movis.contrib.voicevox.make_timeline_from_voicevox#

movis.contrib.voicevox.make_timeline_from_voicevox(audio_dir: str | PathLike, max_text_length: int = 25, extra_columns: tuple[tuple[str, Hashable], ...] = (('slide', 0), ('status', 'n'))) → DataFrame[source]#

Create a pandas.DataFrame based on text files generated by Voicevox.

This function reads .txt files from a directory generated by Voicevox and constructs a DataFrame. Each row of the DataFrame corresponds to a text file and includes columns such as 'character', 'hash', and 'text', as well as any additional columns specified in extra_columns.

Args:
audio_dir:

The directory containing .txt files generated by Voicevox.

max_text_length:

The maximum length of text for each entry in the 'text' column. Strings longer than max_text_length are automatically line-breaked for subtitling convenience. Defaults to 25.

extra_columns:

Additional columns to be added to the DataFrame. Each inner tuple should contain a column name and a default value. Defaults to (("slide", 0), ("status", "n")). For example, extra_columns=(("slide", 0), ("status", "n")) will add two columns named 'slide' and 'status' with default values 0 and 'n', respectively.

Note

The 'character' column is automatically added to the DataFrame based on the filename. For example, if the filename is '001_ずんだもん（ノーマル）.txt', the 'character' column will be set to 'zunda'.

Note

The 'hash' column is automatically added to the DataFrame based on the content of the text file. This is used in merget_timeline to detect changes in the text file.

Returns:

A DataFrame with columns that may include ‘character’, ‘hash’, ‘text’, and any additional columns.

  • character: The identifier for the Voicevox character.

  • hash: A hash prefix based on the content of the text file.

  • text: The text content, potentially divided into multiple parts if it exceeds max_text_length.

  • Any additional columns specified in extra_columns.

Examples:
>>> from movis.contrib.voicevox import make_timeline_from_voicevox
>>> make_timeline_from_voicevox('audio')
character    hash                                      text  slide status
0     zunda  9bc5d6         こんにちは！このサンプルでは、簡単なずんだもん動画\nを作ります。      0      n
1     metan  ca647c                  このような形で対談を行うのが解説動画の特徴です。      0      n
2     zunda  75d695    先生役と生徒役で対談することで、内容をわかりやすく\n視聴者に伝えられます。      0      n
3     metan  4d880b          また疑問を生徒側が投げかけることで、より深い解説を\nスムーズに      0      n
4     metan  e9e4f0                        伝えられるのも利点と言えるでしょう。      0      n
5     zunda  702873  欠点としては、現状のところ深い内容を解説する動画に\nは作られていないことです。      0      n
6     zunda  4a1a86   手間がかかることが理由の一つかもしれませんが、この\nようなライブラリを通して      0      n
7     zunda  cb26fc     より多くのユーザー層へ、専門的な知識を伝えられるこ\nとを期待しています。      0      n