Actionable voice commands within a connected fitness platform

Patent metadata

Field	Value
Publication number	US-12558605-B2
Application number	US-202318545902-A
Country	US
Kind code	B2
Filing date	Dec 19, 2023
Priority date	Dec 20, 2022
Publication date	Feb 24, 2026
Grant date	Feb 24, 2026

Abstract

A connected fitness platform can determine members are recognized during a live class or other live event, and seamlessly perform actions in response to or along with the recognition. The platform may determine a user or member is being recognized by an instructor or leader of the class/event, such as by tokenizing usernames and utilizing a semantic database to match or identify members represented by the usernames. The platform may then perform actions for the identified members.

First claim

What is claimed is:

1 . A method performed by a connected fitness platform that streams exercise classes to members of the connected fitness platform via exercise machines associated with the members, the method comprising:
capturing a voice snippet spoken by an instructor during a live exercise class,
wherein the live exercise class is streamed to multiple remote exercise machines that present the live exercise class to members of the connected fitness platform that are performing exercise activities while participating in the live exercise class;
extracting a username from the voice snippet; tokenizing the extracted username into multiple tokens; comparing the multiple tokens to a database of tokenized usernames associated with the live exercise class; and identifying a specific member that is participating in the live exercise class based on the comparison.

2 . The method of claim 1 wherein identifying a specific member of the live exercise class based on the comparison includes matching at least two tokens of the multiple tokens to tokens contained by the database of tokenized usernames associated with the live exercise class.

3 . The method of claim 1 , wherein extracting a username from the voice snippet includes identifying a context during the live exercise class within which the instructor provides the voice snippet.

4 . The method of claim 1 , further comprising:
performing an action associated with the identified specific member during the live exercise class.

5 . The method of claim 1 , wherein the identified specific member is participating in the live exercise class via an associated exercise machine that presents information for the live exercise class, the method further comprising:
causing the exercise machine associated with the identified specific member to display a graphical element during the live exercise class that indicates the instructor has spoken the username representing the identified specific member.

6 . The method of claim 1 , wherein the identified specific member is participating in the live exercise class via an associated exercise machine that presents information for the live exercise class, the method further comprising:
causing the exercise machine associated with the identified specific member to capture an image of the identified specific member during the live exercise class.

7 . The method of claim 1 , wherein the identified specific member is participating in the live exercise class via an associated exercise machine that displays a leaderboard for the live exercise class, the method further comprising:
causing the exercise machine associated with the identified specific member to display a graphical element overlaying the leaderboard during the live exercise class that indicates the instructor has spoken the username representing the identified specific member.

8 . The method of claim 1 , wherein the database of tokenized usernames associated with the live exercise class is generated after commencement of the live exercise class.

9 . The method of claim 1 , wherein extracting a username from the voice snippet includes:
determining the instructor has spoken one or more keywords associated with identifying users of the live exercise class; and capturing the voice snippet in response to the determination.

10 . The method of claim 1 , wherein extracting a username from the voice snippet includes:
determining the instructor has performed a certain movement associated with identifying users of the live exercise class; and capturing the voice snippet in response to the determination.

11 . A system, comprising:
a processor; and a memory coupled with the processor, the processor configured to cause the system to:
receive a text string based on one or more words spoken by an instructor of a live exercise class,
wherein the live exercise class is streamed to multiple remote exercise machines that present the live exercise class to users performing exercise activities associated with the live exercise class;

tokenize the text string into multiple tokens;
compare the multiple tokens to a database of tokenized usernames associated with the live exercise class; and
identify a member that is participating in the live exercise class based on the comparison.

12 . The system of claim 11 , wherein the text string represents a username or hashtag spoken by the instructor of the live exercise class.

13 . The system of claim 11 , wherein the database of tokenized usernames associated with the live exercise class is generated after commencement of the live exercise class.

14 . A non-transitory, computer-readable medium whose contents, when executed by a computing system, cause the computing system to perform a method, the method comprising:
capturing a voice snippet spoken by an instructor during a live exercise class streamed to multiple remote exercise machines that present the live exercise class to members performing exercise activities while participating in the live exercise class; extracting a username from the voice snippet; tokenizing the extracted username into multiple tokens; comparing the multiple tokens to a database of tokenized usernames associated with the live exercise class; and identifying a specific member that is participating in the live exercise class based on the comparison.

15 . The non-transitory, computer-readable medium of claim 14 , wherein identifying the specific member of the live exercise class based on the comparison includes matching at least two tokens of the multiple tokens to tokens contained by the database of tokenized usernames associated with the live exercise class.

16 . The non-transitory, computer-readable medium of claim 14 , wherein extracting a username from the voice snippet includes identifying a context during the live exercise class within which the instructor provides the voice snippet.

17 . The non-transitory, computer-readable medium of claim 14 , further comprising:
performing an action associated with the identified specific member during the live exercise class.

18 . The metho non-transitory, computer-readable medium of claim 14 , wherein the identified specific member is participating in the live exercise class via an associated exercise machine that presents information for the live exercise class, the method further comprising:
causing the exercise machine associated with the identified specific member to display a graphical element during the live exercise class that indicates the instructor has spoken the username representing the identified specific member.

19 . The non-transitory, computer-readable medium of claim 14 , wherein the identified specific member is participating in the live exercise class via an associated exercise machine that presents information for the live exercise class, the method further comprising:
causing the exercise machine associated with the identified specific member to capture an image of the identified specific member during the live exercise class.

20 . The non-transitory, computer-readable medium of claim 14 , wherein the identified specific member is participating in the live exercise class via an associated exercise machine that displays a leaderboard for the live exercise class, the method further comprising:
causing the exercise machine associated with the identified specific member to display a graphical element overlaying the leaderboard during the live exercise class that indicates the instructor has spoken th

Assignees

• Peloton Interactive Inc

Inventors

• Khandelwal Ashay
• Hemaraj Kiran
• Kang Kenny
• Lai Michael
• Nguyen Nelson
• Patil Sunil

Classifications

• G10L15/1815

  Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning · CPC title

• G10L2015/088

  Word spotting · CPC title

• A63B2220/807

  Photo cameras · CPC title

• A63B2071/063

  Spoken or verbal instructions · CPC title

• G10L2015/223

  Execution procedure of a spoken command · CPC title