en-es
en-fr
en-sl
en
0.25
0.5
0.75
1.25
1.5
1.75
2
Speech Translation in human-to-human Interaction: Skype Translator
Published on Jul 31, 20161161 Views
The error rate of speech recognition has come down enough, and the quality of machine translation has gone up enough to allow us to combine the two and achieve useful results. When building a transla
Related categories
Chapter list
Speech translation in human-to-human interaction: Skype Translator00:00
Breaking down language barriers00:28
Should the interpreter have a persona? - 100:59
Should the interpreter have a persona? - 201:33
Why now?02:20
Skype Translator: What is it?03:34
How people really speak04:29
Pipeline05:07
There is a large perceived need for translation services in the US, across personal and business scenarios06:45
Untitled07:08
Rough grouping of scenarios - 107:50
Rough grouping of scenarios - 208:40
Human interpreter example - 109:02
Human interpreter example - 209:06
Human interpreter example - 309:09
Human interpreter example - 409:12
Human interpreter example - 509:32
Our first attempt - 109:37
Our first attempt - 209:48
Our first attempt - 309:50
User feedback received on that design09:51
The system needs to adjust09:58
People can talk over each other - 110:16
People can talk over each other - 210:21
People can talk over each other - 310:22
People can talk over each other - 410:23
What about audio?10:33
People want to hear the original audio10:50
Ducking: Varying the volume of the original audio11:51
Perfecting the audio settings12:38
Text and Audio User Preference12:44
Typical lifetime of an utterance - 113:25
Typical lifetime of an utterance - 213:43
Typical lifetime of an utterance - 313:51
User chooses terms13:56
This is my transcript14:17
This is my partner’s translated transcript14:21
Users don’t discover that they can type here14:22
Where are we going?14:51
MT quality14:52
Machine Interpretation15:24
Speech Translator on Github15:45
Thank you16:07