Update app.py

app.py

@@ -1,90 +1,109 @@
 import gradio as gr
-import subprocess
-import traceback
+from moviepy.editor import VideoFileClip
 import speech_recognition as sr
-import argostranslate.package
-import argostranslate.translate
+from argostranslate import package, translate
+import os
+import tempfile
 
-# -------------------------------
-# 1. Extract audio from video
-# -------------------------------
+# ---------------------------
+# Setup Argos Translate (offline)
+# ---------------------------
+def setup_translation():
+    # Download English->Urdu model if not exists
+    if not os.path.exists("en_ur.argosmodel"):
+        import urllib.request
+        url = "https://www.argosopentech.com/argospm/models/en_ur.argosmodel"
+        urllib.request.urlretrieve(url, "en_ur.argosmodel")
+        package.install_from_path("en_ur.argosmodel")
+setup_translation()
+
+# ---------------------------
+# Extract audio from video
+# ---------------------------
 def extract_audio(video_path):
-    audio_path = "audio.wav"
-    cmd = f"ffmpeg -y -i '{video_path}' -ar 16000 -ac 1 -f wav {audio_path}"
-    subprocess.run(cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    clip = VideoFileClip(video_path)
+    audio_path = tempfile.mktemp(suffix=".wav")
+    clip.audio.write_audiofile(audio_path, fps=16000, codec="pcm_s16le")
     return audio_path
 
-# -------------------------------
-# 2. Transcribe using CMU Sphinx (offline)
-# -------------------------------
+# ---------------------------
+# Transcribe audio using CMU Sphinx
+# ---------------------------
 def transcribe_audio(audio_path):
+    r = sr.Recognizer()
+    with sr.AudioFile(audio_path) as source:
+        audio = r.record(source)
     try:
-        recognizer = sr.Recognizer()
-        with sr.AudioFile(audio_path) as source:
-            audio = recognizer.record(source)
-        text = recognizer.recognize_sphinx(audio)
-        return text.strip()
-    except Exception:
-        return f"STT Error:\n{traceback.format_exc()}"
+        text = r.recognize_sphinx(audio)
+        return text
+    except sr.UnknownValueError:
+        return "[Could not understand audio]"
+    except sr.RequestError as e:
+        return f"[Sphinx error: {e}]"
 
-# -------------------------------
-# 3. Translate text offline
-# -------------------------------
-def translate_text_offline(text, target_lang):
+# ---------------------------
+# Translate text using Argos Translate
+# ---------------------------
+def translate_text(text, target_lang):
     if target_lang == "original":
         return text
-    try:
-        installed_langs = argostranslate.translate.get_installed_languages()
-        lang_map = {"en":"en","ur":"ur","hi":"hi","ps":"ps","ar":"ar"}
-        target_code = lang_map.get(target_lang,"en")
-        target_lang_obj = next((l for l in installed_langs if l.code==target_code), None)
-        if not installed_langs or not target_lang_obj:
-            return text
-        # assume first installed language is source
-        translated = installed_langs[0].translate(text, target_lang_obj)
+    from_lang = "en"
+    to_lang = target_lang
+    installed_languages = translate.get_installed_languages()
+    from_lang_obj = next((l for l in installed_languages if l.code == from_lang), None)
+    to_lang_obj = next((l for l in installed_languages if l.code == to_lang), None)
+    if from_lang_obj and to_lang_obj:
+        translated = from_lang_obj.get_translation(to_lang_obj).translate(text)
         return translated
-    except Exception:
-        return f"Translation Error:\n{traceback.format_exc()}"
-
-# -------------------------------
-# 4. Main processing
-# -------------------------------
-def process_video(video, lang):
-    try:
-        if not video:
-            return "", "No video uploaded"
-
-        audio_path = extract_audio(video)
-        transcription = transcribe_audio(audio_path)
-        translation = translate_text_offline(transcription, lang)
+    return text
 
-        logs = f"Audio extracted!\nTranscribed text length: {len(transcription)}"
-        return translation, logs
-    except Exception:
-        return "", f"❌ ERROR:\n{traceback.format_exc()}"
-
-# -------------------------------
-# 5. Gradio UI
-# -------------------------------
+# ---------------------------
+# Gradio Interface
+# ---------------------------
 languages = {
-    "original":"Original",
-    "en":"English",
-    "ur":"Urdu",
-    "hi":"Hindi",
-    "ps":"Pashto",
-    "ar":"Arabic"
+    "original": "No Translation",
+    "ur": "Urdu",
+    "hi": "Hindi",
+    "ps": "Pashto",
+    "ar": "Arabic",
+    "en": "English"
 }
 
-with gr.Blocks() as demo:
-    gr.Markdown("## 🎬 Video → Text → Translation (Offline, CPU, Token-free)")
+def process_video(video_file, lang):
+    error_log = ""
+    try:
+        audio_path = extract_audio(video_file.name)
+        error_log += "Audio extracted!\n"
+    except Exception as e:
+        return "", f"Audio extraction error: {e}"
 
-    video_input = gr.Video(label="Upload Video")
-    lang_dropdown = gr.Dropdown(list(languages.keys()), value="original", label="Translate To")
-    btn = gr.Button("Generate Text")
+    try:
+        text = transcribe_audio(audio_path)
+        error_log += f"Transcribed text length: {len(text)}\n"
+    except Exception as e:
+        return "", f"STT Error: {e}"
+
+    try:
+        translated = translate_text(text, lang)
+        error_log += f"Translation done!\n"
+    except Exception as e:
+        translated = text
+        error_log += f"Translation error: {e}\n"
 
-    text_output = gr.Textbox(label="Transcribed / Translated Text", lines=10)
-    debug_box = gr.Textbox(label="Debug Logs", lines=8)
+    return translated, error_log
 
-    btn.click(process_video, inputs=[video_input, lang_dropdown], outputs=[text_output, debug_box])
+demo = gr.Interface(
+    fn=process_video,
+    inputs=[
+        gr.Video(label="Upload Video"),
+        gr.Dropdown(list(languages.keys()), value="original", label="Translate To")
+    ],
+    outputs=[
+        gr.Textbox(label="Transcribed / Translated Text", interactive=False),
+        gr.Textbox(label="Debug / Error Log", interactive=False)
+    ],
+    title="Offline Video Subtitle Generator",
+    description="Upload a video → Extract audio → Generate subtitles → Optional translation → All offline, token-free"
+)
 
 demo.launch()