Update app.py

app.py

@@ -1,119 +1,120 @@
 import gradio as gr
-import os
-import uuid
-import re
+import os, uuid, re, tempfile
 from pydub import AudioSegment, silence
 
-# -------------------------------------------------
-# Clean filename
-# -------------------------------------------------
-def clean_file_name(file_path):
-    base = os.path.basename(file_path)
-    name, ext = os.path.splitext(base)
+# ----------------------------------
+# Utils
+# ----------------------------------
+def clean_name(path, ext):
+    name = re.sub(r'[^a-zA-Z0-9]+', '_', os.path.basename(path))
+    return f"{name}_{uuid.uuid4().hex[:6]}.{ext}"
 
-    name = re.sub(r'[^a-zA-Z0-9]+', '_', name)
-    name = re.sub(r'_+', '_', name).strip('_')
+# ----------------------------------
+# Convert to WAV (FAST PROCESSING)
+# ----------------------------------
+def ensure_wav(path):
+    if path.lower().endswith(".wav"):
+        return path
 
-    uid = uuid.uuid4().hex[:6]
-    return os.path.join(os.path.dirname(file_path), f"{name}_{uid}{ext}")
+    audio = AudioSegment.from_file(path)
+    audio = audio.set_channels(1)
 
-# -------------------------------------------------
-# SAFE Silence Removal (NO WORD LOSS)
-# -------------------------------------------------
-def remove_silence_safe(file_path, keep_silence_ms=250):
-    audio = AudioSegment.from_file(file_path)
+    tmp = tempfile.gettempdir()
+    wav_path = os.path.join(tmp, clean_name(path, "wav"))
+    audio.export(wav_path, format="wav")
+    return wav_path
 
-    # mono for speed, safe for voice
-    audio = audio.set_channels(1)
+# ----------------------------------
+# SAFE Silence Removal
+# ----------------------------------
+def remove_silence_safe(path, keep_ms=250):
+    audio = AudioSegment.from_wav(path)
 
-    # 🔐 conservative threshold (speech friendly)
     silence_thresh = audio.dBFS - 22
 
     chunks = silence.split_on_silence(
         audio,
-        min_silence_len=350,       # ignore natural speech pauses
+        min_silence_len=350,
         silence_thresh=silence_thresh,
-        keep_silence=keep_silence_ms
+        keep_silence=keep_ms
     )
 
     if not chunks:
-        return file_path
-
-    output = AudioSegment.empty()
+        return path
 
-    for chunk in chunks:
-        # soften edges so words never cut
-        chunk = chunk.fade_in(20).fade_out(20)
-        output += chunk
+    out = AudioSegment.empty()
+    for c in chunks:
+        out += c.fade_in(20).fade_out(20)
 
-    output_path = clean_file_name(file_path)
-    output.export(output_path)
-    return output_path
+    out_path = os.path.join(
+        tempfile.gettempdir(),
+        clean_name(path, "wav")
+    )
+    out.export(out_path, format="wav")
+    return out_path
 
-# -------------------------------------------------
-# Duration helper
-# -------------------------------------------------
+# ----------------------------------
+# Duration
+# ----------------------------------
 def duration(path):
-    audio = AudioSegment.from_file(path)
-    return len(audio) / 1000
+    return len(AudioSegment.from_wav(path)) / 1000
 
-# -------------------------------------------------
-# Main processing
-# -------------------------------------------------
+# ----------------------------------
+# Main
+# ----------------------------------
 def process_audio(audio_file, keep_seconds):
     keep_ms = int(max(keep_seconds, 0.25) * 1000)
 
-    before = duration(audio_file)
-    output = remove_silence_safe(audio_file, keep_ms)
-    after = duration(output)
+    wav = ensure_wav(audio_file)
+    before = duration(wav)
+
+    cleaned = remove_silence_safe(wav, keep_ms)
+    after = duration(cleaned)
+
+    # Export final MP3 (user friendly)
+    final_mp3 = cleaned.replace(".wav", ".mp3")
+    AudioSegment.from_wav(cleaned).export(final_mp3, format="mp3")
 
     info = (
-        f"Old Duration : {before:.2f} sec\n"
-        f"New Duration : {after:.2f} sec\n"
-        f"Silence Removed : {before - after:.2f} sec"
+        f"Old Duration : {before:.2f}s\n"
+        f"New Duration : {after:.2f}s\n"
+        f"Silence Removed : {before - after:.2f}s"
     )
 
-    return output, output, info
+    return final_mp3, final_mp3, info
 
-# -------------------------------------------------
-# UI (Hugging Face SAFE)
-# -------------------------------------------------
-with gr.Blocks(title="Safe Silence Remover") as demo:
+# ----------------------------------
+# UI
+# ----------------------------------
+with gr.Blocks(title="Fast Silence Remover") as demo:
     gr.Markdown(
-        """
-        ## 🔇 Safe Silence Remover  
-        ✔ Removes **only silence**  
-        ✔ Words & voice **100% preserved**  
-        """
+        "## ⚡ Fast MP3/WAV Silence Remover (No Word Loss)"
     )
 
     with gr.Row():
         with gr.Column():
-            audio_input = gr.Audio(
-                label="Upload Audio",
+            audio = gr.Audio(
                 type="filepath",
-                sources=["upload", "microphone"]
+                sources=["upload"],
+                label="Upload MP3 or WAV"
             )
 
-            keep_silence = gr.Number(
-                label="Keep Silence (seconds)",
-                value=0.25
+            keep = gr.Number(
+                value=0.25,
+                label="Keep Silence (seconds)"
             )
 
             btn = gr.Button("Remove Silence")
 
         with gr.Column():
-            audio_out = gr.Audio(label="Processed Audio")
-            file_out = gr.File(label="Download")
-            info_out = gr.Textbox(label="Info")
+            out_audio = gr.Audio(label="Processed Audio")
+            out_file = gr.File(label="Download")
+            info = gr.Textbox(label="Info")
 
     btn.click(
         process_audio,
-        [audio_input, keep_silence],
-        [audio_out, file_out, info_out]
+        [audio, keep],
+        [out_audio, out_file, info]
     )
 
-# -------------------------------------------------
-# Hugging Face Entry Point
-# -------------------------------------------------
 demo.queue().launch()