NeuralFalcon's picture
Update app.py
04fc0d0 verified
raw
history blame
14.4 kB
from utils import language_dict
import math
import torch
import gc
import time
from faster_whisper import WhisperModel
import os
import re
import uuid
import shutil
def get_language_name(lang_code):
global language_dict
# Iterate through the language dictionary
for language, details in language_dict.items():
# Check if the language code matches
if details["lang_code"] == lang_code:
return language # Return the language name
return lang_code
def clean_file_name(file_path):
# Get the base file name and extension
file_name = os.path.basename(file_path)
file_name, file_extension = os.path.splitext(file_name)
# Replace non-alphanumeric characters with an underscore
cleaned = re.sub(r'[^a-zA-Z\d]+', '_', file_name)
# Remove any multiple underscores
clean_file_name = re.sub(r'_+', '_', cleaned).strip('_')
# Generate a random UUID for uniqueness
random_uuid = uuid.uuid4().hex[:6]
# Combine cleaned file name with the original extension
clean_file_path = os.path.join(os.path.dirname(file_path), clean_file_name + f"_{random_uuid}" + file_extension)
return clean_file_path
def format_segments(segments):
saved_segments = list(segments)
sentence_timestamp = []
words_timestamp = []
speech_to_text = ""
for i in saved_segments:
temp_sentence_timestamp = {}
# Store sentence information in sentence_timestamp
text = i.text.strip()
sentence_id = len(sentence_timestamp) # Get the current index for the new entry
sentence_timestamp.append({
"id": sentence_id, # Use the index as the id
"text": text,
"start": i.start,
"end": i.end,
"words": [] # Initialize words as an empty list within the sentence
})
speech_to_text += text + " "
# Process each word in the sentence
for word in i.words:
word_data = {
"word": word.word.strip(),
"start": word.start,
"end": word.end
}
# Append word timestamps to the sentence's word list
sentence_timestamp[sentence_id]["words"].append(word_data)
# Optionally, add the word data to the global words_timestamp list
words_timestamp.append(word_data)
return sentence_timestamp, words_timestamp, speech_to_text
def combine_word_segments(words_timestamp, max_words_per_subtitle=8, min_silence_between_words=0.5):
if max_words_per_subtitle<=1:
max_words_per_subtitle=1
before_translate = {}
id = 1
text = ""
start = None
end = None
word_count = 0
last_end_time = None
for i in words_timestamp:
try:
word = i['word']
word_start = i['start']
word_end = i['end']
# Check for sentence-ending punctuation
is_end_of_sentence = word.endswith(('.', '?', '!'))
# Check for conditions to create a new subtitle
if ((last_end_time is not None and word_start - last_end_time > min_silence_between_words)
or word_count >= max_words_per_subtitle
or is_end_of_sentence):
# Store the previous subtitle if there's any
if text:
before_translate[id] = {
"text": text,
"start": start,
"end": end
}
id += 1
# Reset for the new subtitle segment
text = word
start = word_start # Set the start time for the new subtitle
word_count = 1
else:
if word_count == 0: # First word in the subtitle
start = word_start # Ensure the start time is set
text += " " + word
word_count += 1
end = word_end # Update the end timestamp
last_end_time = word_end # Update the last end timestamp
except KeyError as e:
print(f"KeyError: {e} - Skipping word")
pass
# After the loop, make sure to add the last subtitle segment
if text:
before_translate[id] = {
"text": text,
"start": start,
"end": end
}
return before_translate
def custom_word_segments(words_timestamp, min_silence_between_words=0.3, max_characters_per_subtitle=17):
before_translate = []
id = 1
text = ""
start = None
end = None
last_end_time = None
i = 0
while i < len(words_timestamp):
word = words_timestamp[i]['word']
word_start = words_timestamp[i]['start']
word_end = words_timestamp[i]['end']
# Look ahead to check if the next word (i+1) starts with a hyphen
if i + 1 < len(words_timestamp) and words_timestamp[i + 1]['word'].startswith("-"):
# Combine the current word and the next word (i, i+1) if next word starts with a hyphen
combined_text = word + words_timestamp[i + 1]['word'][:] # Skip the hyphen and combine
combined_start_time = word_start
combined_end_time = words_timestamp[i + 1]['end']
