The automatic annotation and analysis of speech

Before using SPPAS...

"Before anything else, preparation is the key to success." A-.G. Bell

Use left and right arrow keys to browse the slides!

Corpus

Robust and reliable corpus creation method suited for SPPAS

Files

Properly prepare your data for the automatic annotations

The important points:

  • Strictly the same name for the audio file, the video file and the transcription file, except the pattern+extension
  • Annotation files: UTF-8 encoding only
  • Audio files are limited to WAV mono (e.g. ONE CHANNEL)
    • do NOT convert a lossy compressed file like mp3 into WAV!!!
  • Video files are limited to the containers mp4, avi, mov, mkv. Notice that not all codecs are supported.

Some of the SPPAS automatic annotations

Use left and right arrow keys to browse the slides!

Speech seg.

The steps to get phonemes and words automatic segmentation

Repetitions

Understanding the other-repetitions auto. detection

Step-by-step corpus annotation process

Use left and right arrow keys to browse the slides!

Annotate with SPPAS

A solution example of corpus annotation process

Others

The latest versions: SPPAS 4.x

Do not except for tutorials to learn how to use the SPPAS Graphical User Interface (GUI).

If you have understood the annotation process, you will have no trouble taking control of the GUI: like in any other GUI, there are buttons, click on them and you will see what's happening. There's no risk to try it!

On the other hand, if you don’t understand what you need to do, look at the tutorials above.

Trick question: How can you analyze annotations if you do not understand how they are obtained?

Slide shows of SPPAS 2.x

All the features of the SPPAS current version were already in SPPAS 2.x. The main principles on how to use SPPAS and the recommendations are still valid. However, new features are constantly added and the Graphical User Interface was entirely changed.

Click left-right keys to browse the slides...

See these tutorials →