SuperwhisperSuperwhisper
May 13

Separate voice language option modes

Currently, voice language is part of the mode, but it's not ideal for people who speak multiple languages. We can set this option to Automatic, but it's not working all the time, especially with local models. For me, it often outputs results in English anyway when I'm using my native language. I tested it on Ultra models both local and cloud, cloud works most of the time well. In some cases, when I define a custom mode with additional LLM processing, I need to create two copies: one for English and one for my native language, using the same prompts etc. This is inconvenient. I was thinking that maybe it's a good idea to separate the voice language from the mode and allow us to select the language in the dictation window, similar as we can do currently with modes. It could work like a quick access list, allowing us to define/select a few languages in settings and switch between them (to avoid a list of 100+ languages), or select automatic option. What do you think?
PendingPending
Jan 14, 2026

This would be a great feature. I tried some of the multi-language models and they just don't work quite well enough to be used in production. What I would really like to see is the option to configure different keys for different languages.

Jun 4, 2025

Yes, I'm switching between Chinese and English constantly, but somehow it works on the iOS.

May 29, 2025

My current workaround for this is to add this instruction to the custom prompt in each mode: - If user message starts or ends with instruction sentence: "Language specific language", translate whole message to that language, and remove that instruction sentence from final output.

May 20, 2025

multi lingual input recognition would be great!!