Update app.py

app.py

@@ -10,7 +10,7 @@ from urllib.parse import urlparse
 # Clone and install faster-whisper from GitHub
 subprocess.run(["git", "clone", "https://github.com/SYSTRAN/faster-whisper.git"], check=True)
 subprocess.run(["pip", "install", "-e", "./faster-whisper"], check=True)
-subprocess.run(["pip", "install", "yt-dlp"], check=True)
+subprocess.run(["pip", "install", "yt-dlp pytube ffmpeg-python"], check=True)
 
 # Add the faster-whisper directory to the Python path
 sys.path.append("./faster-whisper")
@@ -21,71 +21,134 @@ import yt_dlp
 
 def download_audio(url):
     parsed_url = urlparse(url)
-    if parsed_url.netloc == 'www.youtube.com' or parsed_url.netloc == 'youtu.be':
-        # YouTube video
-        ydl_opts = {
-            'format': 'bestaudio/best',
-            'postprocessors': [{
-                'key': 'FFmpegExtractAudio',
-                'preferredcodec': 'mp3',
-                'preferredquality': '192',
-            }],
-            'outtmpl': '%(id)s.%(ext)s',
-        }
-        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
-            info = ydl.extract_info(url, download=True)
-            return f"{info['id']}.mp3"
+    if parsed_url.netloc in ['www.youtube.com', 'youtu.be', 'youtube.com']:
+        return download_youtube_audio(url)
     else:
-        # Direct MP3 URL
-        response = requests.get(url)
-        if response.status_code == 200:
-            with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as temp_file:
-                temp_file.write(response.content)
-                return temp_file.name
-        else:
-            raise Exception(f"Failed to download audio from {url}")
+        return download_direct_audio(url)
 
