app.py

import gradio as gr
import whisper

# define function for transcription
def whisper_transcript(model_size, audio_file):
    
    source = audio_file
    language = "english"
    loaded_model = whisper.load_model(model_size)
    transcript = loaded_model.transcribe(source, language=language)

    return transcript["text"]

# define Gradio app interface
gradio_ui = gr.Interface(
    fn=whisper_transcript,
    theme="Nymbo/Nymbo_Theme",
    title="Transcribir audios en inglés a texto",
    description="**Cómo usar**: Elegir uno de los modelos, subir un audio y clicar el botón de Submit.",
    article="**Nota**: Exclusivo para audios en inglés.",
    inputs=[
        gr.Dropdown(
            label="Select Model",
            choices=[
                "tiny.en",
                "base.en",
                "small.en",
                "medium.en",
            ],
            value="base",
        ),
        gr.Audio(label="Upload Audio File", sources=["upload", "microphone"], type="filepath"),
    ],
    outputs=gr.Textbox(label="Whisper Transcript"),
)

gradio_ui.queue().launch()