Add speech input, voice input, dictate function for keyboard

Bug #1545714 reported by Peter Bittner
46
This bug affects 11 people
Affects Status Importance Assigned to Milestone
ubuntu-keyboard
New
Wishlist
Unassigned
ubuntu-keyboard (Ubuntu)
Confirmed
Undecided
Unassigned

Bug Description

Typing on a mobile device is tedious. In some situations it's advantageous to dictate some text, use speech input that is translated to text (for texting, for writing documents, tasks, emails, etc.).

A dictate function should be available in a integrated fashion with the keyboard, everywhere on Ubuntu Touch where text input is needed.

Details
=======

This could be implemented as an integral feature (as in iOS, I believe) or via a pluggable API (on Android, I believe).

Changed in ubuntu-keyboard:
importance: Undecided → Wishlist
Revision history for this message
Peter Bittner (peter-bittner) wrote :

The Houndify API [1] allows free speech2text with the "Speech To Text Only" domain [2], as they call it.

[1] https://www.houndify.com/
[2] https://www.houndify.com/domains/detail/71/

Revision history for this message
Launchpad Janitor (janitor) wrote :

Status changed to 'Confirmed' because the bug affects multiple users.

Changed in ubuntu-keyboard (Ubuntu):
status: New → Confirmed
Revision history for this message
Peter Bittner (peter-bittner) wrote :

Is there any progress with this? Are there any, at least vague, plans to make this happen anytime soon?

Revision history for this message
Peter Bittner (peter-bittner) wrote :

Just found this video from 23-11-2016: "Ubuntu Keyboard - Speech Recognition"

- https://www.youtube.com/watch?v=vglBQanR_t4

Revision history for this message
Michael Sheldon (michael-sheldon) wrote :

That video's just from a bit of experimentation in my free time; once we've got the input method interface for snapd finished up I'm planning on making it available for people to experiment with on the desktop, but it's not really suitable for use on a phone. Any phone based solution is likely to require some sort of cloud service providing the actual text-to-speech processing.

Revision history for this message
Peter Bittner (peter-bittner) wrote :

That sounds promising!

Why isn't offline processing suitable for use on a phone? If it's working on a desktop, and we have convergence, we should be able to assume it will work on any (better) smartphone in 2018, shouldn't we?

Revision history for this message
Michael Sheldon (michael-sheldon) wrote :

The laptop I recorded that demo on has a high end i7, which was operating at 100% on all 8 cores (it can work with much lower CPU usage if CUDA is available though); so even when mobile processors get to an equivalent performance level it'd be terrible for battery life. The only exception would be if phones started shipping with CUDA capable processors (nVidia do have some mobile oriented CUDA capable chips, but they're not widely used at the moment). So unfortunately for the foreseeable future the most practical solution for mobile systems is to have a CUDA backed server processing requests remotely I'm afraid.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.