search

readbeyond / aeneas

aeneas is a Python/C library and a set of tools to automagically synchronize audio and text (aka forced alignment)
http://www.readbeyond.it/aeneas/
GNU Affero General Public License v3.0
2.45k stars 218 forks source link

Not working for audio files of longer lengths #243

Closed sidvud98 closed 3 years ago

sidvud98 commented 4 years ago

When I tried to create .srt file from an audiobook and text file, the subtitles and audio went out of sync after 20 minutes or so. I tried multiple times with no use.

Details about the input files- audiobook : mp3 of 5 hours 27 minutes length text File : Obtained by conversion from .mobi using calibre The command I used: python -m aeneas.tools.execute_task some_audioBook.mp3 The_ebook_in_text.txt "task_language=eng|is_text_type=plain|os_task_file_format=srt" subtitles.srt

ps. Of course I cleaned the just-converted text file for getting rid of Contents, Acknowledgements, About the author,etc. My input text file to aenas only contains the words in mp3 file. Nothing more, nothing less.

pettarin commented 4 years ago

Are you using Windows? aeneas on Windows is not officially supported, and it has bugs, for example you might be affected by https://github.com/readbeyond/aeneas/issues/190

sidvud98 commented 4 years ago

No, I'm using Ubuntu 18.04. Can you check out why that's happening?

kdorichev commented 4 years ago

I have a similar result: SRTs are out of sync on audio files after 10th minute -- subtitles tend to start not on the phrase start, but on the end of the previous phase. On shorter audios it seems better.

pettarin commented 4 years ago

I can have a look, if you provide me with the audio, text, and the command(s) you are running. Send me a link to an FTP / Dropbox / OneDrive / etc. URL where I can download a ZIP file containing all the necessary files to reproduce.

Have you tried using the aeneas Web application https://aeneasweb.org/ ? If you get the same results, then there is an issue in aeneas, otherwise it is a problem in your setup/settings (I have aligned up to 2 hours of audio).

readbeyond commented 3 years ago

Closing due to inactivity.