owoi-audiotoclip 1.0.1

Package to convert audio to video with google speech to text and google image search

OWOI_AudioToClip

Python module used for the school project OWOI (One Word One Image)

Installation

After git cloning the repository, you can install the dependencies with the following command:

poetry install

Credentials

Please provide your credentials in the following environment variables:

export GOOGLE_APPLICATION_CREDENTIALS="path/to/credentials.json"
export GOOGLE_IMAGES_SEARCH_TOKEN="token"
export GOOGLE_SEARCH_ID="id"

Classes

TranscriptFactory

This class is used to create a transcript from a text file. It will create a list of words and a list of timestamps.

from audio_to_clip import ClipMakerFactory

transcript_factory = TranscriptFactory(gcs_uri="gs://bucket/file.mp3")

Methods:

• transcribe_audio_to_text() -> list[dict]: transcribe audio to text from the gcs_uri and returns a list of dict with the following keys: "word", "start_time" and "end_time"
• get_word_timestamps() -> list[dict]: returns a list of dict with the following keys: "word", "start_time" and "end_time"

This Class should be used to create a transcript from a text file before creating a clip with the ClipMakerFactory.

ClipMakerFactory

This class is used to create a clip from a transcript.

from audio_to_clip import ClipMakerFactory

clip_maker_factory = ClipMakerFactory(video_name, username, transcript, gcs_bucket, local_storage, gcs_audio_path)

Params:

• video_name:str -> name of the video
• username:str -> name of the user
• transcript:list[WordTimestamp] -> list of WordTimestamp
• gcs_bucket_dest:str -> name of the gcs bucket destination
• local_storage:str -> path to the local storage destination
• gcs_audio_path:str -> path to the audio file in the gcs bucket

Methods:

• clip_maker(word_timestamps: list[WordTimestamp]) -> VideoFileClip: creates a clip from the transcript and returns a VideoFileClip
• upload_video_to_gcs() -> None: uploads the video to the gcs bucket with the name: username/video_name.mp4, should be called after clip_maker