Update app.py

app.py

@@ -2,6 +2,7 @@ import gradio as gr
 from pytube import YouTube
 from transformers import pipeline
 import os
+from moviepy.editor import VideoFileClip
 
 
 pipe = pipeline(model="GIanlucaRub/whisper-tiny-it-6")  # change to "your-username/the-name-you-picked"
@@ -21,8 +22,17 @@ def populate_metadata(link):
   yt = YouTube(link)
   return yt.thumbnail_url, yt.title
 
-title="Youtube Whisperer"
-description="Speech to text transcription of Youtube videos using OpenAI's Whisper"
+def transcribe_video(video):
+    clip = VideoFileClip(video)
+    audio = video[:-4] + ".mp3"
+    clip.audio.write_audiofile(audio)
+    clip.close()
+    os.remove(video)
+    text = transcribe_audio(audio)
+    os.remove(audio)
+    
+    return text
+
 block = gr.Blocks()
 
 with block:
@@ -30,10 +40,10 @@ with block:
         """
             <div style="text-align: center; max-width: 500px; margin: 0 auto;">
               <div>
-                <h1>Youtube Whisperer</h1>
+                <h1>Whisper Tiny Italian</h1>
               </div>
               <p style="margin-bottom: 10px; font-size: 94%">
-                Speech to text transcription of Youtube videos using OpenAI's Whisper
+                Realtime demo for Italian speech recognition using a fine-tuned Whisper tiny model.
               </p>
             </div>
         """
@@ -46,19 +56,22 @@ with block:
               lines=5)
         
           microphone=gr.Audio(source="microphone", type="filepath")
-          
           with gr.Row().style(mobile_collapse=False, equal_height=True): 
               btn_microphone = gr.Button("Transcribe microphone audio")
                 
           audio_uploaded=gr.Audio(source="upload", type="filepath")
-          
           with gr.Row().style(mobile_collapse=False, equal_height=True): 
               btn_audio_uploaded = gr.Button("Transcribe audio uploaded")
+          
+          video_uploaded = gr.Video(source = "upload") 
+          with gr.Row().style(mobile_collapse=False, equal_height=True): 
+              btn_video_uploaded = gr.Button("Transcribe video uploaded")
+            
                 
           link = gr.Textbox(label="YouTube Link")
           with gr.Row().style(mobile_collapse=False, equal_height=True): 
               btn_youtube = gr.Button("Transcribe Youtube video") 
-        
+    
           with gr.Row().style(mobile_collapse=False, equal_height=True):
             title = gr.Label(label="Video Title", placeholder="Title")
             img = gr.Image(label="Thumbnail")
@@ -69,6 +82,7 @@ with block:
           btn_youtube.click(transcribe_yt, inputs=[link], outputs=[text])
           btn_microphone.click(transcribe_audio, inputs=[microphone], outputs=[text])
           btn_audio_uploaded.click(transcribe_audio, inputs=[audio_uploaded], outputs=[text])
+          btn_video_uploaded.click(transcribe_video, inputs=[video_uploaded], outputs=[text])
           link.change(populate_metadata, inputs=[link], outputs=[img, title])
 
 block.launch(debug=True)