FAQs that are ready for people and assistants.

Drag VoxChimp into Applications to install.

1. Open your Downloads folder and double-click VoxChimp.dmg.
2. In the window that appears, drag VoxChimp into the Applications folder.
3. Open Applications and launch VoxChimp. The first time you launch it, macOS may ask you to confirm opening an app downloaded from the internet. Click Open.

VoxChimp needs a few macOS permissions to work from anywhere:

• Microphone access to hear your dictation.
• Accessibility access to paste text at the active cursor.
• Input Monitoring to detect your activation hotkey.

VoxChimp requires macOS 14 Sonoma or later. It runs natively on Apple Silicon and Intel Macs.

VoxChimp works system-wide with apps that accept text input. That includes Slack, Gmail, VS Code, Notion, iMessage, Pages, browser text fields, and most other places where your Mac cursor can type.

Transcription and translation run on local models, so core dictation does not need an internet connection. Agent Mode needs a connection when it talks to a cloud provider, but can also be configured with local model providers for offline-style workflows.

Agent Mode in VoxChimp lets you speak an instruction while selected text or clipboard content is passed to your chosen assistant profile. It is built for fast, repeatable workflows like rewriting an email, summarising notes, or turning rough ideas into structured output.

Dictation captures what you say and pastes text at your cursor. Agent Mode captures an instruction, combines it with selected text or clipboard content, and asks a configured AI provider to transform it before you paste the result.

Chimps are configurable profiles for formatting or transforming text. A chimp can be tuned for email, Slack, code comments, clean transcription, translation, or another repeatable writing job.

Agent Mode can use configured cloud providers such as Anthropic, OpenAI, Gemini, Grok, or GitHub Copilot, and local model providers such as LM Studio or Ollama. The right option depends on whether you prioritise speed, privacy, cost, or model quality.

Dictation pastes the transcribed text where your cursor is. In Agent Mode, VoxChimp can use selected text or clipboard text as input, then replace the clipboard with the transformed result so you can paste it where you need it.

Copy or select the text you want to transform, hold the Agent Mode shortcut, speak the instruction, and let the configured chimp produce the result.

Yes. Search-backed chimps can use a configured search provider such as SearXNG, Brave, Tavily, SerpAPI, or Serper for current facts, rates, news, or other live information before generating an answer.

Yes. VoxChimp can translate dictated text locally, including workflows where you speak in English and output text in another language.

Add your OpenAI API key in VoxChimp Settings under Providers. Credentials are stored locally and used only when an assistant action is triggered.

• Open Settings, then Providers.
• Paste your OpenAI API key into the OpenAI field.
• Test the connection.
• Pick a default model for Agent Mode.

Open Settings, choose Providers, paste your Anthropic API key, test the connection, and choose the Claude model you want your chimps to use.

LM Studio is a free desktop app for running local AI models with a built-in server. Set it up once and VoxChimp uses it for every Agent Mode request.

1. Download LM Studio from lmstudio.ai and install it.
2. Pick a chat model in LM Studio that fits your Mac's RAM (see "Which local model should I choose?" for guidance).
3. Open the Developer tab in LM Studio and start the local server. Note the base URL (default: http://localhost:1234/v1).
4. In VoxChimp, open Settings > Agent and select "LM Studio (Local)".
5. Paste the base URL, pick your loaded model, and test the connection.

Ollama is a free, open-source tool that runs large language models locally. It's the quickest way to get Agent Mode working without an API key.

1. Install Ollama: brew install ollama (or download from ollama.ai).
2. Pull a model: ollama pull llama3.2.
3. Ollama runs automatically on localhost:11434. No extra setup needed.
4. In VoxChimp, open Settings > Agent and select "Ollama (Local)".
5. Default model is llama3.2. Hit "Test connection" and you're good to go.

SearXNG is a free, self-hosted search engine that runs in Docker on your Mac. It gives search-backed chimps live web results without needing an API key.

1. Install Docker Desktop from docker.com/products/docker-desktop. It's free for personal use. Open it once to finish setup.
2. Download our ready-made setup files: searxng-setup.zip. Unzip anywhere on your Mac.
3. Open Terminal (in Applications > Utilities), navigate to the unzipped folder, and run docker compose up -d. This downloads and starts SearXNG.
4. In VoxChimp, open Settings > Web Search and select "SearXNG".
5. SearXNG restarts automatically with Docker Desktop, so you only set it up once.

Prefer not to install Docker? You can use a paid search provider (Brave, Tavily, SerpAPI, or Serper) with just an API key, configured in the same settings panel.

Match the model to your Mac's RAM. Use smaller models for short formatting tasks; use larger or quantised models for longer rewrites, reasoning, or multi-step Agent Mode workflows.

