INPUT
Hold global hotkey and record instantly.
oiper-shell v1.0
$ hotkey --hold --listen
recording: active
$ hotkey --release
transcription: 1.5s
injecting text into focused application
result: "Can you share the deployment notes before lunch?"_
TRANSCRIBE
Release key and execute low-latency local speech pipeline.
INJECT
Insert text into the currently active desktop application.
OPTIMIZE
Apply local cleanup or optional online refinement.
MODULE 01
Quick Recording & Transcription
- Global press-and-release dictation flow
- Instant text injection into active apps
- Optional local or online text cleanup
MODULE 02
Privacy-First Architecture
- Transcription runs locally on your machine
- Audio and activity logs stay on-device
- No outbound data without explicit choice
MODULE 03
Flexibility & Control
- Local or online optimization paths
- Backend preference: auto, cpu, gpu
- Multiple LLM options for extra accuracy
SECURITY BLOCK // LOCAL PROCESSING
DATA REMAINS ON DEVICE
Transcription executes locally. Audio and activity logs are retained on your machine for full ownership and predictable privacy.SECURITY BLOCK // OPTIONAL ONLINE
ENABLE WITH YOUR API KEY
Online optimization stays off until you configure a provider. You can set base URL, key, and model name, then disable at any time.SETTINGS.CONFIG
speech.model = selectable; downloadable; switchable
backend.mode = auto | cpu | gpu
provider.base_url = custom
provider.api_key = user_managed
provider.model_name = configurable
optimization.route = local | online
advanced.llm_transcription = enabled_for_technical_terms
PERFORMANCE / 30 SEC ENGLISH AUDIO
OiPer Desktop1.5s
Lemonfox API3.27s
Python Faster-Whisper3.55s
OpenAI Whisper 1 API6.46s
OiPer achieves the fastest transcription while keeping your data path transparent and user-controlled.