Mastering Pronunciation

Tweaking the robot’s speech for a natural prosody with TTS (text to speech)
Mastering Pronunciation

What the lesson is about

Speech is the quickest and easiest way to give and receive information, and is therefore one of the most important means of communication for Pepper and NAO. As humanoid robots, they are expected to be able to speak flawlessly, so any error will definitely be noticed.
In this lesson we are going to learn how to get the pronunciation and prosody right to make the robot’s speech more natural.

What you will learn

You are going to learn how to modify Text to Speech (TTS) to correct pronunciation errors and make the prosody more natural by changing the following:

TTS tags:

  • Pause
  • Volume
  • Speed
  • Pitch
  • Reset
  • Emphasis
  • Prosodic boundary
  • End of sentence

What you will need

Warning about robot version and language restrictions

Unless otherwise stated, it means that sections are applicable to NAOqi 2.1 to 2.8. Version 2.9 is an exception, which is covered further in this lesson (Japanese only).
See the Tweaking in 2.9 section in Advanced Tweaks (step2)

The Japanese language package is from a different provider to all other languages, so some of the section will not be applicable to the Japanese language package. Some others are restricted to a few languages only, so be sure to check the restrictions before trying out the tools of each section.

Applicable use cases

  • Any use case that requires the robot to speak is applicable for this lesson, which means almost all of them. For example:

    • Welcoming clients in shops and malls
    • Making speeches on stage for various events
    • Showcasing a given brand
    • Answering specific Q&A
  • Another representative use case is the dialog part of the Basic Channel, called Basic Dialog: a non-proactive dialog enabling the robot to give technical informations, answer small talk and questions about himself, move on command, and launch applications.

  • And obviously any previously identified business use cases:

    • Host
    • Brand Ambassador
    • Advisor
    • Sales Associate
    • Promoter
    • Service Provider
    • Survey Conductor
    • Entertainer