666
edits
Connected Devices Weekly Update/2016-10-13: Difference between revisions
Jump to navigation
Jump to search
Connected Devices Weekly Update/2016-10-13 (view source)
Revision as of 16:13, 13 October 2016
, 13 October 2016deep speech
Benfrancis (talk | contribs) (HomeWeb update) |
(deep speech) |
||
| Line 24: | Line 24: | ||
** [[File:Haiku.jpg|300px]] | ** [[File:Haiku.jpg|300px]] | ||
|- | |- | ||
| | | Platform and Tools | ||
| Sandip/Nicole | | Sandip/Nicole/Mahe | ||
| N/A | | N/A | ||
! style="background:#00EC00;" | On Target | ! style="background:#00EC00;" | On Target | ||
| Line 32: | Line 32: | ||
* Bulleted list item | * Bulleted list item | ||
|- | |- | ||
| Platform | | [[Voice_Interface|Voice Platform - Deep Speech]] | ||
| Sandip/Nicole | | Sandip/Nicole | ||
| N/A | | N/A | ||
! style="background:#00EC00;" | On Target | ! style="background:#00EC00;" | On Target | ||
| | | | ||
* | * Integrated two training sets into Deep Speech | ||
* | ** TED Talk speech corpus (300 hours of speech) | ||
** LibriVox speech corpus (1000 hours of speech) | |||
* Implemented WER (word error rate) evaluation code for train, validation, and test data sets to track WER progress as the model evolves. | |||
* Started implementation of reporting GUI to track training progress | |||
[[File:Deepspeech WER.png|500px]] | |||
|- | |- | ||
| [[Connected_Devices/Projects/Abigail|Abigail]] | | [[Connected_Devices/Projects/Abigail|Abigail]] | ||