Upload app.py

app.py

@@ -0,0 +1,40 @@
+import os
+import gradio as gr
+import librosa
+import soundfile as sf
+
+
+def inference(audio_1, audio_2):
+    # Load the audio files
+    translated_audio, sr_tr = librosa.load(audio_1, sr=None)
+    background_audio, sr_bg = librosa.load(audio_2, sr=None)
+
+    # Ensure both audio files have the same sample rate
+    if sr_bg != sr_tr:
+        translated_audio = librosa.resample(translated_audio, sr_tr, sr_bg)
+        sr = sr_bg
+    else:
+        sr = sr_bg  # or sr_tr, they're the same
+
+    # Pad or truncate the audio to make them the same length
+    max_len = max(len(background_audio), len(translated_audio))
+    background_audio = librosa.util.fix_length(background_audio, max_len)
+    translated_audio = librosa.util.fix_length(translated_audio, max_len)
+
+    # Mix the audio
+    full_audio = background_audio + translated_audio
+
+    # Normalize to prevent clipping
+    full_audio = librosa.util.normalize(full_audio)
+
+    # Write the output
+    return full_audio
+    
+title = "음성 합성"
+
+gr.Interface(
+    inference, 
+    [gr.Audio(type="filepath", label="Vocals"),gr.Audio(type="filepath", label="배경음")],
+    gr.Audio(type="numpy", label="합성 결과"),
+    title=title,
+    ).launch(enable_queue=True, debug=True)