🎬 Auto Subtitle Generator
If you have a large video, upload the audio instead, it's much faster to upload.
😇 Run on Google Colab
""")
with gr.Row():
with gr.Column():
upload_media = gr.File(label="Upload Audio or Video File")
input_lang = gr.Dropdown(label="Source Language", choices=source_lang_list, value="English")
generate_btn = gr.Button("🚀 Generate Subtitle", variant="primary")
with gr.Accordion("⚙️ Translation using Google Translator", open=False):
output_lang = gr.Dropdown(label="Translate Into", choices=target_lang_list, value="English")
with gr.Column():
default_srt = gr.File(label="🎯 Original Subtitles (Default Generated by whisper)")
customized_srt = gr.File(label="📝 Readable Subtitles (Multi-line)")
word_level_srt = gr.File(label="🔠 Word-by-Word Subtitles")
with gr.Accordion("🌍 Other Subtitle Formats", open=False):
translated_srt = gr.File(label="🌐 Translated Subtitles")
shorts_srt = gr.File(label="📱 Shorts/Reels Subtitles")
transcript_txt = gr.File(label="📄 Full Transcript (Text File)")
subtitle_json= gr.File(label="📄 Full Transcript (JSON File) To make .ass file")
word_json= gr.File(label="📄 Shorts Transcript (JSON File) To make .ass file")
transcript_box = gr.Textbox(label="🗒️ Transcript Preview", lines=4,show_copy_button=True)
generate_btn.click(
fn=subtitle_maker,
inputs=[upload_media, input_lang, output_lang],
outputs=[default_srt, translated_srt, customized_srt, word_level_srt, shorts_srt, transcript_txt, subtitle_json,word_json,transcript_box]
)
input_lang.change(
fn=update_target_lang,
inputs=input_lang,
outputs=output_lang
)
# Add example runs (optional)
# gr.Examples(
# examples=[
# ["sample.mp3", "English", "English"],
# ["sample.wav", "English", "Hindi"],
# ],
# inputs=[upload_media, input_lang, output_lang]
# )
return demo
def prompt_translation(language):
"""
Generates a dubbing-friendly translation prompt for an .srt subtitle file.
Tailored for natural speech and timing accuracy.
"""
prompt = f"""
-------------- You are a professional subtitle translator for **video dubbing**.
Translate the following `.srt` subtitle file into **{language}** while preserving timing, meaning, and emotional tone.
Output in JSON format exactly like this:
```json
{{
"subtitle sequence number": {{
"timestamp": "original timestamp",
"actual subtitle text": "original English subtitle line",
"dubbing": "natural, dubbing-friendly {language} translation"
}}
}}
```
**Guidelines for Translation:**
1. **Understand the full context** before translating — read the entire subtitle file first.
2. Translate into **natural, conversational {language}**, not a direct word-for-word translation.
6. Keep translations **roughly similar in length** to the original so lip movements sync naturally.
"""
return prompt
def prompt_fix_grammar(language="English"):
"""
Generates a dubbing-friendly grammar correction prompt for an .srt subtitle file.
Tailored for natural speech and timing accuracy.
"""
prompt = f"""
-------------- You are a professional subtitle editor for **video dubbing**.
Fix the grammar, spelling, and awkward phrasing in the following `.srt` subtitle file while preserving timing, meaning, and emotional tone.
Do NOT translate — keep everything in {language}.
Output in JSON format exactly like this:
```json
{{
"subtitle sequence number": {{
"timestamp": "original timestamp",
"actual subtitle text": "original {language} subtitle line",
"dubbing": "natural, dubbing-friendly corrected {language} line"
}}
}}
```
**Guidelines for Grammar Fixing:**
1. **Understand the full context** before editing — read the entire subtitle file first.
2. Correct grammar, spelling, and phrasing errors while keeping the same meaning.
4. Keep corrections **roughly similar in length** to the original so lip movements sync naturally.
"""
return prompt
def prompt_srt_to_romanized(language="Hindi"):
"""
Generates a prompt for converting a .srt subtitle file
from any language to a Romanized (Latin letters) version,
preserving timing, meaning, punctuation, and formatting.
"""
prompt = f"""
-------------- You are a professional subtitle editor tasked with converting subtitles to Romanized text.
Your task is to convert a `.srt` subtitle file from {language} to **Romanized {language}**,
keeping everything exactly the same except using Latin letters for all words.
**Instructions:**
1. Preserve the original timestamp of each subtitle.
2. Keep the original meaning, punctuation, and formatting intact.
3. Convert **only the original subtitle text** to Roman letters, word by word.
4. Do not add, remove, or change any words.
5. Output in strict JSON format exactly like this:
```json
{{
"subtitle sequence number": {{
"timestamp": "original timestamp",
"original subtitle text": "original {language} subtitle line",
"dubbing": "Romanized, {language} line of original subtitle text"
}}
}}
````
Focus entirely on **accurate Romanization**; do not modify anything else.
"""
return prompt
import pysrt
def prompt_maker(srt_path, target_language, task="Translation"):
txt_path = srt_path.replace(".srt", ".txt")
subs = pysrt.open(srt_path, encoding='utf-8')
with open(txt_path, 'w', encoding='utf-8') as f:
for sub in subs:
f.write(f"{sub.index}\n")
f.write(f"{sub.start} --> {sub.end}\n")
f.write(f"{sub.text}\n\n")
if task == "Translation":
f.write(prompt_translation(target_language))
if task=="Romanization":
f.write(prompt_srt_to_romanized(target_language))
else:
f.write(prompt_fix_grammar(target_language))
with open(txt_path, 'r', encoding='utf-8') as f:
content = f.read()
# return both prompt text and original path
return content, srt_path
import pysrt
import json
import os
def json_to_srt(json_script, srt_path):
"""
Convert dubbing-friendly JSON back into .srt
Uses original srt_path to name output