added more 3 vars for LLM post-processing transcript

[khanhicetea]

Before Submitting This PR

Please confirm you have done the following:

• I have searched existing issues and pull requests (including closed ones) to ensure this isn't a duplicate
• I have read CONTRIBUTING.md

If this is a feature or change that was previously closed/rejected:

Human Written Description

We have more variables in prompt when post-processing:

$time_local : current time local string
$current_app : use for identify the purpose of work, can teach the LLM about in prompt template
$short_prev_transcript : a trim about 200 words last transcript in same app (or expired in 5 mins), so LLM has more context about what next.
$language : Selected transcription language for Whisper models (e.g., "en", "zh-Hans"), or "auto" for other models (Parakeet, Moonshine)

Related Issues/Discussions

Fixes #168
Discussion:

Community Feedback

Testing

Using this prompt:

Current App : ${current_app}
Current Local Time : ${time_local}

Clean this transcript:
1. Fix spelling, capitalization, and punctuation errors
2. Convert number words to digits (twenty-five → 25, ten percent → 10%, five dollars → $5)
3. Replace spoken punctuation with symbols (period → ., comma → ,, question mark → ?)
4. Remove filler words (um, uh, like as filler)
5. Keep the language in the original version of ’Transcript' (if it was french, keep it in french for example)

Preserve exact meaning and word order. Do not paraphrase or reorder content.

Return only the cleaned transcript.

Last transcript in same app (for better understanding context) : ${short_prev_transcript}

Transcript:
${output}

Screenshots/Videos (if applicable)

AI Assistance

• No AI was used in this PR
• AI was used (please describe below)

If AI was used:

• Tools used: Claude Opus 4.5 Thinking
• How extensively:

[khanhicetea]

btw, I plan to have 1 more var named '${language}' which is picked by user in settings for multi-lang model. Should I push it ?

[cjpais]

Sure. If you can, can you also add documentation of this feature in the readme or somewhere

[khanhicetea]

• Added $language (for Whisper models), fallback into 'auto' for everything else
• Added docs into readme (Experimental > Post Processing section)

I tested on my mac works (not sure about Win and Linux)

This template can provide more ways to do with transcribe (fix typo, grammar, even translation or dual translation output) 😄

[VirenMohindra]

hmm so macOS impl returns app name via localizedName (ie "Safari"), but Windows uses GetWindowTextW which returns the window title (ie "Stack Overflow - How to do X - Google Chrome"). this means ${current_app} has different semantics per platform, and ${short_prev_transcript} context keying by app name would create separate entries per browser tab on Windows. maybe consider using GetModuleFileNameExW or querying the process name instead?

[VirenMohindra]

the change logs the entire prompt including transcription text AND app name. this is a privacy concern since the log file could contain sensitive speech content

recommend reverting

[VirenMohindra]

crate is deprecated in favor of objc2 which is already a transitive dep in the project. maybe we look into objc2-app-kit or a simpler approach for the macOS active app detection

[VirenMohindra]

nit: entries are inserted in start() and removed in stop(), but CancelAction doesn't clean up its entry. over many cancelled recordings this could accumulate stale entries. minor since the values are small strings, but worth adding cleanup in the cancel path too

[VirenMohindra]

solid feature idea, the context variables will make post-processing much more useful. some notes below~

merge conflicts + coordination with #706

this PR is currently conflicting against main and also has a semantic conflict with #706 (structured outputs). that PR splits prompts into system prompt + user content, so the variable substitutions here (${current_app}, ${time_local}, etc) will need to work in both the legacy ${output} path and the structured output system prompt path. recommend merging #706 first and rebasing this on top