TiVo Voice Features & Benefits

• “Margin Call”
• “Terminator 3”
• “MI3”
• “The Simpsons”
• “How I met your Mother”

• Recognizes between actual titles and subscriber utterance
• Will return what you said in the right language if there are multiple versions of the title

• "Mission Impossible"
• "Terminator"
• "Fast and Furious"
• "Bourne movies"
• "Star Wars movies"

• The metadata doesn't provide the linkage
• The franchise name isn't in one of the titles, such as "The Empire Strikes Back"

• "Red Sox"
• "Golden State Warriors"

• Channel surfing
• Calling customer support

• "When are the Celtics playing next?"
• "Find the Warriors Cavaliers game tonight"

• Find an upcoming team
• Game schedules

• "James bond movies"

• That contain a specific character
• Where the system understands the relevance of the name 
• Where the system knows when a user is searching for a role versus a real person, providing a more accurate experience

• "World Series"
• "NBA Playoffs"
• "US Open Tennis"
• "The Masters"            

• Find tournaments and games for events that may be titled differently from how they're scheduled
• This works when events are shown as upcoming in the Guide.

Alternate Title (alias) Search

• "New York Yankees (Yanks)"

• "Warriors (Golden State Warriors)"

• "Cards (Arizona Cardinals)"

• "The Oscars (Academy Awards)"

• "Show me something funny (comedy genre)"

• "Scary movies (horror genre)"

• "Tune to HGTV, or tune to Home and Garden TV"

• "7 (Seven)"

• Find these without knowing exactly how they are labelled or titled:

  • Teams  
  • Shows
  • Events
  • Channels
  • Movies

Popular Abbreviation Search

• "MI3"

Understand abbreviations of:

• Teams
• Shows
• Events
• Genres
• Channels
• Movies

Graph Connections Search

• "Movies with Jack Gleeson from Game of Thrones (e.g.: A Shine of Rainbows, All Good Children)"

• "Movies with Julia Louis from Seinfeld and James Gandolfini (e.g.: Enough Said)"

• Find items you are looking for across contexts even when the system provided incomplete information 

• Use the context to figure out the intended entity

• Account for incorrect speech to text transcriptions based on the context
   

Actor Title Search

• "Tom Cruise"

• "Penelope Cruz"

• Find the most relevant titles where the actor is in their most likely preferred order

• Provide lower priority to non-leading roles and previously viewed titles

Crew Search

"Movies directed by Tom Hanks"

• Order results based on the importance of the role within
  the program 

• De-duplicate between the production house and the individual producer on a query

• Find items where the individual provides the significant contribution to the work

Personality Intersections

"Tom Hanks movies with Julia Roberts"

Find movies and shows based on multiple cast members

Personality Genre Intersections

"Tom Hanks comedy movies"

Find movies or shows

• With actors you like
• The genre you're in the mood for
• Personalized to your interests

Keyword Search

• “Movies about the financial crisis” (such as Margin Call,

• “Tom Hanks plane crash movie” (such as Castaway, Sully),

• “Friends episodes with the beach” (such as The One with the Jellyfish, The One at the Beach)

Find:

• Shows/movies that you can't remember the title of based on what it was about

• New things to watch based on more granular interests

Sports Genre Search

• "Football Games"

• "Basketball"

Find sports: 

• Across every league you care about

• The order personalized to their tastes.

Age appropriate content search

• "Show me some PG rated movies"

• "Show me shows for my 3 year old"

Provides flexibility so you can create specialized areas like kids zones.

Negative Filters

• "Comedy movies without Tom Hanks"

• "Superman movies without Brandon Routh"

• "Comedy movies not rated TVMA"

Understand when items should be excluded from 

Personality Keyword Intersections

“Which Mob movies was Robert DeNiro in?”

Find movies or shows with:

• Actors you like

• In the genre you are in the mood for

• Personalized to your interests

Selection filters

• "Show me the latest episode of Bones"
   

Helps you

• Find and navigate
  to the exact content you want based on your criteria

• Be able to apply qualifiers to searches that can be configured by business rules
   

Availability attribute filters

• "Celtics live games"

• "Next live football game"

Helps you:

• find the exact version of the program you  want

• By filtering on airing information such as:

• Live

• HD, with closed captions,

• etc.

