Spaces:
Build error
Build error
| import torch | |
| import gradio as gr | |
| from transformers import pipeline | |
| import os | |
| import pandas as pd # Importa o Pandas | |
| import json # Importa o módulo json para ler o arquivo de palavras-chave | |
| MODEL_NAME = "openai/whisper-large-v2" | |
| BATCH_SIZE = 8 | |
| device = 0 if torch.cuda.is_available() else "cpu" | |
| pipe = pipeline( | |
| task="automatic-speech-recognition", | |
| model=MODEL_NAME, | |
| chunk_length_s=30, | |
| device=device, | |
| ) | |
| # Função para carregar palavras-chave do arquivo JSON | |
| def load_keywords(json_path): | |
| with open(json_path, 'r') as file: | |
| data = json.load(file) | |
| keywords = data.get('keywords', []) + data.get('key_words_Telus', []) | |
| return keywords | |
| # Caminho para o arquivo JSON de palavras-chave | |
| keywords_path = os.path.join(os.path.dirname(__file__), "keywords.json") | |
| keywords = load_keywords(keywords_path) | |
| def 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.") | |
| result = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True) | |
| text = result["text"] | |
| # Identifica as palavras-chave na transcrição | |
| found_keywords = [keyword for keyword in keywords if keyword.lower() in text.lower()] | |
| # Cria um DataFrame para armazenar os dados | |
| df = pd.DataFrame({'Transcription': [text], 'Found Keywords': [", ".join(found_keywords)]}) | |
| # Define o caminho onde o arquivo CSV será salvo | |
| csv_save_path = os.path.join(os.path.dirname(__file__), "transcription.csv") | |
| # Salva o DataFrame como um arquivo CSV | |
| df.to_csv(csv_save_path, index=False) | |
| # Retorna o texto da transcrição, as palavras-chave encontradas e o caminho do arquivo CSV | |
| return text, ", ".join(found_keywords), csv_save_path | |
| # Ajusta as interfaces Gradio para incluir a saída das palavras-chave encontradas | |
| mf_transcribe = gr.Interface( | |
| fn=transcribe, | |
| inputs=[ | |
| gr.Audio(type="filepath"), | |
| gr.Radio(choices=["transcribe", "translate"], label="Task"), | |
| ], | |
| outputs=[ | |
| gr.Textbox(label="Transcription Text"), | |
| gr.Textbox(label="Found Keywords", visible=False), # Oculta para a aba de microfone | |
| gr.File(label="Download CSV File", visible=False), # Oculta o download do CSV para a aba de microfone | |
| ], | |
| title="GexelView.AI - Transcriber", | |
| description="Transcribe or translate audio inputs with a single click.", | |
| allow_flagging="never", | |
| ) | |
| file_transcribe = gr.Interface( | |
| fn=transcribe, | |
| inputs=[ | |
| gr.Audio(type="filepath", label="Audio file"), | |
| gr.Radio(choices=["transcribe", "translate"], label="Task"), | |
| ], | |
| outputs=[ | |
| gr.Textbox(label="Transcription Text"), | |
| gr.Textbox(label="Found Keywords"), # Exibe as palavras-chave encontradas | |
| gr.File(label="Download CSV File"), # Torna o download do arquivo CSV visível para a aba de arquivo | |
| ], | |
| title="GexelView.AI - Transcriber", | |
| description="Transcribe or translate audio files with a single click.", | |
| allow_flagging="never", | |
| ) | |
| demo = gr.Blocks() | |
| with demo: | |
| gr.TabbedInterface([mf_transcribe, file_transcribe], ["Microphone", "Audio file"]) | |
| demo.launch() |