app.py

import gradio as gr
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import torch

# Cargar el modelo y el procesador
model = Wav2Vec2ForCTC.from_pretrained("openai/whisper-large-v2")
processor = Wav2Vec2Processor.from_pretrained("openai/whisper-large-v2")

def asr(audio_file_path):
    # Cargar archivo de audio
    input_audio, _ = librosa.load(audio_file_path, sr=16000)

    # Preprocesar audio
    input_values = processor(input_audio, return_tensors="pt", sampling_rate=16000).input_values

    # Realizar inferencia
    logits = model(input_values).logits

    # Decodificar los logits a texto
    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.decode(predicted_ids[0])

    return transcription

# Crear interfaz de Gradio
iface = gr.Interface(fn=asr, inputs=gr.inputs.Audio(source="microphone", type="file"), outputs="text")
iface.launch()