sourcetoTarget.py

# Transformers installation
# pip install transformers

import os
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer


# define tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("facebook/nllb-200-distilled-600M")
model = AutoModelForSeq2SeqLM.from_pretrained(
        "facebook/nllb-200-distilled-600M")


# translation function for single sentence
def translateSentence(sentence):
    inputs = tokenizer(sentence, return_tensors="pt")
    translated_tokens = model.generate(
        **inputs, forced_bos_token_id=tokenizer.lang_code_to_id["deu_Latn"], max_length=400
    )
    translatedSentence = tokenizer.batch_decode(
        translated_tokens, skip_special_tokens=True)[0]
    return translatedSentence

def translateAllTranscripts():

    # source text directory
    directory = os.fsencode("whisper")

    # go through source text files
    # translate them to target language (German)
    #  save translated text files to translatedTranscripts file
    for file in os.listdir(directory):
        filename = os.fsdecode(file)
        print(f'reading {filename}')
        if (filename.endswith(".txt")):
            file_path = f"whisper/{filename}"
            with open(file_path, 'r') as f:
                sourceText = f.read()  # read source text
                sentences = sourceText.split('.')
                translationArr = []
                for sentence in sentences:
                    #  translate each sentence
                    translatedSentence = translateSentence(sentence)
                    translationArr.append(translatedSentence)
                # join translated sentences
                translation = " ".join(translationArr)
                # save translated transcript
                fname = "translatedTranscripts/" + filename
                with open(fname, "w", encoding="UTF8") as ft:
                    ft.write(translation)
                    print("written to: " + fname)


def englishToGerman(videoId):
    file_path = f"whisper/{videoId}.txt"
    with open(file_path, 'r') as f:
        sourceText = f.read()  # read source text
        sentences = sourceText.split('.')
        translationArr = []
        for sentence in sentences:
            #  translate each sentence
            translatedSentence = translateSentence(sentence)
            translationArr.append(translatedSentence)
        # join translated sentences
        translation = " ".join(translationArr)
        # save translated transcript
        fname = f"translatedTranscripts/{videoId}.txt" 
        with open(fname, "w", encoding="UTF8") as ft:
            ft.write(translation)
            print("written to: " + fname)