The Elements of Voice First Style

The Elements of Voice First Style

by Ahmed Bouzid, Weiye Ma
Released May 2022
Publisher(s): O'Reilly Media, Inc.
ISBN: 9781098119591

Read it now on the O’Reilly learning platform with a 10-day free trial.

O’Reilly members get unlimited access to books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.

Buy on Amazon Buy on ebooks.com
Start your free trial

Book description

If you're a new or experienced designer of conversational voice first experiences, this handy reference provides actionable answers to key aspects of eyes-busy, hands-busy, voice-only user interfaces. Designed as a companion to books about conversational voice design, this guide includes important details regarding eyes-free, hands-free, voice-only interfaces delivered by Amazon Echo, Google Nest, and a variety of in-car experiences.

Authors Ahmed Bouzid and Weiye Ma provide far-field voice best practices and recommendations in a manner similar to The Elements of Style, the popular American English writing style guide. Like that book, The Elements of Voice First Style provides direct, succinct explanations that focus on the essence of each topic. You'll find answers quickly without having to spend time searching through other sources.

With this guide, you'll be able to:

  • Craft just the right language to enable your voicebot to effectively communicate with humans
  • Create conversational voice interfaces that are robust enough to handle errors and failures
  • Design highly usable conversational voice interfaces by paying attention to small details that can make or break the experience
  • Build a design for a voice-only smart speaker that doesn't require customers to use their eyes or hands

