Cantonese Context Accuracy #1916
TheBlueKingLP
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Hello, I am trying to integrate Whisper into Home Assistant. With some specific sentence/word, I am getting accuracy issue.
For example, if I say "saan1 閂 dang1 燈(turn off the lights)" in Cantonese, with Whisper I get the following output
[00:00.000 --> 00:02.200] 山燈
山燈, pronounced "saan1 dang1" Which literally means "mountain light", which makes no sense.
In Cantonese, the pronunciation is exactly the same as "turn off the lights" (saan1 閂 dang1 燈)
saan1 閂 -> to close/turn off
dang1 燈 -> (the) light(s)
Please note that the above words are not the only issue. There are other words with similar pronunciation, for example, if I say "閂房燈", I get
[00:00.000 --> 00:02.200] 山晃燈
from whisper.房(fong2, means room, implies bedroom when spoken) becomes 晃(fong2, means to sway/to shake) in Whisper
Is there any way to correct this issue?
Beta Was this translation helpful? Give feedback.
All reactions