Toggle Mode
Start recording, speak at your pace, then stop when you are done. This is the normal hands-free flow for longer dictation, notes, replies, and drafts.
Features Overview
See exactly how speech becomes final text.
Every capture moves through clear stages: local Whisper transcription, your rules, optional AI cleanup, final output, then history for search and reprocessing.
Audio
→
Whisper
→
Rules
→
Optional LLM Cleanup
→
Final Output
→
History / Reprocessing
Dictation Modes
Three ways to capture what you want to say.
Use AimeFlux like a normal dictation tool, hold a shortcut for intentional capture, or save thoughts to Collections when you do not want to interrupt your current app.
Push to Talk
Hold a key or key combo while you speak, then release to finish. It is built for short commands, quick edits, and moments when you want capture to be deliberate.
Quick Capture
Save a one-off thought or idea to history without inserting text into the app you are using. Quick Captures land in Collections so you can organize them later.
Quick Capture is also a lightweight entry point into
Collections:
capture the thought now, then sort it into the right project, client, or idea collection when you have time.
History & Processing
Keep transcripts useful after the first run.
Search, processing diffs, reprocessing, and parent comparisons answer the practical question: what happened to this transcript, and what can I still do with it?
Find previous transcripts without hunting for files.
The first run is not the end. History gives you a searchable place to revisit transcripts, inspect output, and reuse earlier material.
- Search across earlier transcripts and outputs
- Review the mode and settings used at the time
- Keep prior results available for later reuse
Collections
Turn saved transcripts into an organized workspace.
Collections help history become more than a timeline. Group related captures, review uncategorized items in Inbox, and let rules keep recurring work organized automatically.
Group transcripts by project, client, idea, meeting, or workflow
Use Inbox to review uncategorized captures when you have time
Create automatic rules from source, mode, keywords, regex, or metadata
Use local semantic search to find related history and collection context
Mark sensitive collections private so MCP tools cannot search or summarize them
Rules & Modes
One capture flow. Different output rules.
Replacements, vocabulary, cleanup prompts, translation settings, and auto activation work together as reusable mode behavior.
Same voice capture. Different output behavior.
Modes let each recurring task use its own language, translation setting, prompts, and cleanup rules without changing how you capture.
- Set language and translation behavior per mode
- Keep prompts and cleanup behavior scoped to the task
- Use one app across coding, writing, support, and bilingual work
Privacy & Local AI
Local transcription first. AI only where you choose it.
AimeFlux keeps the core path on your Mac: transcription, rules, vocabulary, modes, and encrypted history. Cleanup can stay no-cloud through local-compatible providers, or use OpenAI, Anthropic, and xAI only when you configure them.
Core path stays local
Whisper transcription, replacements, vocabulary, encrypted history, and reusable modes stay on your Mac.
No-cloud cleanup
Ollama and compatible local providers can handle AI cleanup without sending text to cloud APIs.
Optional cloud providers
OpenAI, Anthropic, and xAI (Grok) are available only when a mode or cleanup setting uses them.
Private agent scopes
Private modes and collections stay local in the app while hidden from MCP transcript, collection, and semantic search tools.
Keys stay protected
Provider keys and local history encryption keys are stored in macOS Keychain. Cloud cleanup sends transcript text only.
Default flow
Audio
→
Whisper local
→
Rules local
→
Encrypted history local
→
Private MCP scopes
Cloud boundary
Audio is not uploaded. Cloud cleanup sends transcript text only when enabled.
Community Packages
Reusable settings you can move or share.
Packages bundle modes, vocabularies, cleanup rules, and terminology so useful setups can move between machines or be shared with others.
Medical terminology packs
Bilingual workflow bundles
Engineering vocabulary and cleanup kits
PR writing and support-response modes
Import and export for sharing across devices or teams