Publisher resources

View/Submit Errata

Table of contents

  1. Preface
    1. Who Should Read This Book?
    2. Why We Wrote This Book
    3. Navigating This Book
    4. O’Reilly Online Learning
    5. How to Contact Us
    6. Acknowledgments
  2. Introduction
  3. 1. Why Voice First
    1. Eyes-Free
    2. Hands-Free
    3. Ephemerality
    4. Wealth
    5. Passivity
    6. Minimal Effort
    7. Broadcasting
    8. Nonliteracy
  4. 2. When Voice First
    1. Environment
    2. Content
    3. User State
      1. Physical State
      2. Competency
      3. Availability
      4. Willingness
    4. Channels
    5. Some Scenarios
      1. Scenario 1
      2. Scenario 2
      3. Scenario 3
  5. 3. Why Voice First Automation
    1. Reduce Costs
    2. Handle Spikes
    3. Increase Customer Satisfaction
    4. Increase Agent Satisfaction
    5. Increase Revenue
    6. Enable Personalization
    7. Facilitate Task Completion
    8. Secure Privacy
    9. Increase Security
  6. 4. The Three Core Characteristics of the VUI
    1. Time Linearity
    2. Unidirectionality
    3. Invisibility
  7. 5. The Elements of Conversation
    1. The Ontology of Conversations
      1. Participant
      2. Statement
      3. Turn
      4. Conversation
    2. The Conversational Actions
    3. The Conversational States
    4. The Internal Conversational Context
    5. Conversational Signaling
      1. Signaling States
      2. Signaling Transitions
  8. 6. The Rules of Conversation
    1. The Cooperative Principle
    2. The Maxim of Quality
    3. The Maxim of Quantity
    4. The Maxim of Relevance
    5. The Maxim of Manner
  9. 7. The Basic Tenets
    1. The Voicebot Is Not a Human
    2. The Voicebot Should Be as Smart as the Data It Has—and No More!
    3. The Voicebot Should Be Consistent
    4. The Voicebot Should Be Transparent
    5. The Voicebot Should Behave Respectfully
  10. 8. The Extra-Conversational Context
    1. The State of the User
      1. Emotional State
      2. Linguistic Competence
      3. Level of Familiarity
      4. Technical Savviness
      5. Physical Readiness
      6. Task-Relevant Properties
    2. The Physical Context
    3. The Social Context
    4. The Recent Context
    5. User Patterns
    6. User-Base Patterns
  11. 9. The UI Use Case Fit
    1. An Illustrative Use Case
    2. Basic Heuristics
  12. 10. The Elements of Starting
    1. Be Brief
    2. Use an Audio Icon
    3. Drop the “Welcome to…”
    4. Never Ever Say, “Please Listen Carefully as Our Options Have Changed”
    5. Have the Voicebot Refer to Itself in the First Person
    6. Drop “You Can Interrupt Me at Any Time”
    7. Keep the Origination Context in Mind
    8. Remember the User’s Preferences
    9. Anticipate User-Specific Requests
    10. Anticipate General User-Base Requests
  13. 11. The Elements of Prompting
    1. Prompt Types
    2. Writing Effective Prompts
      1. Use Language That Is Commonly Used in Conversations
      2. Remember That the User Will Mimic the Voicebot
      3. Unless It’s Essential to the Use Case, Don’t Use Slang or Jargon
      4. Put the Most Important Information First
      5. Use Want Instead of Wish
      6. Avoid Using Speak
      7. Use Contractions
      8. Be Consistent in Your Wording
      9. Avoid Mixing Recorded and TTS Speech
      10. It’s OK for a Sentence to End in a Preposition
      11. Avoid Using Whom
      12. Minimize the Use of Please
      13. Use Incremental Prompts When Dealing with Expert Users
      14. Use Tapering Prompts to Minimize on Wordy Repetitions
      15. Request an Explicit Confirmation Only When Necessary
  14. 12. Choices
    1. Present the Most Requested Items First
    2. Keep the Menu List to Three Items or Less
    3. Keep the Menu Depth to Three Levels or Less
    4. Avoid the Construction of “for/to X, Say X; for/to Y, Say Y; for/to Z, Say Z”
    5. Don’t Use, “Please Select from the Following Options”
    6. Use the Same Part of Speech/Clausal Form When Listing Menu Options
    7. Let Users Ask, “What Are My Choices?”
    8. Let Users “Climb Back” the Menu
    9. Offer to Repeat the Menu Options After a 3-Second Pause
    10. Turn on Barge-In for Expert Users
    11. Include and Teach Shortcuts
  15. 13. Managing Failure
    1. Types of Failure
      1. No-Input
      2. No-Match
      3. Misrecognition
      4. System Failure
    2. Causes of Failure
      1. No-Input
      2. No-Match
      3. Misrecognition
      4. System
    3. Best Practices
      1. Always Have the Voicebot Take the Blame
      2. Give the User Three Chances
      3. Offer Explicit Examples of How to Respond
      4. Be Careful When You Reprompt
      5. Establish “Safety Points”
      6. Never Terminate a Conversation Unilaterally—Especially During Recovery
      7. Don’t Be Repetitive During Recovery
      8. Orient the User About Where They Are
      9. Give the User Information About the Issue
      10. Do Not Be Overly Apologetic
  16. 14. Help Strategies
    1. Tell the User That Help Is Available
    2. Detect When the User Needs Help
    3. Structure Your Help
      1. Mention the Basic Task the Voicebot Is Trying to Solve
      2. Offer Help to the Most Frequently Encountered Problems First
    4. Return from Where You Left Off After Giving Help
    5. Be Concise and Specific with Your Help
    6. Use Context to Guide Your Explanations
    7. Illustrate Your Explanations with Examples
    8. Offer Help Only When It Is Needed
  17. 15. Verbal Dialogue Marking
    1. Acknowledge Receipt of Information
    2. Announce That the User Is About to Receive Some Information
    3. Mark Sequences
    4. Mark the Beginning and End of a Section
    5. Mark Failures
    6. Show Light at the End of the Tunnel
    7. Indicate Implicitly That the Voicebot Still Owns the Turn
    8. Tell the User Explicitly That They Are Being Placed on Hold
    9. Don’t Repeat the Same Marker Twice in a Row
    10. Pay Attention to the Markers After a Failure Strategy
  18. 16. Nonverbal Dialogue Marking
    1. Types of Nonverbal Audio
