Overview

HPE IDOL Speech Server allows you to process training data audio files and labels in such a way that any sensitive or private information is hidden, so that the data is available for DNN model training.

To obfuscate training data, you must carry out the following steps:

  1. Normalize the transcription files.
  2. Create a language model based on the normalized transcription files.
  3. Perform speech-to-text using the latest language pack and the transcript language model that you built in step 2.
  4. Convert the audio files into acoustic feature files.
  5. Align the transcripts.
  6. Run audio analysis.
  7. Perform data obfuscation. The DataObfuscation task takes the files that you specify in the action parameters, and uses them to produce the obfuscated and randomized training data.
NOTE:

For this procedure, your HPE IDOL Speech Server license must include the amadaptadddata module.


_HP_HTML5_bannerTitle.htm