3 ms·

I would prefer to have camera read my lips rather than read my fingers as I type on nothing.

by reportgunner 7y ago

Quequau 7y ago
As I already talk to myself a lot, I'm way closer to this sort of strategy working than I would have thought before you mentioned it. Though I don't enunciate at all when I talk to myself, so whatever was doing the detecting would need to work on subvocalization. It reminds me of a talk someone gave about this ad-hoc verbal shorthand someone invented so that they could use speech-to-text to code. The talk sounded really funny but the process did seem fairly productive.
golergka 7y ago
Spelling something like AbstractFactory<Factory<IEnumerable<TDataType>>> must be fun. (But not as fun as it would be if I had more experience with Perl).
- bmn__ 7y ago
  cap abstract cap factory join leb cap factory leb capital letter india cap enumerable join leb capital letter tango cap data join cap type join reb reb reb slap https://youtube.com/watch?v=Mz3JeYfBTcY&t=465 https://youtube.com/watch?v=Mz3JeYfBTcY&t=465
  - Mirioron 7y ago
    While I think that the way he has figured out how to input symbols quickly is really cool, I wonder if there wouldn't be a better way. Our autocomplete systems can close brackets and all those nice things. Maybe voice commands could understand structures like templates when inputting commands with your voice?
    - bmn__ 7y ago
      Indeed, and the example above is just one of many possible ways to achieve the same result. That one works out of the box, but is tedious and cumbersome as you noticed. To reduce voice strain and make input more reliable, one would spend a little bit of time to set up shortcuts so that e.g. "i enumerable" produces 'IEnumerable' and "of type" expands to '<‸>' and places the cursor at the caret. The voice input system is fully programmable at different levels of abstraction.

3 ms·

I would prefer to have camera read my lips rather than read my fingers as I type on nothing.

by reportgunner 7y ago

I would prefer to have camera read my lips rather than read my fingers as I type on nothing.

Quequau 7y ago
As I already talk to myself a lot, I'm way closer to this sort of strategy working than I would have thought before you mentioned it. Though I don't enunciate at all when I talk to myself, so whatever was doing the detecting would need to work on subvocalization. It reminds me of a talk someone gave about this ad-hoc verbal shorthand someone invented so that they could use speech-to-text to code. The talk sounded really funny but the process did seem fairly productive.
golergka 7y ago
Spelling something like AbstractFactory<Factory<IEnumerable<TDataType>>> must be fun. (But not as fun as it would be if I had more experience with Perl).
- bmn__ 7y ago
  cap abstract cap factory join leb cap factory leb capital letter india cap enumerable join leb capital letter tango cap data join cap type join reb reb reb slap https://youtube.com/watch?v=Mz3JeYfBTcY&t=465 https://youtube.com/watch?v=Mz3JeYfBTcY&t=465
  - Mirioron 7y ago
    While I think that the way he has figured out how to input symbols quickly is really cool, I wonder if there wouldn't be a better way. Our autocomplete systems can close brackets and all those nice things. Maybe voice commands could understand structures like templates when inputting commands with your voice?
    - bmn__ 7y ago
      Indeed, and the example above is just one of many possible ways to achieve the same result. That one works out of the box, but is tedious and cumbersome as you noticed. To reduce voice strain and make input more reliable, one would spend a little bit of time to set up shortcuts so that e.g. "i enumerable" produces 'IEnumerable' and "of type" expands to '<‸>' and places the cursor at the caret. The voice input system is fully programmable at different levels of abstraction.