|
import gradio as gr |
|
import whisper |
|
|
|
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'{result["text"]}' |
|
|
|
gr.Interface( |
|
title="", |
|
thumbnail="", |
|
css=""" |
|
footer {visibility: xhidden} |
|
.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(debug = True) |