31 lines
764 B
Python
31 lines
764 B
Python
import os
|
|
import glob
|
|
import datetime
|
|
|
|
import whisper
|
|
|
|
from datahandler import DataHandler
|
|
|
|
|
|
class Transcriber:
|
|
def __init__(self, model_name="base"):
|
|
self.model_name = model_name
|
|
self.DataHandler = DataHandler()
|
|
|
|
def load_model(self):
|
|
self.model = whisper.load_model(name=self.model_name, download_root="models/")
|
|
|
|
def transcribe(self, audio_file):
|
|
self.transcription = self.model.transcribe(
|
|
audio=audio_file,
|
|
verbose=True,
|
|
fp16=False)
|
|
self.raw_text = self.transcription["text"]
|
|
|
|
def persist(self):
|
|
self.DataHandler.save_text_as_file(self.raw_text)
|
|
|
|
def cleanup(self):
|
|
self.DataHandler.get_audio_files()
|
|
self.DataHandler.remove_audio_files()
|