Text-to-Speech configuration during software distribution

[jo-elimu]

@tuancoltech Just sharing some thoughts here on the various Text-to-Speech scenarios we might encounter when preparing devices for distribution. As well as ways to make the software installation as frictionless as possible for newcomers (even non-technical users).

1. The device supports the language, but a TTS voice has not been downloaded.
   • In a future version of the Content Provider app, we can prompt user to download a compatible TTS voice.
2. A TTS voice has been downloaded, but its language has not been configured in the device settings.
   • If we rely on the default settings of the device, this step will be needed. But if we handle it programmatically, the user can probably configure whatever they want here.
3. The system language is different from the chosen TTS voice language.
   • In most cases, devices are configured to use the TTS voice of the default system language. If this system language is different from the Content Provider language, we will have to choose it programmatically.
4. The system language and the TTS voice language are the same.
   • In this case, it will work if the system language on the device is the same as the one chosen in the Content Provider app. If not, we will have to choose it programmatically.
5. The device does not support TTS for the language.
   • We can probably ignore this last point (TTS language not supported) for now, since we have freedom to choose devices that will work best. But in the future (for better device compatibility) we might have to add instructions to download an open-source TTS voice from a 3rd-party app somewhere.

My recommendation would be to not require the user to change the system language or the TTS language of the device. Optimally, yes, the system-wide language should be the same as the language being taught to the student, but it doesn't have to be a requirement. The way I see it, we only have to check one thing: Has a TTS voice been downloaded for the language chosen in the Content Provider or not. If not, assist the user with the download.

Thoughts?

[tuancoltech]

@tuancoltech Just sharing some thoughts here on the various Text-to-Speech scenarios we might encounter when preparing devices for distribution. As well as ways to make the software installation as frictionless as possible for newcomers (even non-technical users).

1. The device supports the language, but a TTS voice has not been downloaded.

   • In a future version of the Content Provider app, we can prompt user to download a compatible TTS voice.

2. A TTS voice has been downloaded, but its language has not been configured in the device settings.

   • If we rely on the default settings of the device, this step will be needed. But if we handle it programmatically, the user can probably configure whatever they want here.

3. The system language is different from the chosen TTS voice language.

   • In most cases, devices are configured to use the TTS voice of the default system language. If this system language is different from the Content Provider language, we will have to choose it programmatically.

4. The system language and the TTS voice language are the same.

   • In this case, it will work if the system language on the device is the same as the one chosen in the Content Provider app. If not, we will have to choose it programmatically.

5. The device does not support TTS for the language.

   • We can probably ignore this last point (TTS language not supported) for now, since we have freedom to choose devices that will work best. But in the future (for better device compatibility) we might have to add instructions to download an open-source TTS voice from a 3rd-party app somewhere.

My recommendation would be to not require the user to change the system language or the TTS language of the device. Optimally, yes, the system-wide language should be the same as the language being taught to the student, but it doesn't have to be a requirement. The way I see it, we only have to check one thing: Has a TTS voice been downloaded for the language chosen in the Content Provider or not. If not, assist the user with the download.

Thoughts?

@jo-elimu I think you have covered all the possible scenarios.

The way I see it, we only have to check one thing: Has a TTS voice been downloaded for the language chosen in the Content Provider or not. If not, assist the user with the download.

That's technically feasible. We can do so with below code snippet:

tts = TextToSpeech(this@SwipeUpDownActivity.applicationContext) { status ->
            if (status == TextToSpeech.SUCCESS) {
                val designedLocale = Locale("th", "TH")
                val voices = tts?.voices
                val matchingVoice = voices?.any { voice ->
                    voice.locale.language == designedLocale.language &&
                            voice.isNetworkConnectionRequired.not() &&
                            voice.features.contains("notInstalled").not()
                } ?: false
                Log.d("", "in downloaded: $matchingVoice")
                if (!matchingVoice) {
                    val intent = Intent(TextToSpeech.Engine.ACTION_INSTALL_TTS_DATA)
                    intent.setPackage("com.google.android.tts")
                    context.startActivity(intent)
                }
            } else {
                // TTS initialization failed
            }