i += 1 # Skip the next word (i+1) since it has been combined
# Look ahead for the next non-hyphenated word, check further if needed (i+2, i+3, etc.)
while i + 1 < len(words_timestamp) and words_timestamp[i + 1]['word'].startswith("-"):
combined_text += words_timestamp[i + 1]['word'][:] # Add word excluding hyphen
combined_end_time = words_timestamp[i + 1]['end']
i += 1 # Skip the next hyphenated word
else:
# No hyphen at the next word, just take the current word
combined_text = word
combined_start_time = word_start
combined_end_time = word_end
# Check if the combined text exceeds the maximum character limit
if len(text) + len(combined_text) > max_characters_per_subtitle:
# If accumulated text is non-empty, store it as a subtitle
if text:
before_translate.append({
"word": text.strip(),
"start": start,
"end": end
})
id += 1
# Start a new subtitle with the combined text
text = combined_text
start = combined_start_time
else:
# Accumulate text
if not text:
start = combined_start_time
text += " " + combined_text
# Update the end timestamp
end = combined_end_time
last_end_time = end
# Move to the next word
i += 1
# Add the final subtitle segment if text is not empty
if text:
before_translate.append({
"word": text.strip(),
"start": start,
"end": end
})
return before_translate
def convert_time_to_srt_format(seconds):
""" Convert seconds to SRT time format (HH:MM:SS,ms) """
hours = int(seconds // 3600)
minutes = int((seconds % 3600) // 60)
secs = int(seconds % 60)
milliseconds = int((seconds - int(seconds)) * 1000)
return f"{hours:02}:{minutes:02}:{secs:02},{milliseconds:03}"
def write_subtitles_to_file(subtitles, filename="subtitles.srt"):
# Open the file with UTF-8 encoding
with open(filename, 'w', encoding='utf-8') as f:
for id, entry in subtitles.items():
# Write the subtitle index
f.write(f"{id}\n")
if entry['start'] is None or entry['end'] is None:
print(id)
# Write the start and end time in SRT format
start_time = convert_time_to_srt_format(entry['start'])
end_time = convert_time_to_srt_format(entry['end'])
f.write(f"{start_time} --> {end_time}\n")
# Write the text and speaker information
f.write(f"{entry['text']}\n\n")
def word_level_srt(words_timestamp, srt_path="world_level_subtitle.srt",shorts=False):
punctuation_pattern = re.compile(r'[.,!?;:"\–—_~^+*|]')
with open(srt_path, 'w', encoding='utf-8') as srt_file:
for i, word_info in enumerate(words_timestamp, start=1):
start_time = convert_time_to_srt_format(word_info['start'])
end_time = convert_time_to_srt_format(word_info['end'])
word=word_info['word']
word =re.sub(punctuation_pattern, '', word)
if word.strip() == 'i':
word = "I"
if shorts==False:
word=word.replace("-","")
srt_file.write(f"{i}\n{start_time} --> {end_time}\n{word}\n\n")
def generate_srt_from_sentences(sentence_timestamp, srt_path="default_subtitle.srt"):
with open(srt_path, 'w', encoding='utf-8') as srt_file:
for index, sentence in enumerate(sentence_timestamp):
start_time = convert_time_to_srt_format(sentence['start'])
end_time = convert_time_to_srt_format(sentence['end'])
srt_file.write(f"{index + 1}\n{start_time} --> {end_time}\n{sentence['text']}\n\n")
def get_audio_file(uploaded_file):
global temp_folder
file_path = os.path.join(temp_folder, os.path.basename(uploaded_file))
file_path=clean_file_name(file_path)
shutil.copy(uploaded_file, file_path)
return file_path
def whisper_subtitle(uploaded_file,Source_Language,max_words_per_subtitle=8):
global language_dict,base_path,subtitle_folder
#Load model
if torch.cuda.is_available():
# If CUDA is available, use GPU with float16 precision
device = "cuda"
compute_type = "float16"
# compute_type="int8_float16"
else:
# If CUDA is not available, use CPU with int8 precision
device = "cpu"
compute_type = "int8"
faster_whisper_model = WhisperModel("deepdml/faster-whisper-large-v3-turbo-ct2",device=device, compute_type=compute_type)
audio_path=get_audio_file(uploaded_file)
if Source_Language=="Automatic":
