Back to results
US 7,680,853 B2Granted
Clickable snippets in audio/video search results
AssigneeMicrosoft Corporation
InventorsRoger Peng Yu, Frank Torsten Seide, Kaijiang Chen
Life of this patent
Estimating the term…
Abstract
Search results are provided in a format that allows users to efficiently determine whether audio or video documents identified from a search query actually contain the words in the query. This is achieved by returning snippets of text around query term matches and allowing the user to play a segment of the audio signal by selecting a word in the snippet. In other embodiments, markers are placed on a timeline that represents the duration of the audio signal. Each marker represents a query term match and when selected causes the audio signal to begin to play near the temporal location represented by the marker.
Claims15
A method comprising:
- receiving a search query comprising at least one query term;
- using the query term to retrieve a document identifier and a start time for the query term in an audio signal by accessing an index formed from text recognized from audio signals associated with spoken documents;
- using the start time to retrieve a string of words that includes the query term and that was recognized from the audio signal for the document represented by the document identifier;
- and
- generating on a display device, a user interface displaying the string of words including displaying the query term such that the string of words displayed in the user interface does not represent the entire time span of the audio signal and at least one word in the string of words is a selectable link such that when a user selects the at least one word in the user interface, playback of the audio signal is initiated at the start time of the query term in the audio signal, and further displaying a timeline having a length representing the duration of the audio signal and displaying a marker for the query term on the timeline to indicate the temporal position of the query term in the audio signal, wherein displaying a marker on the timeline comprises multiplying a ratio of the start time of the query term to the duration of the audio signal by the length of the displayed timeline to determine a location for the marker on the timeline.
2 more elements
A computer-readable storage medium tangibly embodying a program of instructions executable by a processor to perform steps comprising:
- searching an inverted index formed from words recognized from audio signals to locate query terms in a search query and producing query matches for occurrences of the query terms in the inverted index;
- identifying text snippets from the query matches, each text snippet containing at least one query term and at least one word that is not part of the search query, all of the text snippets for an audio signal together representing less than the duration of the audio signal;
- displaying the query terms that were used to perform the search;
- displaying the text snippets associated with a selected audio signal such that a displayed query term in a displayed text snippet is a selectable link such that when a user selects the query term, playback of the audio signal is initiated at a start time of the query term in the audio signal;
- displaying a timeline having a length representing the duration of the selected audio signal and displaying a separate marker for each query match on the timeline to indicate the temporal position of each query match in the selected audio signal, wherein displaying a marker on the timeline comprises multiplying the ratio of the temporal position of the query term to the duration of the selected audio signal by the length of the displayed timeline to determine a location for the marker on the timeline and wherein displaying a marker comprises setting the height of the marker based on a word score for the search match such that the marker is taller for higher word scores and such that different markers on the timeline have different heights;
- based on an indication that a user has selected a word in a text snippet, requesting a segment of an audio signal, the segment containing less than the entire time span of the audio signal and starting at the start time of the query term;
- and
- based on an indication that a user has selected a marker on the timeline, requesting a second segment of the audio signal, the second segment containing less than the entire time span of the audio signal.
5 more elements
A method comprising:
- a processor determining temporal locations of a query term in a plurality of audio signals;
- and
- generating a user interface on a display device, the user interface comprising a separate timeline for each audio signal and a set of markers for each timeline, wherein each timeline has a length representing a time span of the respective audio signal and the markers for the timeline are positioned along the length of the timeline to show the temporal locations of the query terms within the audio signal, wherein displaying a marker on the timeline comprises multiplying a ratio of a temporal location of the query term to the duration of the audio signal by the length of the displayed timeline to determine a location for the marker on the timeline and setting a brightness value for the marker based on a word score associated with the temporal location of the query term such that the marker is made brighter for a higher word score and such that different markers along the timeline have different brightness values.
Dependencies are inferred from the claim text.