Introducing Subtitle Formatting and Font Color Options in Whisper

[MatteoFasulo]

This enhancement proposal builds upon the advancements of a previous pull request (PR #1786) and aims to further enhance the capabilities of the Whisper repository by introducing new features for subtitle customization. The primary additions include a subtitle_format option, allowing users to style subtitles with underline, bold, or italicized formatting. Additionally, a font_color option enables users to specify the font color for subtitles using hex values.

The core modification is made to the iterate_result function in utils.py, which now accommodates the new options. When using subtitle_format, users can choose from "u" for underlined, "b" for bolded, and "i" for italicized subtitles. In conjunction with this, the font_color option accepts hex color values (e.g., "#FFFFFF" for white), providing users with the flexibility to personalize the appearance of subtitles.

The default behavior is set to None for subtitle_format and white for font_color if these options are not explicitly specified. This ensures backward compatibility and seamless integration with third-party software like FFMPEG.

To implement these enhancements, users can simply include the desired options in the word_options dictionary when utilizing the iterate_result function. An illustrative example is provided in the usage section, showcasing how to integrate these options into your Whisper workflow.

This proposed enhancement not only enhances the visual appeal of generated subtitles but also empowers users with greater control over the styling aspects, contributing to a more versatile and user-friendly experience with the Whisper library. Your feedback on this enhancement proposal is highly encouraged and welcomed in the discussion section.

Example:

import whisper
from whisper.utils import get_writer 

audio = './audio.mp3'
model = whisper.load_model(model='small')
result = model.transcribe(audio=audio, language='en', word_timestamps=True, task="transcribe")

# Set VTT Line and words width
word_options = {
    "highlight_words": False,
    "max_line_count": 1,
    "max_line_width": 42,
    "subtitle_format": "b", # bold word highlighting
    "font_color": "#fff000", # yellow font color
}
vtt_writer = get_writer(output_format='vtt', output_dir='./')
vtt_writer(result, audio, word_options)