IBM Watson™ Ideas

Welcome to the IBM Watson™ Ideas Portal

We welcome and appreciate your feedback on IBM Watson™ Products to help make them even better than they are today!

If you are looking for troubleshooting help or wondering how to use our products and services, please check the IBM Watson™ documentation. Please do not use the Ideas Portal for reporting bugs - we ask that you report bugs or issues with the product by contacting IBM support.

Before you submit an idea, please perform a search first as a similar idea may have already been reported in the portal.

If a related idea is not yet listed, please create a new idea and include with it a description which includes expected behavior as well as why having this feature would improve the service and how it would address your use case.

Support STT edge use when not connected to IBM Cloud

Our WPA Automotive clients need to use speech to text when not connected to cloud.  Like turn on and turn off the lights commands in the car when for example the car is in a tunnel and doesn't have internet connectivity to WPA and IBM CLoud services.

  • Sep 11 2017
  • Needs review
Why is it useful?
Who would benefit from this IDEA?
How should it work?
Idea Priority
Priority Justification
Customer Name
Submitting Organization
Submitter Tags
  • Attach files
  • Jakub Krchák commented
    October 05, 2017 13:33

    The issue is broader than just speech, the WPA team would like to have (limited) conversational capability offline.

    Most relevant speech related task:
    - create offline STT SDK
    -- well-defined API
    -- documentation
    -- offline capabilities
    --- low-energy keyword activation
    --- speech barge-in into TTS
    --- platform support - most importantly BLAS (or equivalent) libraries on ASM level
    -- automotive acoustic models
    -- tools for LM pruning (basic LM smaller than for service)
    -- more languages :)
    -- platform CI

    - create offline TTS SDK
    -- well-defined API
    -- documentation
    -- offline capabilities
    --- CELP voice support
    --- smaller CELP voices (tooling)
    --- RNN prosody & phrasebreak speed up / reimplementation
    --- solve expressive for embedded
    --- parametric (voice transformation) for embedded

  • BHAVIK SHAH commented
    October 11, 2017 15:32

    Hybrid Model is not just Offline speech -- Need a Hybrid Conversation Framework (that should include STT, Conversation, NLU, TTS) - an Edge Model.  Low energy keyword activation can be an first phase of this entire Edge Model Program. For this we would need to know details of how it can be exposed - i.e. open source, library, compiled code and what programming paradigm will it support: (i.e. IOS, Android, Raspberry Pi, Auto-specific models?)

    Right now this work will not be prioritized for Q4 and Q1. We will start having strategic discussions on this in Q4 and have a goal to have something on roadmap by end or Q4 for edge computing.

  • Derek Carroll commented
    October 24, 2017 07:51