update

main.py

@@ -37,12 +37,13 @@ def process_uploaded_file(
         filename: str,
         silence_time: float = 0.3,
         longest_activate: float = 3.0,
+        speech_pad_time: float = 0.03,
         threshold: float = 0.5,
 ) -> Tuple[str, str]:
     if vad_engine == "nx_vad":
         return run_nx_vad(filename, silence_time, longest_activate)
     elif vad_engine == "silero_vad":
-        return run_silero_vad(filename, silence_time, threshold)
+        return run_silero_vad(filename, silence_time, speech_pad_time, threshold)
     else:
         return f"vad engine invalid: {vad_engine}", ""
 
@@ -67,13 +68,17 @@ def run_nx_vad(filename: str, silence_time: float = 0.3, longest_activate: float
     return vad_timestamps, raw_vad_result
 
 
-def run_silero_vad(filename: str, silence_time: float = 0.3, threshold: float = 0.5) -> Tuple[str, str]:
+def run_silero_vad(filename: str,
+                   silence_time: float = 0.3,
+                   speech_pad_time: float = 0.03,
+                   threshold: float = 0.5
+                   ) -> Tuple[str, str]:
     filename = Path(filename).as_posix()
 
     main_logger.info("do silero vad: {}".format(filename))
 
-    cmd = "vad_bins/silero_vad --filename {} --silence_time {} --threshold {}".format(
-        filename, silence_time, threshold
+    cmd = "vad_bins/silero_vad --filename {} --silence_time {} --speech_pad_time {} --threshold {}".format(
+        filename, silence_time, speech_pad_time, threshold
     )
     raw_vad_result = Command.popen(cmd)
 
@@ -123,8 +128,9 @@ def main():
 
                 with gr.Row():
                     uploaded_vad_engine = gr.Dropdown(choices=["nx_vad", "silero_vad"], value="nx_vad", label="vad_engine")
-                    uploaded_silence_time = gr.Slider(minimum=0.0, maximum=1.0, value=0.3, step=0.01, label="silence time")
-                    uploaded_longest_activate = gr.Slider(minimum=0.0, maximum=20.0, value=3.0, step=0.1, label="longest activate")
+                    uploaded_silence_time = gr.Slider(minimum=0.0, maximum=1.0, value=0.3, step=0.01, label="silence_time")
+                    uploaded_longest_activate = gr.Slider(minimum=0.0, maximum=20.0, value=3.0, step=0.1, label="longest_activate")
+                    uploaded_speech_pad_time = gr.Slider(minimum=0.0, maximum=20.0, value=3.0, step=0.1, label="speech_pad_time")
                     uploaded_threshold = gr.Slider(minimum=0.0, maximum=1.0, value=0.5, step=0.1, label="threshold")
                 upload_button = gr.Button("Run VAD", variant="primary")
 
@@ -139,6 +145,7 @@ def main():
                         uploaded_file,
                         uploaded_silence_time,
                         uploaded_longest_activate,
+                        uploaded_speech_pad_time,
                         uploaded_threshold,
                     ],
                     outputs=[
@@ -155,6 +162,7 @@ def main():
                         uploaded_file,
                         uploaded_silence_time,
                         uploaded_longest_activate,
+                        uploaded_speech_pad_time,
                         uploaded_threshold,
                     ],
                     outputs=[