Spaces:
Sleeping
Sleeping
import streamlit as st | |
import os | |
import tempfile | |
import whisper | |
import speech_recognition as sr | |
from pydub import AudioSegment | |
from audio_recorder_streamlit import audio_recorder | |
# Function to convert mp3 file to wav | |
def convert_mp3_to_wav(mp3_path): | |
audio = AudioSegment.from_mp3(mp3_path) | |
wav_path = mp3_path.replace('.mp3', '.wav') | |
audio.export(wav_path, format="wav") | |
return wav_path | |
# Function to transcribe audio using OpenAI Whisper | |
def transcribe_whisper(model_name, file_path): | |
model = whisper.load_model(model_name) | |
result = model.transcribe(file_path) | |
return result["text"] | |
# Function to transcribe audio using Google Speech API | |
def transcribe_speech_recognition(file_path): | |
r = sr.Recognizer() | |
with sr.AudioFile(file_path) as source: | |
r.adjust_for_ambient_noise(source, duration=0.5) # Adjust ambient noise threshold | |
audio = r.record(source) | |
try: | |
result = r.recognize_google(audio, language='spanish') | |
return result | |
except sr.UnknownValueError: | |
return "No se pudo reconocer ning煤n texto en el audio." | |
# Function to perform transcription based on selected method | |
def perform_transcription(transcription_method, model_name, audio_path): | |
if transcription_method == 'OpenAI Whisper': | |
return transcribe_whisper(model_name, audio_path) | |
else: | |
return transcribe_speech_recognition(audio_path) | |
# Function to handle uploaded file transcription | |
def handle_uploaded_file(uploaded_file, transcription_method, model_name): | |
file_details = {"FileName": uploaded_file.name, "FileType": uploaded_file.type, "FileSize": uploaded_file.size} | |
st.write(file_details) | |
# Save uploaded file to temp directory | |
os.makedirs("temp", exist_ok=True) # Create temp directory if it doesn't exist | |
file_path = os.path.join("temp", uploaded_file.name) | |
with open(file_path, "wb") as f: | |
f.write(uploaded_file.getbuffer()) | |
with st.spinner('Transcribiendo...'): | |
if uploaded_file.name.endswith('.mp3') and transcription_method != 'OpenAI Whisper': | |
# Convert mp3 to wav if Google Speech API is selected and file is in mp3 format | |
file_path = convert_mp3_to_wav(file_path) | |
# Perform transcription | |
transcript = perform_transcription(transcription_method, model_name, file_path) | |
st.text_area('Resultado de la Transcripci贸n:', transcript, height=200) | |
def main(): | |
st.title('Transcriptor de Audio') | |
# Choose the transcription method and model | |
option = st.selectbox('Escoger Modelo de Transcripci贸n', ('Subir un archivo', 'Grabar audio en tiempo real')) | |
transcription_method = st.selectbox('Escoge el m茅todo de transcripci贸n', ('OpenAI Whisper', 'Google Speech API')) | |
if transcription_method == 'OpenAI Whisper': | |
model_name = st.selectbox('Escoge el modelo de Whisper', ('base', 'small', 'medium', 'large', 'tiny')) | |
if option == 'Subir un archivo': | |
uploaded_file = st.file_uploader("Sube tu archivo de audio para transcribir", type=['wav', 'mp3']) | |
if uploaded_file is not None: | |
handle_uploaded_file(uploaded_file, transcription_method, model_name) | |
elif option == 'Grabar audio en tiempo real': | |
duration = 5 | |
# duration = st.slider("Selecciona la duraci贸n de la grabaci贸n (segundos)", 1, 10, 5) | |
# st.write("Duraci贸n de la grabaci贸n:", duration, "segundos") | |
audio_bytes = audio_recorder(pause_threshold=duration, sample_rate=16_000) | |
if audio_bytes: | |
st.write("Grabaci贸n finalizada. Transcribiendo...") | |
with st.spinner('Transcribiendo...'): | |
# Save recorded audio to a temporary file | |
with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as temp_audio: | |
temp_path = temp_audio.name | |
temp_audio.write(audio_bytes) | |
# Perform transcription | |
transcript = perform_transcription(transcription_method, model_name, temp_path) | |
st.text_area('Resultado de la Transcripci贸n:', transcript, height=200) | |
if __name__ == "__main__": | |
main() | |