Episode Attribute Search

• "Record the season premiere for Game of Thrones"

You can refer to specific episodes to find:

• the one you care about, the season premiere, the latest one,

• only new ones (vs reruns),

• etc.
   

Sports Temporal Relevance

• "When are the Cardinals playing ( Arizona Cardinals if NFL)"

• "When do the Giants play? (San Francisco vs NY Giants)"

Find the right team based on the season

Genre Search 

• "Comedy shows"

• "Action movies"

• Receive personalized recommendations based on user-friendly genres  such as

  • "scary movies" or

  • "funny shows"

• Which aren't typical tags on shows and movies

Decade or
Release Year
filters

"Comedy movies from the '90s"

• Dig deeper into the provider's catalogue of content

• Find personalized recommendations for shows and movies you may have missed the first time around

Temporal Filter Search

• "Shows on TV tonight"

• "Movies this weekend"

• "What is on this weekend?"

• "What movies are on this weekend from the 80s?"

• "What movies are on this weekend from the 80s on Netflix?"

• Find content in the future across multiple combos. 

• Find the shows and movies in the future that you are most likely to be interested in, with an ease not available in a guide navigation

Source-based Content Search

• "Shows on TV"

• "Movies on Netflix"

• "Movies On Demand"

• "Shows on Netflix"

• Limit the search to the service being searched on

• Show personalized recommendations across any content source without needing to go into a separate application or interface.

Content Recommendations

• "What's on TV?"

• "Anything good on Netflix?"

• "What can I watch tomorrow morning?"

Provide personalized recommendations for anytime

• day

• channel

• source

• etc.
   

Content to Content Recommendations

"Show me movies like The Lego Movie"

• Show similar titles to those you have recently watched and without requiring changes to the User Interface treatment,

• Decide how you want to see that "Because you watched…" type recommendations

Feature

Examples

Function

Station title Search 

• “Watch FXX”  

• Skip scrolling through screen after
  screen to tune to the right one on the first

• Attempt using the user's entitlements to find
  the ones that are particular in your locale 

Voice-activated Scrolling

“Show me movies on tonight” > “more”

• You aren’t limited to a single screen When many results are available

• Makes it very simple for you to navigate without requiring a wake-up word

Navigation

• “Launch Netflix”

• “Home”

• “Guide”

• Launch any app by saying its name

• See the list of supported apps and UI screens: UI Launch Points and Apps

List Item Selection 

“Tom Hanks movies” -> “pick the first one”

Maintain context across queries and understand what item you are selecting from the list of results.

Results Refinement

“Tom Hanks movies” -> “the one released in 2012”

• Maintain context across queries so you don’t have to repeat the initial request and try to combine it in a single query

• Making it much easier to speak to the system more like you would a person.

 Results non-refinement

“Show me movies with Tom Cruise” -> “Movies with Tom Hanks” -> “Adventure movies on tonight”

Understands when you have switched context 

Pronoun based refinement

“Tom Cruise movies with Nicole Kidman” -> “his comedy movies”

Speak to the feature as if it's a person that understands how subsequent queries are related

Context sensitive Entity Recognition

"When do the Kings play next?" (expect: game intent) 
VS 
"When is Kings releasing?" (expect: movie intent)

Understand which entity you care about based on what information you want for it

Speech-to-text Error Corrections 

• “Tom thanks” (correct to Hanks)

• “red socks games” (correct to Red Sox games)

• Correct misspellings that may come through from the Automated Speech Recognition service

• The system understands enough about the domain that even when the upstream systems may have errors, it will auto-correct to the most likely user intent

Results Refinement (in the presence of errors)

“Yankees” > “their games with Rangers” (should get NY Yankees and Texas Rangers games - deduplicating the name vs the Rangers of the NHL)

Returns results when there is a chance of being ambiguous with entities to match against

Gibberish detection

“Get it” (respond with: “Sorry, I couldn’t get anything for you”)

The system

• can tell when something irrelevant has been entered

• can provide a graceful response to nonsensical input

Feature

Examples

Function

Quotes Search 

• “Here's looking at you kid” (for example, Casablanca)

• “I'm gonna make him an offer he can't refuse” (for example, The Godfather)

Find content based on popular catch-phrases and quotes