Create app.py

app.py

@@ -0,0 +1,24 @@
+import gradio as gr
+import whisper
+from langcodes import *
+
+def speech_to_text(tmp_filename, uploaded, model_size):
+    model = whisper.load_model(model_size)
+    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"]}'
+
+gr.Interface(
+    title="",
+    thumbnail="",
+    css="""
+    footer {visibility: hidden}
+    .gr-prose p{text-align: center;}
+    .gr-button {background: black;color: white}
+    """,
+    description="",
+    fn=speech_to_text,
+    inputs=[
+        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()