You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm thinking of adding some issue templates in order to make it as easy as possible for folks to:
report sounds that "say the wrong things" (e.g. pit mode / modo foso in Spanish - is roughly equivalent to saying "moat mode" en English, it just makes no sense unless you know that a pit and a moat are kinda similar in a context that's not remote-controlled aircraft.)
report sounds that "don't sound right" (e.g. es-CL transcribes hertz as jertz for phonetic purposes, even if the word in written hertz in Spanish, that's sometimes necessary depending on what voice is used by the TTS service. More on that below.)
request new locales to be supported (e.g. Son Belges #62)
request new voices to be supported (why not?)
Other (a default template)
"says the wrong thing"
It seems like providing voice packs in different locales can make EdgeTX more accessible to more people.
Relying on machine translations allows to provide some meaningful level of support to non-English speakers, but good translations require both language and EdgeTX proficiency.
So there will be some mediocre translations in the voice packs that receive less attention (see 👍 in #17 despite "mode fosse" in fr-FR and "modo foso" in es-ES which are wrong.)
Review mistakes will happen, and not all sounds will make sense to all reviewers every time and that should be okay. I think we can make the trade-off relying on unreliable machine translations acceptable by making sure that reporting errors is easy, so they get fixed over time.
"doesn't sound right"
Sometimes terms that we use are jargon that a general-purpose text-to-speech service may not render correctly. The current pipeline allows the text to be adjusted in order to compensate for that, and it's an easy thing to fix, assuming that someone reported it.
I hope that having a template to report those issues will help lowering the amount of effort folks need to put into contributing to making the voice packs better in their own language.
new locale please
That one seems obvious. If there are no voice packs for your locale, and adding them is not too hard / costly, it should be easy to request them. 🙂
new voice please
There are different voices for some of the locales already (e.g. en_US), so I suppose that's something that people request from time to time.
default template
Of course, providing issue templates fro some common requests / problems shouldn't prevent anyone from reporting the other ones.
The text was updated successfully, but these errors were encountered:
There is also a secondary goal to this. I'd like to start distinguishing between the different stages of the pipeline that produces the voice packs, and these issue templates could help introducing the idea that translation and text-to-speech are two distinct stages of that pipeline, with their own particularities, procedures and issues.
Overview
I'm thinking of adding some issue templates in order to make it as easy as possible for folks to:
pit mode
/modo foso
in Spanish - is roughly equivalent to saying "moat mode" en English, it just makes no sense unless you know that a pit and a moat are kinda similar in a context that's not remote-controlled aircraft.)es-CL
transcribeshertz
asjertz
for phonetic purposes, even if the word in writtenhertz
in Spanish, that's sometimes necessary depending on what voice is used by the TTS service. More on that below.)"says the wrong thing"
It seems like providing voice packs in different locales can make EdgeTX more accessible to more people.
Relying on machine translations allows to provide some meaningful level of support to non-English speakers, but good translations require both language and EdgeTX proficiency.
So there will be some mediocre translations in the voice packs that receive less attention (see 👍 in #17 despite "mode fosse" in
fr-FR
and "modo foso" ines-ES
which are wrong.)Review mistakes will happen, and not all sounds will make sense to all reviewers every time and that should be okay. I think we can make the trade-off relying on unreliable machine translations acceptable by making sure that reporting errors is easy, so they get fixed over time.
"doesn't sound right"
Sometimes terms that we use are jargon that a general-purpose text-to-speech service may not render correctly. The current pipeline allows the text to be adjusted in order to compensate for that, and it's an easy thing to fix, assuming that someone reported it.
I hope that having a template to report those issues will help lowering the amount of effort folks need to put into contributing to making the voice packs better in their own language.
new locale please
That one seems obvious. If there are no voice packs for your locale, and adding them is not too hard / costly, it should be easy to request them. 🙂
new voice please
There are different voices for some of the locales already (e.g.
en_US
), so I suppose that's something that people request from time to time.default template
Of course, providing issue templates fro some common requests / problems shouldn't prevent anyone from reporting the other ones.
The text was updated successfully, but these errors were encountered: