Error
There is an error when downloading the tiny model for Spanish. It will not let me download it (Download failed: HTTP 404 for encoder_model.ort)
Also, when choosing the default language I want to translate from (i.e., Español), when I close and reopen the app, my preferences for that language are not saved. Also, it doesn’t seem to pick up any Spanish being spoken. If I use the default language (Japanese, I think), it does detect audio of the speaker speaking. But since they aren’t speaking Japanese, it’s not useful.
Even though no words in the audio are being detected, I don’t see a location where I can see the saved recordings either. It isn’t under “saved transcripts”, unless that is only for when words are detected.
When clicking “start”, the 0:00 on the right never increases. I’m not sure the cause.
The green bar indicating it is detecting sound, it will show words, but only on the default language. And I have to click “TV Mode” first and then exit
Also, when choosing the default language I want to translate from (i.e., Español), when I close and reopen the app, my preferences for that language are not saved. Also, it doesn’t seem to pick up any Spanish being spoken. If I use the default language (Japanese, I think), it does detect audio of the speaker speaking. But since they aren’t speaking Japanese, it’s not useful.
Even though no words in the audio are being detected, I don’t see a location where I can see the saved recordings either. It isn’t under “saved transcripts”, unless that is only for when words are detected.
When clicking “start”, the 0:00 on the right never increases. I’m not sure the cause.
The green bar indicating it is detecting sound, it will show words, but only on the default language. And I have to click “TV Mode” first and then exit
Show more
Response from developer
Thanks for your feedback, we will correct the issues you mentioned. Some are already addressed with the current release. For Spanish you can use the moonlight base model as the tiny one is not available from the moonlight voice project actually.