10 +18,10 @@ logging.getLogger("numba").setLevel(logging.WARNING)
 
 
 def create_vc_fn(tgt_sr, net_g, vc, if_f0, file_index, file_big_npy):
-    def vc_fn(vc_microphone, vc_transpose, vc_f0method, vc_index_ratio):
+    def vc_fn(vc_audio_file, vc_transpose, vc_f0method, vc_index_ratio):
         try:
-            # Get the recorded audio from the microphone
-            audio, sr = vc_microphone.record(num_frames=16000)  # Adjust the sample rate if needed
+            # Load the audio file uploaded via Gradio
+            audio, sr = librosa.load(vc_audio_file.name, sr=None, mono=True)
 
             # Your existing processing logic for audio
             times = [0, 0, 0]
@@ -118,8 +119,8 @@ if __name__ == '__main__':
                         )
                     with gr.Row():
                         with gr.Column():
-                            # Use microphone instead of file upload
-                            vc_microphone = gr.Microphone(label="Record your voice")
+                            # Use file upload instead of microphone
+                            vc_audio_file = gr.File(label="Upload your audio file", type="audio")
                             vc_transpose = gr.Number(label="Transpose", value=0)
                             vc_f0method = gr.Radio(
                                 label="Pitch extraction algorithm, PM is fast but Harvest is better for low frequencies",
@@ -139,5 +140,5 @@ if __name__ == '__main__':
                             vc_output1 = gr.Textbox(label="Output Message")
                             vc_output2 = gr.Audio(label="Output Audio")
 
-                vc_submit.click(vc_fn, [vc_microphone, vc_transpose, vc_f0method, vc_index_ratio], [vc_output1, vc_output2])
-        app.queue(concurrency_count=1, max_size=20, api_open=args.api).launch(share=args.share)
+                vc_submit.click(vc_fn, [vc_audio_file, vc_transpose, vc_f0method, vc_index_ratio], [vc_output1, vc_output2])
+        app.queue(concurrency_count=1, max_size=20, api_open=args.api).launch(share=args.share)