segments,d = faster_whisper_model.transcribe(audio_path, word_timestamps=True)
lang_code=d.language
src_lang=get_language_name(lang_code)
else:
lang=language_dict[Source_Language]['lang_code']
segments,d = faster_whisper_model.transcribe(audio_path, word_timestamps=True,language=lang)
src_lang=Source_Language
sentence_timestamp,words_timestamp,text=format_segments(segments)
if os.path.exists(audio_path):
os.remove(audio_path)
del faster_whisper_model
gc.collect()
torch.cuda.empty_cache()
word_segments=combine_word_segments(words_timestamp, max_words_per_subtitle=max_words_per_subtitle, min_silence_between_words=0.5)
shorts_segments=custom_word_segments(words_timestamp, min_silence_between_words=0.3, max_characters_per_subtitle=17)
#setup srt file names
base_name = os.path.basename(uploaded_file).rsplit('.', 1)[0][:30]
save_name = f"{subtitle_folder}/{base_name}_{src_lang}.srt"
original_srt_name=clean_file_name(save_name)
original_txt_name=original_srt_name.replace(".srt",".txt")
word_level_srt_name=original_srt_name.replace(".srt","_word_level.srt")
customize_srt_name=original_srt_name.replace(".srt","_customize.srt")
shorts_srt_name=original_srt_name.replace(".srt","_shorts.srt")
generate_srt_from_sentences(sentence_timestamp, srt_path=original_srt_name)
word_level_srt(words_timestamp, srt_path=word_level_srt_name)
word_level_srt(shorts_segments, srt_path=shorts_srt_name,shorts=True)
write_subtitles_to_file(word_segments, filename=customize_srt_name)
with open(original_txt_name, 'w', encoding='utf-8') as f1:
f1.write(text)
return original_srt_name,customize_srt_name,word_level_srt_name,shorts_srt_name,original_txt_name
#@title Using Gradio Interface
def subtitle_maker(Audio_or_Video_File,Source_Language,max_words_per_subtitle):
try:
default_srt_path,customize_srt_path,word_level_srt_path,shorts_srt_name,text_path=whisper_subtitle(Audio_or_Video_File,Source_Language,max_words_per_subtitle=max_words_per_subtitle)
except Exception as e:
print(f"Error in whisper_subtitle: {e}")
default_srt_path,customize_srt_path,word_level_srt_path,shorts_srt_name,text_path=None,None,None,None,None
return default_srt_path,customize_srt_path,word_level_srt_path,shorts_srt_name,text_path
import gradio as gr
import click
base_path="."
subtitle_folder=f"{base_path}/generated_subtitle"
temp_folder = f"{base_path}/subtitle_audio"
if not os.path.exists(subtitle_folder):
os.makedirs(subtitle_folder, exist_ok=True)
if not os.path.exists(temp_folder):
os.makedirs(temp_folder, exist_ok=True)
source_lang_list = ['Automatic']
available_language=language_dict.keys()
source_lang_list.extend(available_language)
@click.command()
@click.option("--debug", is_flag=True, default=False, help="Enable debug mode.")
@click.option("--share", is_flag=True, default=False, help="Enable sharing of the interface.")
def main(debug, share):
description = """**Note**: Avoid uploading large video files. Instead, upload the audio from the video for faster processing.
You can find the model at [faster-whisper-large-v3-turbo-ct2](https://huggingface.co/deepdml/faster-whisper-large-v3-turbo-ct2)"""
# Define Gradio inputs and outputs
gradio_inputs = [
gr.File(label="Upload Audio or Video File"),
gr.Dropdown(label="Language", choices=source_lang_list, value="Automatic"),
gr.Number(label="Max Word Per Subtitle Segment [Useful for Vertical Videos]", value=8)
]
gradio_outputs = [
gr.File(label="Default SRT File", show_label=True),
gr.File(label="Customize SRT File", show_label=True),
gr.File(label="Word Level SRT File", show_label=True),
gr.File(label="SRT File For Shorts", show_label=True),
gr.File(label="Text File", show_label=True)
]
# Create Gradio interface
demo = gr.Interface(fn=subtitle_maker, inputs=gradio_inputs, outputs=gradio_outputs, title="Auto Subtitle Generator Using Whisper-Large-V3-Turbo-Ct2",description=description)
# Launch Gradio with command-line options
demo.queue().launch(debug=debug, share=share)
if __name__ == "__main__":
main()