Documentation

Documentation

Everything you need to install WhisperARC, dictate, work with files, manage your history and solve common problems.

Download for Windows
Configure

Local, Cloud and OpenAI

Compare Local Mode, paid Cloud credits and your own OpenAI key for transcription, translation and rewriting.

Last updated

A tactile graphics card, processor, waveform ribbon and settings controls

Local Mode

Local Mode runs the selected Whisper model through whisper.cpp on the configured CPU or GPU. It is available on the Free plan.

Local Qwen translation and Assistant modes, glossary corrections, media conversion, queue state and History also run on this device.

  • Microphone and system-audio recordings remain on the computer during speech recognition.
  • Imported audio and video files are processed locally.
  • Performance depends on the active model and hardware.

Cloud

Paid plans add Cloud for online transcription with Whisper Large v3 Turbo. When a Cloud dictation starts, WhisperARC checks the available quota while audio capture begins. If validation fails, the recording stops without producing a transcription. WhisperARC reports errors without switching modes.

  • Long file jobs are split into smaller sections for reliable processing.
  • A valid paid license and internet connection are required.

Own OpenAI key

OpenAI KEY is available on every plan, including Free. Audio and text requests go directly to OpenAI using your API key. These requests use separate OpenAI billing and do not consume WhisperARC Cloud credits.

  1. Open Model and select OpenAI KEY.
  2. Paste your API key and save it. Wait for WhisperARC to verify the key.
  3. Select the OpenAI KEY card again to activate this source after verification.
  4. Choose Transcription and Assistant models separately, then use dictation or file transcription normally.
  5. To replace a verified key, remove it with the delete button and add the new key.

Shared monthly Cloud budget

Pro and Annual each include 9,000 Cloud credits per calendar month, shared across transcription, translation, Assistant and the license's devices. The balance and next reset are shown under Model > Cloud. Credits reset on the first day of each month in UTC without rollover.

One display credit represents one minute of equivalent Whisper Turbo transcription, up to approximately 150 hours for transcription alone. Translation and Assistant consume the same budget according to text usage. Local and OpenAI KEY do not consume these credits.

When Cloud credits run out, choose Local if its models are installed, or OpenAI KEY with your own account. A Cloud processing failure does not automatically retry using another source.

Switch modes

  1. Open Model and find Local, Cloud and OpenAI KEY.
  2. Select Local for on-device processing, Cloud for included paid credits, or OpenAI KEY for your own API account.
  3. If Cloud is locked, open Licenses and activate a paid license.

Compare processing modes

CapabilityLocal ModeCloudOpenAI KEY
PlanEvery planPaid plansEvery plan with a verified key
Speech recognitionThis computerOnline via WhisperARCDirectly via OpenAI
InternetNot for recognitionRequiredRequired
UsageNo Cloud quota9,000 shared credits per monthSeparate OpenAI billing
Hardware impactDepends on model and CPU/GPULower local recognition loadLower local recognition load
Failure behaviorLocal processing errorNo automatic local retryNo automatic local retry