app.py

import spaces
import torch
import gradio as gr
from transformers import pipeline
from transformers.pipelines.audio_utils import ffmpeg_read

MODEL_NAME = "openai/whisper-large-v3"
BATCH_SIZE = 8
FILE_LIMIT_MB = 1000

device = 0 if torch.cuda.is_available() else "cpu"

pipe = pipeline(
    task="automatic-speech-recognition",
    model=MODEL_NAME,
    chunk_length_s=30,
    device=device,
)

@spaces.GPU
def audio_transcribe(inputs, task):
    if inputs is None:
        raise gr.Error("No audio file submitted! Please upload or record an audio file before submitting your request.")

    text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
    return  text

with gr.Blocks() as transcriberUI:
    gr.Markdown(
    """
    # Ola!
    Clicar no botao abaixo para selecionar o Audio a ser transcrito!
    Ambiente Demo disponivel 24x7. Running on ZeroGPU with openai/whisper-large-v3
    """)
    inp = gr.File(label="Arquivo de Audio", show_label=True, type="file_path", file_count="single", file_types=["mp3"])
    transcribe = gr.Textbox(label="Transcricao", show_label=True, show_copy_button=True)
    inp.upload(audio_transcribe, inp, transcribe)

if __name__ == "__main__":
    transcriberUI.launch()