    2. Opening the Dialogue
    3. Signaling That It’s the User’s Turn to Speak
    4. Signaling That the Voicebot Is Busy Doing Something and Is Holding the Turn
    5. Waiting for the User to Give an Answer
    6. After a No-Input
    7. Announcing a List of Choices
    8. Entering a New Section
    9. Marking Transition from One List Item to the Next
    10. Announcing Help
    11. Ending the Conversation
  19. 17. Language Design
    1. On “Naturalness”
    2. Key Terms
    3. Designing an Effective Language Model
    4. Clearly Define the Problems That Your Voicebot Can Help the User With
    5. Communicate Why the Voicebot Exists and What It Can Help the User Do Outside of the Voicebot
    6. Spend Time Building a Clean Ontology
    7. Do Not Design Your Language from the Armchair
    8. Go Explicit When Recovering from a Language Error
  20. 18. On Silence
    1. Prior to Listing Options
    2. Between Options in a Menu List
    3. Between Categories of Options
    4. When Interacting with Power Users
    5. After Echoing
    6. Before and After TTS Prompts
  21. 19. The Elements of Closing
    1. Allow the Users to Explicitly End the Dialogue
    2. Allow the User to Request a Human
    3. When the User Has to Wait, Provide a Waiting Time Estimate
    4. Provide the Option to Cancel a Transfer to a Human
    5. Keep the “While-You-Wait” Audio Relevant
    6. Understand the User’s State of Mind When You Play the “While-You-Wait” Audio
    7. Never Say, “Your Call Is Important to Us”
    8. Don’t Make the User Repeat to the Human Information They Provided to the Voicebot
    9. Make the Human Agent Aware That the Customer Was Interacting with the Voicebot
    10. Avoid Transferring Users from One Voicebot to Another
    11. Don’t Play Phone Rings Unless You Are Transferring Directly to a Human
    12. Reassure Users of Success
    13. Don’t Provide Any Crucial New Information
    14. Give the User a Quick Tip
    15. Offer to Reach Back
  22. 20. Voice First Notifications
    1. Fundamental Considerations
    2. Key Notification Attributes
      1. Notification Urgency
      2. Notification Content
      3. Delivery Context
    3. Key Form Factors
      1. Far-Field Voice
      2. Near-Field Voice
      3. In-Ear Voice
    4. Some Best Practices
      1. Be Mindful of the Receiver’s Time Zone
      2. Provide Some Context
      3. Cut to the Chase
      4. Repeat the Important Information
      5. Provide the User with a Way to Get More Details
    5. Some Scenarios
      1. Scenario 1
      2. Scenario 2
      3. Scenario 3
  23. 21. Laying Out the Foundations
    1. Bring Together All the Key Players
    2. Define the Business Goals
    3. Define User Needs and Intent
      1. Identify User Tasks
      2. Identify Usage Patterns
    4. Define the Voicebot’s Voice Register
  24. 22. The Key to Successful Product Launches
    1. Write Everything Out in Full Sentences
    2. Your Press Release Needs to Be Crystal Clear
    3. Your Answers Are Given in One or Two Paragraphs at Most, and Not Much More
    4. Answer the Basic Questions First
    5. Describe Clearly the Research You Have Done
    6. Be Modest and Cautious in Your Claims and Statements
    7. Make Your Document Readable by Everyone
    8. List the Functional Requirements in Terms of What the User Can Do
    9. Describe the Intended Minimum Viable Product (MVP) in Detail
  25. 23. The Elements of Deployment
    1. Product Management
    2. Product Marketing
    3. UX Research
    4. UX Design
    5. Development
    6. Quality Engineering
      1. Functional Testing
      2. Traversal Testing
      3. Stress Testing
      4. Beta Testing
    7. Program Management
  26. 24. Post-Launch Monitoring
    1. Sources of Information
      1. Interaction Logs
      2. The Audio of the Interactions
      3. End Users
    2. The Basic Questions
      1. Where Are Users Abandoning the Session?
      2. Where Are Users Asking to Be Connected to a Human Agent?
      3. Where Are Users Saying the Wrong Things?
      4. Where Are Users Not Saying Anything?
      5. Where Are Users Speaking Too Soon?
      6. What Is the Noise Level of Your User’s Environment?
      7. What Options Are Your Users Asking For?
      8. How Are Users Feeling About the Voicebot?
  27. 25. The Elements of Voice First Success
    1. Abandonment Rate
    2. Automation Rate
    3. Average Number of Failures per Session
    4. Average Number of Failures per Task
    5. Average Task Completion Time
    6. Containment Rate
    7. First-Use Resolution Rate
    8. Task Completion Rate
    9. Task Initiation Rate
    10. Time to Task
  28. 26. Coda
  29. A. The 10 Sources of Voice First Failures!
  30. B. Demonstrating Voice First
    1. Remove All Prompts That Explicitly Talk About Failure
    2. Don’t Speak Over Prompts
    3. Don’t Have the Voicebot Talk for More Than 10 Seconds When It Starts Without Giving the Turn Back to You
    4. Test the Voicebot with the Same Equipment You Will Use in the Demo
    5. Have Two or Three Backup Ways to Demo the Voicebot
    6. Test the Application in the Same Room and Environment Where You Will Do the Demo
    7. Know How to Gracefully End a Telephony-Based Voicebot
    8. Ask for Silence
    9. Never Improvise or Show Off While Demoing
    10. If, for Whatever Reason, the Voicebot Fails, Be Honest About Why It Failed
    11. No Cheat Sheets!
    12. Speak Normally
    13. Don’t Leave Long Silences Between Words When You Speak
    14. Keep It Really Short
  31. C. Useful Matrices
    1. The Assessment Matrix Cheat Sheet
    2. Value-Usability Matrix
    3. Feature Expectation Matrix
    4. Automation-Complexity Matrix
      1. Quadrant 1
      2. Quadrant 2
      3. Quadrant 3
      4. Quadrant 4
  32. A Voice First Glossary
  33. References
  34. Index
  35. About the Authors

Product information

  • Title: The Elements of Voice First Style
  • Author(s): Ahmed Bouzid, Weiye Ma
  • Release date: May 2022
  • Publisher(s): O'Reilly Media, Inc.
  • ISBN: 9781098119591