search

robotemi / sdk

temi is an unparalleled robotic platform introducing a new dimension of development - movement. Using temi’s SDK, developers can create new functionalities and introduce new use cases via temi’s Android tablet. temi’s movement and navigation capabilities run off of it's Linux computer and uses a set of 16 sensors including a Lidar, depth cameras, driving cameras, and microphones. We encourage our developer community to suggest and request expanded functionality within the SDK and we will adhere to your needs. We want to enable you to create skills like never seen before! We will be constantly improving the SDK and its documentation. Please feel free to reach out to us with any questions or thoughts at developers@robotemi.com
https://www.robotemi.com/
201 stars 85 forks source link

How to trigger asr listener without using hey temi #427

Open mubashir-shk opened 8 months ago

mubashir-shk commented 8 months ago

I am currently working on implementing a question-answer session with Temi. However, I want to eliminate the need for the user to say 'Hey Temi' before asking a question. I have two specific concerns:

1) How can I trigger the ASR listener without using the 'Hey Temi' command? 2) Will Temi always be in listening mode, or does it go to sleep after answering a question, similar to the behavior with 'Hey Temi'? I want to avoid the scenario where the user has to say 'Hey Temi' for each question, and instead, I want Temi to consistently listen and answer because users can ask questions at any time.

zjn0505 commented 8 months ago

wakeup() will make temi start to listen.

You will need to call wakeup() again to keep it listening, after TTS/ASR listeners giving you the event that conversation ends.

Perhaps add person detection listener to only start to listen when there is person in front.

Also your app can be kiosk app, so the converstaion layer can be overriden by your app.

You can look into this page https://github.com/robotemi/sdk/wiki/Speech for APIs you can use.

mubashir-shk commented 8 months ago

Thank you so much for your reply. Just one more question: as the conversation ends after the first question, I will call wakeup() again. Suppose the user didn't instantly ask a second question and asked after 5 seconds, will Temi wait for the user until 5 seconds or will it go to sleep? How much time will Temi wait for user input on the ASR listener after calling wakeup()?