ver 1.9

app.py

@@ -10,6 +10,7 @@ import gradio as gr
 from uroman import uroman
 import concurrent.futures
 from pydub import AudioSegment
+from pydub.effects import speedup
 from datasets import load_dataset
 from IPython.display import Audio
 from scipy.signal import butter, lfilter
@@ -184,12 +185,10 @@ def generate_audio_with_pause(srt_file_path):
         desired_duration = time_to_seconds(end_time) - time_to_seconds(start_time)
         current_duration = len(audio_data) / 16000
         
-        # Adjust speed
-        if current_duration != desired_duration:
-            factor = desired_duration / current_duration
-            indices = np.round(np.arange(0, len(audio_data), factor)).astype(int)
-            indices = indices[indices < len(audio_data)]
-            audio_data = audio_data[indices]
+        # Adjust audio speed by speedup
+        if desired_duration < current_duration:
+            speedup_factor = current_duration / desired_duration
+            audio_data = speedup(audio_data, playback_speed=speedup_factor)
         
         audio_clips.append(audio_data)
 
@@ -290,7 +289,7 @@ with gr.Blocks(css=css) as demo:
         height=100
     )
 
-    inp.change
+    inp.change(check_input_files, inputs=inp)
     btn.click(fn=srt_to_audio_multi, inputs=inp, outputs=out)
     download_btn.click(fn=download_all, inputs=out, outputs=download_out)