IMO, if user switches to Korean or Zhuyin keyboard.
They should know the symbol on the keyboards as expectation.
So I think "aahh" for Korean "ㅏ" in English is reasonable.

Not sure if I understand comment 4 correctly. 

I was indeed suggesting "aahh" as a reasonable way to describe the Korean letter "ㅏ" in an English or French voice, instead of silence. The problem here is that currently no names or descriptions are announced by ChromeVox when touch-exploring these keys. It just says "button", so besides the word "button" itself, there's silence in place of any names or descriptions.

This may also happen for other ChromeVox - Keyboard language pairs, besides {English, French} ChromeVox voice on {Zhuyin, Korean} keyboards that I managed to verify.

I got your meaning.
I think ChromeVox lack the sound of "ㅏ".
We should works with ChromeVox team to add the sounds of these characters.

Could you help to collect all characters used in keyboard layouts.
And ask ChromeVox to provide at least two sounds of them.
One is the sound in English. The other is the sound in native language.

This sounds like a text-to-speech (TTS) issue then. 

Ideally TTS voice for every language should be able to say the name of foreign characters of popular writing scripts when there's an alphabet or syllabary, and at least say something descriptive instead of silence for non-alphabet ones. Maybe emojis need similar treatment, too.

I'm reassigning to ChromeVox team first. Please help assess and forward to the TTS team as appropriate. Thanks.

The issue isn't quite as described above. The crucial piece that isn't supported is dynamic language switching. Given korean, for example, we would need:
1. a hint from the VK that the button's text is in Korean.
2. the ability for the tts engine to switch to a given language (Korean in this case) for a single piece of text

2 is not supported by Google tts and 1 is probably not exposed correctly.

I'm afraid I was not referring to dynamic language switching. Don't think that's required here...

We just need the TTS ability to read letter names (for those with an alphabet or syllabary e.g. reading the names of Greek letters, Russian letters, Japanese kanas, Korean stand-alone hangeul jamos) or at least some meaningful descriptive hints instead of silence for less "finite" character set (e.g. just say "some Chinese characters" or the like).

As an example, currently French TTS voice can say the names in a French voice of Greek letters found on a Greek VK (e.g. saying "alpha button" when seeing the "α" key). This can be extended in the same manner, without the need for dynamic language switch, esp. when VK use cases are usually single standalone letters.

I believe this is a feasible request to the TTS team. This can then be extended to emojis for the VK and will aid a11y a great deal. Each emoji has a standard Unicode description phrase e.g. "grinning face", but I'm not sure if these are read when the emoji is encountered or there's just silence, not to mention whether they're localised.

Btw, from my understanding of TTS, there's usually fallback mechanism to deal with out-of-vocabulary situations, including foreign tokens like proper names, neologisms and so on, all still in the voice language. What we need here should fit in the realm of out-of-vocabulary handling by TTS.

In fact, we purposely don't want to switch the voice language. We want the letter names to be uttered in whatever way and accent understandable to speakers of the current voice language (e.g. English voice saying the name of alpha "α" in English way, not necessarily the same as how it's called in native Greek).

Consultation in progress with the TTS team to confirm the issue, possible solutions and their feasibility.