vikit.prompt.recorded_prompt_subtitles_extractor¶

Overview¶

Classes¶
`RecordedPromptSubtitlesExtractor`	A class to extract subtitles from a sound recording,

Classes¶

class vikit.prompt.recorded_prompt_subtitles_extractor.RecordedPromptSubtitlesExtractor¶

Bases: vikit.prompt.subtitle_extractor.SubtitleExtractor

A class to extract subtitles from a sound recording, merge short subtitles into longer ones, or extract them as text tokens

Here is how we do it: - We extract the audio slice from the audio file and cap the duration of the slice to x seconds so that replicate can process it (today n = 300 seconds, i.e. 5minutes) - We save the generated subtitles to a temporary SRT file - We concatenate the temporary SRT files to a prompt wide srt file

Yes this is kind of hacky, but it works for now.

Overview

Methods¶
`extract_subtitles_async`(recorded_prompt_file_path, ml_models_gateway)	Generate subtitles from a recorded audio file

Members

async extract_subtitles_async(recorded_prompt_file_path, ml_models_gateway: vikit.gateways.ML_models_gateway.MLModelsGateway = None)¶: Generate subtitles from a recorded audio file

inputs: NA returns: Subtitle Rip File object