-def transcribe_audio(input_source, batch_size):
-    # Initialize the model
-    model = WhisperModel("cstr/whisper-large-v3-turbo-int8_float32", device="auto", compute_type="int8")
-    batched_model = BatchedInferencePipeline(model=model)
-
-    # Handle input source
-    if isinstance(input_source, str) and (input_source.startswith('http://') or input_source.startswith('https://')):
-        # It's a URL, download the audio
-        audio_path = download_audio(input_source)
+def download_youtube_audio(url):
+    methods = [
+        youtube_dl_method,
+        pytube_method,
+        youtube_dl_alternative_method,
+        ffmpeg_method
+    ]
+
+    for method in methods:
+        try:
+            return method(url)
+        except Exception as e:
+            print(f"Method {method.__name__} failed: {str(e)}")
+    
+    raise Exception("All download methods failed. Please try a different video or a direct audio URL.")
+
+def youtube_dl_method(url):
+    ydl_opts = {
+        'format': 'bestaudio/best',
+        'postprocessors': [{
+            'key': 'FFmpegExtractAudio',
+            'preferredcodec': 'mp3',
+            'preferredquality': '192',
+        }],
+        'outtmpl': '%(id)s.%(ext)s',
+    }
+    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+        info = ydl.extract_info(url, download=True)
+        return f"{info['id']}.mp3"
+
+def pytube_method(url):
+    from pytube import YouTube
+    yt = YouTube(url)
+    audio_stream = yt.streams.filter(only_audio=True).first()
+    out_file = audio_stream.download()
+    base, ext = os.path.splitext(out_file)
+    new_file = base + '.mp3'
+    os.rename(out_file, new_file)
+    return new_file
+
+def youtube_dl_alternative_method(url):
+    ydl_opts = {
+        'format': 'bestaudio/best',
+        'postprocessors': [{
+            'key': 'FFmpegExtractAudio',
+            'preferredcodec': 'mp3',
+            'preferredquality': '192',
+        }],
+        'outtmpl': '%(id)s.%(ext)s',
+        'no_warnings': True,
+        'quiet': True,
+        'no_check_certificate': True,
+        'prefer_insecure': True,
+        'nocheckcertificate': True,
+    }
+    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+        info = ydl.extract_info(url, download=True)
+        return f"{info['id']}.mp3"
+
+def ffmpeg_method(url):
+    output_file = tempfile.mktemp(suffix='.mp3')
+    command = ['ffmpeg', '-i', url, '-vn', '-acodec', 'libmp3lame', '-q:a', '2', output_file]
+    subprocess.run(command, check=True, capture_output=True)
+    return output_file
+
+def download_direct_audio(url):
+    response = requests.get(url)
+    if response.status_code == 200:
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as temp_file:
+            temp_file.write(response.content)
+            return temp_file.name
     else:
-        # It's a local file path
-        audio_path = input_source
-
-    # Benchmark transcription time
-    start_time = time.time()
-    segments, info = batched_model.transcribe(audio_path, batch_size=batch_size)
-    end_time = time.time()
-
-    # Generate transcription
-    transcription = ""
-    for segment in segments:
-        transcription += f"[{segment.start:.2f}s -> {segment.end:.2f}s] {segment.text}\n"
-
-    # Calculate metrics
-    transcription_time = end_time - start_time
-    real_time_factor = info.duration / transcription_time
-    audio_file_size = os.path.getsize(audio_path) / (1024 * 1024)  # Size in MB
-
-    # Prepare output
-    output = f"Transcription:\n\n{transcription}\n"
-    output += f"\nLanguage: {info.language}, Probability: {info.language_probability:.2f}\n"
-    output += f"Duration: {info.duration:.2f}s, Duration after VAD: {info.duration_after_vad:.2f}s\n"
-    output += f"Transcription time: {transcription_time:.2f} seconds\n"
-    output += f"Real-time factor: {real_time_factor:.2f}x\n"
-    output += f"Audio file size: {audio_file_size:.2f} MB"
-
-    # Clean up downloaded file if it was a URL
-    if isinstance(input_source, str) and (input_source.startswith('http://') or input_source.startswith('https://')):
-        os.remove(audio_path)
-
-    return output
+        raise Exception(f"Failed to download audio from {url}")
+
+def transcribe_audio(input_source, batch_size):
+    try:
+        # Initialize the model
+        model = WhisperModel("cstr/whisper-large-v3-turbo-int8_float32", device="auto", compute_type="int8")
+        batched_model = BatchedInferencePipeline(model=model)
+
+        # Handle input source
+        if isinstance(input_source, str) and (input_source.startswith('http://') or input_source.startswith('https://')):
+            # It's a URL, download the audio
+            audio_path = download_audio(input_source)
+        else:
+            # It's a local file path
+            audio_path = input_source
+
+        # Benchmark transcription time
+        start_time = time.time()
+        segments, info = batched_model.transcribe(audio_path, batch_size=batch_size)
+        end_time = time.time()
+
+        # Generate transcription
+        transcription = ""
+        for segment in segments:
+            transcription += f"[{segment.start:.2f}s -> {segment.end:.2f}s] {segment.text}\n"
+
+        # Calculate metrics
+        transcription_time = end_time - start_time
+        real_time_factor = info.duration / transcription_time
+        audio_file_size = os.path.getsize(audio_path) / (1024 * 1024)  # Size in MB
+
+        # Prepare output
+        output = f"Transcription:\n\n{transcription}\n"
+        output += f"\nLanguage: {info.language}, Probability: {info.language_probability:.2f}\n"
+        output += f"Duration: {info.duration:.2f}s, Duration after VAD: {info.duration_after_vad:.2f}s\n"
+        output += f"Transcription time: {transcription_time:.2f} seconds\n"
+        output += f"Real-time factor: {real_time_factor:.2f}x\n"
+        output += f"Audio file size: {audio_file_size:.2f} MB"
+
+        return output
+
+    except Exception as e:
+        return f"An error occurred: {str(e)}"
+
+    finally:
+        # Clean up downloaded file if it was a URL
+        if isinstance(input_source, str) and (input_source.startswith('http://') or input_source.startswith('https://')):
+            try:
+                os.remove(audio_path)
+            except:
+                pass
 
 # Gradio interface
 iface = gr.Interface(
@@ -102,6 +165,7 @@ iface = gr.Interface(
         ["https://example.com/path/to/audio.mp3", 16],
         ["path/to/local/audio.mp3", 16]
     ],
+    cache_examples=False  # Prevents automatic processing of examples
 )
 
 iface.launch()