Update app.py

app.py

@@ -1,11 +1,14 @@
 import gradio as gr
-from transformers import pipeline
+from faster_whisper import WhisperModel
 from pydub import AudioSegment
 import os
 import tempfile
+from transformers import pipeline
+
+# הגדרת המודל לתמלול
+model = WhisperModel("ivrit-ai/faster-whisper-v2-d4")
 
-# יצירת pipeline לתמלול ולסיכום
-transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-base", task="transcribe")
+# הגדרת pipeline לסיכום
 summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
 
 def summarize_audio_or_video(file_path):
@@ -16,14 +19,9 @@ def summarize_audio_or_video(file_path):
         else:
             audio_file = file_path
 
-        # חלוקת האודיו לקטעים של 30 שניות
-        segments = split_audio(audio_file)
-
-        # תמלול כל קטע והצטרפות התמלול הסופי
-        transcript = ""
-        for segment in segments:
-            segment_text = transcriber(segment, return_timestamps=True)["text"]
-            transcript += " " + segment_text
+        # תמלול האודיו
+        segments, _ = model.transcribe(audio_file, language="he")
+        transcript = " ".join([segment.text for segment in segments])
 
         # סיכום התמלול
         summary = summarizer(transcript, max_length=50, min_length=25, do_sample=False)[0]["summary_text"]
@@ -44,16 +42,6 @@ def convert_video_to_audio(video_file):
     video.export(temp_audio, format="wav")
     return temp_audio
 
-def split_audio(audio_file, segment_length=30 * 1000):  # 30 שניות במילישניות
-    audio = AudioSegment.from_file(audio_file)
-    segments = []
-    for i in range(0, len(audio), segment_length):
-        segment = audio[i:i + segment_length]
-        temp_segment = tempfile.mktemp(suffix=".wav")
-        segment.export(temp_segment, format="wav")
-        segments.append(temp_segment)
-    return segments
-
 # הגדרת ממשק Gradio
 interface = gr.Interface(
     fn=summarize_audio_or_video,