Out of curiousity and slight enthusiasm, I signed up for translate.mycroft.ai to try to create a translation for Esperanto (code: eo), but there is no option to add a new language.
I also read something on Github about editing some mycroft core files but I’m not sure if this is the good way to start translating to a language that’s not yet listed.
Any advice on where I can start?
We only allow admins to add languages as it is a fairly intensive process. As such we don’t want languages being added and removed all the time, particularly when there might already be heavy load on the server.
So posting here was a great place to start - Esperanto (eo) is now available
I would like to support this project too and I think that the Esperanto Version of the Common Voice Project by Mozilla could be very useful for this. It collects sound samples to train speech-to-text systems, I believe they also collaborate with mycroft for many other languages
I would like to add Mycroft voice commands to this project, but they have to be public domain. Is this the case or do mycroft command have any sort oft copyright?
We do indeed work with Mozilla on a few projects. However I wanted to clarify what data you are referring to specifically?
If you opted-in to the Open Data Set when creating your account or on your Profile page then samples from your account would be shared with Mozilla for their voice work. Any samples you collect yourself are completely yours.
If you mean voice commands being the text used to define intents eg “set a timer” then these are covered by the license on each Skill’s repository. Mycroft built Skills are under the Apache 2.0 license and all Skill in the Marketplace must be under an appropriate open source license. They aren’t however in the public domain, so may not meet Mozilla’s requirements for this particular project.
I was talking about the licenses of the texts. I thought it might be useful to record the commands directly.
Ah I see so I won’t use them. Maybe I will write down similar voice commands manually because I believe that this could be a useful thing for the dataset. The german version of common voice already contains a list of sentences thats looks like commands for a unspecified voice assistant.