Spaces:
Running
Running
import os | |
import io | |
import gradio as gr | |
import subprocess | |
from google.cloud import speech | |
from google.api_core.client_options import ClientOptions | |
# Obtener la API Key desde las variables de entorno | |
API_KEY = os.getenv("GOOGLE_API_KEY") | |
# Verificar si la API Key está configurada | |
if not API_KEY: | |
raise ValueError("La API Key de Google no está configurada. Configúrala en la variable de entorno GOOGLE_API_KEY.") | |
def convert_to_wav(input_file): | |
"""Convierte archivos de audio a formato WAV LINEAR16 si es necesario.""" | |
output_file = input_file + ".wav" | |
command = [ | |
"ffmpeg", "-y", "-i", input_file, | |
"-acodec", "pcm_s16le", "-ar", "44100", "-ac", "1", output_file | |
] | |
subprocess.run(command, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL) | |
return output_file | |
def transcribe(file_path): | |
"""Transcribe audio a texto usando Google Cloud Speech-to-Text con API Key.""" | |
if file_path is None: | |
return '', '' | |
# Convertir a formato WAV si es necesario | |
if not file_path.endswith(".wav"): | |
file_path = convert_to_wav(file_path) | |
# Configurar el cliente de Speech-to-Text con API Key | |
client_options = ClientOptions(api_key=API_KEY) | |
client = speech.SpeechClient(client_options=client_options) | |
# Configuración de la solicitud | |
config = speech.RecognitionConfig( | |
encoding=speech.RecognitionConfig.AudioEncoding.LINEAR16, | |
sample_rate_hertz=44100, | |
audio_channel_count=1, | |
language_code="es-AR", | |
) | |
# Cargar el audio en binario | |
with io.open(file_path, "rb") as audio_file: | |
content = audio_file.read() | |
audio = speech.RecognitionAudio(content=content) | |
# Realiza la transcripción | |
response = client.recognize(config=config, audio=audio) | |
transcript = [] | |
confidence = [] | |
# Lee la respuesta de la API | |
for result in response.results: | |
confidence.append(str(result.alternatives[0].confidence)) | |
transcript.append(result.alternatives[0].transcript) | |
return ' '.join(transcript), '\n'.join(confidence) | |
# Configuración de la interfaz Gradio con opciones de grabación y subida de archivos | |
output1 = gr.Textbox(label='Transcripción') | |
output2 = gr.Textbox(label='Confianza') | |
demo = gr.Interface( | |
transcribe, | |
[ | |
gr.Audio(sources=["microphone"], type="filepath", label='Grabar audio aquí', streaming=False), | |
gr.File(label="Subir archivo de audio") | |
], | |
[output1, output2], | |
title='Demo Reconocimiento de Voz con Google', | |
description='<p>Grabar o subir un archivo de audio para convertir voz a texto usando Google Cloud Speech-to-Text.</p>' | |
) | |
demo.launch() | |