.

app.py

@@ -2,9 +2,10 @@ import gradio as gr
 import whisper
 from langcodes import *
 
-def speech_to_text(tmp_filename, model_size):
+def speech_to_text(tmp_filename, uploaded, model_size):
     model = whisper.load_model(model_size)
-    result = model.transcribe(tmp_filename)
+    source = uploaded if uploaded is not None else tmp_filename
+    result = model.transcribe(source)
     return f'Detected language: {Language.make(language=result["language"]).display_name()}\n\n You said: {result["text"]}'
 
 
@@ -20,6 +21,7 @@ gr.Interface(
     fn=speech_to_text,
     inputs=[
         gr.Audio(label="Record your voice on your mic",source="microphone", type="filepath"),
+        gr.Audio(source="upload", type="filepath", label="Upload Audio"), 
         gr.Dropdown(label="Select model size",value="base",choices=["tiny", "base", "small", "medium", "large"])],
     outputs="text").launch()