version 0.9..4: When the translation direction is reversed, the text in the input window stay (e.g. the Czech text stays in the upper window which is now Ukrainian), and it is translated to cs (usually remains the same, but it is confusing). Suggestion: move the translated text from the output window to the input window at each reversal.

If the user says OK to giving us their data, generate and use an anonymous identifier unigue per device instead of IP address.

• better privacy
• better logging, beucase we get all the data grouped by that one user, not their various IP addresses as the phone moves around.

When the microphone icon is pressed, the input field should be erased.

The input field in the second screenshot should be empty: The current situation can be confusing - users can expect that the recorded text is appended (inserted) into the current text in the input field.

When the input text consists of several sentences (tested with idnes text with fullstops at the end of sentences), they end up joined without space after the fullstop in the output window. The browser UI does not do it, so it has to do something with the app.

New icon, "Charles Translator".

• [ ] pick one (Slack)
• [ ] get feedback for that one (Slack)
• [ ] have the author improve it based on the feedback
• [ ] check with the author that the icon conforms to the guidelines

1. press the mic button
2. say a word: "jedna"
3. then press cross to remove word "1" (now the input area is empty)
4. then say another word: "dva"
5. finally, "1 2" appear in the input area

The app crashes when clicking on some history items. Other items work OK, but I could not find any rule which items cause the crashes and which not. I have Chrome OS 100.0.4896.82, Acer N15Q8. I have tried just a few Android apps and all worked correctly, but I admit that using an old chromebook is an edge case. So maybe we can close this issue as "wont fix" (and similarly #11) if no one else encounters the same bug.

I am using a rather unusual keyboard called MessagEase. With this keyboard, diacritics are added to already written characters, i.e. I write "s", then do "add háček" and the result is "š". This does not work here, it writes a standalone "ˇ" character. Often, the following character also got malformed. I use the keyboard for several years and I have never encountered an issue like this.

version 0.9.7, Samsung M22, Android 11, Kernel 4.14.186-22735277

I translated the word "Ahoj":

Then I pressed history icon:

and then a card with the history item containing the translation of "Ahoj":

I expected that the first screenshot would appear.

version 0.9.4: when using the ASR called from the Google keyboard, it often gets truncated (sometimes even after the first word). This does not happen when speaking through the Gogole keyboard to the web version of the translation. So it must be some sync issue within the app.

Version 0.9.4 The app's "speak" buton invokes ASR, but cannot be switched from the system language (e.g. US English) to anything else. Should follow the input language of the translation.

A user reported via email (in Ukrainian) that "When the application is opened, a black window appears immediately. After some time, it starts working again." I guess it will be difficult to debug without knowing the exact Android version and whether the black screen appears always and for how long, but let's create this issue anyway, maybe there are other users experiencing the same behavior.

A dedicated mode for fluent conversation. See Google, Microsoft or Apple apps.

• interface to display running conversation
  • side by side, or mirrored halves of screen?
  • a switch for automatic synthesis after each translation
  • manual "turns" by pressing microfones (like chess clock)?
• features for conversation
  • identify replicas (speech start + language detection)

may involve other issues, like #18 .

This is just an idea for improvement. It would be nice to have one button for both actions:

• text to speech button
• play translated text

Like a small simulation of a human translator, who listens to a sentence, then translates it in mind and then says it.

For text: https://developers.google.com/ml-kit/language/identification/android

For audio I don't know whether it is public, but Google has it working very nicely in the conversation mode in their translation app. So please look whether there is a method for it that you could use. It would enable a future option for a fully automatic conversation mode, just like Google and Apple translation apps have.

The input field does not support standard Android copy-pasting features (copy / cut / paste / select all) on double-tap or using a keyboard.

I noticed that there is a paste button when the field is empty (and there is also a copy button when there is some text). But I don't think this is the best way to do it, because I often need to paste something after some existing text or copy/cut a text segment.