Add conversion to 16k mono

app.py

@@ -5,6 +5,32 @@ from typing import List
 import soundfile as sf
 import gradio as gr
 import tempfile
+import subprocess
+
+def convert_to_16kHz_mono(input_file, output_file):
+    """
+    Converts an audio file to 16KHz sample rate and single channel (mono) using ffmpeg.
+
+    Parameters:
+    input_file (str): Path to the input audio file.
+    output_file (str): Path to the output WAV file.
+    """
+    try:
+        # Run the ffmpeg command
+        subprocess.run(['ffmpeg', '-y', '-i', input_file, '-ar', '16000', '-ac', '1', output_file], check=True)
+        print(f"Conversion complete: {output_file}")
+        return output_file
+    except subprocess.CalledProcessError as e:
+        print(f"An error occurred during conversion: {e}")
+
+def create_temp_wav_file():
+    # Create a temporary file using NamedTemporaryFile
+    temp_file = tempfile.NamedTemporaryFile(suffix=".wav", delete=False)
+    
+    # Get the path of the temporary file
+    temp_file_path = temp_file.name
+    
+    return temp_file_path
 
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 knn_vc = torch.hub.load('bshall/knn-vc', 'knn_vc', prematched=True, trust_repo=True, pretrained=True, device=device)
@@ -12,6 +38,11 @@ knn_vc = torch.hub.load('bshall/knn-vc', 'knn_vc', prematched=True, trust_repo=T
 
 def convert_voice(src_wav_path:str, ref_wav_paths, top_k:int):
 
+    tmp_src_wav_path = create_temp_wav_file()
+    tmp_ref_wav_path = create_temp_wav_file()
+    src_wav_path = convert_to_16kHz_mono(src_wav_path, tmp_src_wav_path)
+    ref_wav_paths = convert_to_16kHz_mono(ref_wav_paths, tmp_ref_wav_path)
+
     query_seq = knn_vc.get_features(src_wav_path)
     matching_set = knn_vc.get_matching_set([ref_wav_paths])
     out_wav = knn_vc.match(query_seq, matching_set, topk=int(top_k))