Notice

This page show a previous version of the article

Contribute audio for Tatoeba

If you are able to provide audio files in mp3 format, with filenames in the format sentence_number.mp3 (where sentence_number represents the number of the matching sentence), then we can add them to the site. Unfortunately, we do not have any tools at the moment that run on all major platforms and save files with the correct audio and filename format. Thus, you must be able to convert the files yourself or send them to someone else who can.

Note that the Shtooka Recorder only runs on Windows platforms older than Windows Vista, and the Windows version of swac-recorder does not allow you to record. Thus, the following instructions will be useful to you only if you can work on Windows XP, Linux, or some other platform that lets you use these tools.

** Note by CK: I wonder if the above it true. If you want to help, please contact me via http://tatoeba.org/eng/private_messages/write/CK and I'll try to help you.

Instructions

1) You need to have a good microphone because we care about sound quality.

2) Download one of these audio recorders:

  • Shtooka recorder:

    • download the installer
    • get familiar with the tool via video tutorials:

      • youtube.com/watch?v=AcJoLBjUOaY (made by AmberShadow).
      • bit.ly/shtooka (made by CK) = video and documentation

or

  • swac-recorder:

    • for Windows (XP, Vista, Win7, Win8), download the 32-bit or 64-bit binary (though as noted above, these may not allow you to actually record audio)

    • for Ubuntu (13.10, 12.04): download the 32-bit or 64-bit binary

    • for Fedora (20, 19), download the source package

3) Pick a few random sentences (just two or three), record them, and send us the samples at team@tatoeba.org, with the title "Audio for Tatoeba in language_name". This way we can evaluate whether the sound quality is good enough.

4) If we tell you that the quality of your audio is good enough, you can go ahead and record audio for more sentences. Verify that all the sentences sound natural to you. If a sentence doesn't sound natural, post a comment on it.

Using Shtooka Recorder to record audio for the site

1) In Shtooka Recorder, copy-paste the final list into the "Words to record", fill in the info in the "Speaker" tab, go back to the "Words" tab and click "Continue".

2) Press the space bar to indicate that you are ready, and read the sentence that is highlighted in red. The software will detect when to start and stop recording, and will jump to the next sentence automatically. All you have to do is read what's highlighted in red. If you want to take a break, you can press space to pause, so that it doesn't record something unrelated. If you need to listen to a sentence's audio, simply select it and press "Enter". (You can navigate with the directional keys: up, down, left, right.)

3) Save the audio files in MP3 format, converting them if necessary. Verify that the files are named properly. They should be named after the id of the sentence. For instance, the audio file for the sentence with id 123 should be named 123.mp3. (Note that we used to ask for the files to be in FLAC format, but this is no longer the case.)

4) If you can upload your files somewhere, please do, and give us the link to download them. If this is not possible, send the files by e-mail to team@tatoeba.org. We will then include your audio in Tatoeba as soon as possible. Eventually, we hope to make it possible to record/upload directly from Tatoeba.

Related link: http://blog.tatoeba.org/2010/04/audio-for-tatoeba-sentences-in.html