select language

app.py

@@ -1,14 +1,16 @@
 import gradio as gr
 import whisper
+from whisper import tokenizer
 from pytube import YouTube
 
 loaded_model = whisper.load_model("base")
 current_size = 'base'
-def inference(link):
+AUTO_DETECT_LANG = "Auto Detect"
+
+def inference(link,language):
   yt = YouTube(link)
   path = yt.streams.filter(only_audio=True)[0].download(filename="audio.mp4")
-  options = whisper.DecodingOptions(without_timestamps=True)
-  results = loaded_model.transcribe(path)
+  results = loaded_model.transcribe(path,without_timestamps=True,language=language)
   return results['text']
 
 def change_model(size):
@@ -41,9 +43,15 @@ with block:
     with gr.Group():
         with gr.Box():
           sz = gr.Dropdown(label="Model Size", choices=['base','small', 'medium', 'large'], value='base')
-          
-          link = gr.Textbox(label="YouTube Link")
-          
+          with gr.Row(mobile_collaps=False,equal_height=True):
+            link = gr.Textbox(label="YouTube Link")
+            available_languages = sorted(tokenizer.TO_LANGUAGE_CODE.keys())
+            available_languages = [AUTO_DETECT_LANG]+available_languages
+            language = gr.Dropdown(label="Language",choices=available_languages,value=AUTO_DETECT_LANG)
+            
+            if language==AUTO_DETECT_LANG:
+              language=None
+
           with gr.Row().style(mobile_collapse=False, equal_height=True):
             title = gr.Label(label="Video Title", placeholder="Title")
             img = gr.Image(label="Thumbnail")
@@ -55,7 +63,7 @@ with block:
               btn = gr.Button("Transcribe")       
           
           # Events
-          btn.click(inference, inputs=[link], outputs=[text])
+          btn.click(inference, inputs=[link,language], outputs=[text])
           link.change(populate_metadata, inputs=[link], outputs=[img, title])
           sz.change(change_model, inputs=[sz], outputs=[])