• 8 GB RAM: Qwen 2.5 7B, Phi-4 Mini 3.8B, or Nemotron Nano 4B.
• 16 GB RAM: Gemma 3 12B or Gemma 4 26B MoE (quantised).
• 24 GB RAM: Gemma 4 31B, Qwen 3.5 27B, or Gemma 4 26B MoE (full).
• 32+ GB RAM: any model at higher quantisation (Q8_0) for best quality.

On Apple Silicon, both Ollama and LM Studio use Metal automatically, so larger models often run faster than RAM alone would suggest.

This is almost always a permissions issue. Fix it in System Settings:

1. Open System Settings.
2. Go to Privacy & Security > Input Monitoring.
3. Make sure VoxChimp is in the list and enabled.
4. If VoxChimp isn't listed, click the + button and add it from Applications.
5. Restart VoxChimp.

VoxChimp needs Accessibility permission to paste text at the cursor. Fix it in System Settings:

1. Open System Settings.
2. Go to Privacy & Security > Accessibility.
3. Make sure VoxChimp is in the list and enabled.
4. If VoxChimp isn't listed, click + and add it from Applications.
5. Restart VoxChimp.

Check both microphone permission and your input device:

1. Open System Settings.
2. Go to Privacy & Security > Microphone and enable VoxChimp.
3. Go to Sound > Input and confirm the correct microphone is selected.
4. Speak and watch the input level meter to confirm audio is reaching macOS.

Try these tips:

• Speak clearly and at a moderate pace.
• Reduce background noise.
• Use a quality microphone; external mics often work better than built-in.
• Position the microphone closer to your mouth.
• For domain-specific terms, add them to your Vocabulary in Settings > Vocabulary so VoxChimp normalises them after transcription.

If you're still seeing accuracy issues, send us a message describing the problem.

Change the activation key in Settings:

1. Click the VoxChimp icon in your Dock or menu bar.
2. Open Settings.
3. Under "Hotkey", select your preferred key.
4. Available options: Left Option, Right Option, Left Control, Right Control, Caps Lock.

Manage your subscription from VoxChimp:

1. Open VoxChimp and go to Settings > License.
2. Click Manage subscription.
3. This opens the Stripe customer portal where you can cancel.
4. Your access continues until the end of your billing period.

If you have trouble, email [email protected] and we'll help.

When you turn on Translate after transcription, VoxChimp translates your words into another language using Apple's on-device translation. That needs macOS 15 or later and the target language pack installed. If the pack is missing, translation fails or times out and VoxChimp shows a warning: a sticky banner on the Dictate tab and an error card in Settings > Translation, both with the same recovery button.

1. Click Download. On macOS 26 and later this button reads Open Language & Region.
2. On macOS 15 to 25 this opens the Translate app, where you add the language. On macOS 26 and later Apple removed the standalone Translate app, so it opens System Settings > General > Language & Region > Translation Languages instead. Add the language there.
3. Return to VoxChimp. It rechecks the moment you bring it back to the front and clears the warning once the pack is ready.

This is about the language VoxChimp translates into, not the language you speak.

Translate to English (in Settings > Translation) is different from translating into another language. It transcribes speech in any language straight to English, on-device, using WhisperKit, not Apple translation. Because only WhisperKit can do this, the option is hidden when:

• your transcription engine is Apple Speech or Parakeet, since neither can translate to English, or
• the language you speak is already English.

If the WhisperKit model is still downloading or loading, VoxChimp tells you so and asks you to try again shortly, rather than quietly transcribing in the original language. To use it, wait for the model to finish, or pick a WhisperKit model in Settings > Transcription.

Your recognition language is the language you speak. It is separate from translation: recognition is the input, translation is the output. Open the language picker (titled Languages) and either:

• leave Auto-detect on to let VoxChimp identify the language for you, or
• turn Auto-detect off and search the list for the language you speak.

VoxChimp's default WhisperKit engine understands the full set of around 100 languages. Apple Speech and Parakeet cover fewer, so for a less common language choose WhisperKit in Settings > Transcription and use Auto-detect or select the language directly.

Live transcription shows your words as you speak. Choose how in Settings > Recording > Live transcription: Off, Show inline in bar, Show on recording bar, or Stream to cursor (the default). It runs on-device and follows your transcription engine; while a Whisper model is still loading, VoxChimp uses Apple Speech for that recording and may show "Using Apple Speech while Whisper finishes downloading." If live text does not appear, check for one of these:

• "Live transcription needs Speech Recognition access." Grant it in System Settings > Privacy & Security > Speech Recognition, and make sure Microphone is allowed too.
• "Live transcription is unavailable for this language." The live preview cannot handle that language for this take, but your final transcript is unaffected.
• You may briefly see "Live transcription model is still loading." or a "Live transcription failed" message while a model loads. Wait a moment, then try again.

You can also switch the live transcription mode, or set it to Off, if you would rather not see text while recording.

If you switch Whisper models in Settings > Transcription and start recording straight away, VoxChimp may hold off loading the new model until after your recording. Compiling a model is heavy, so deferring it avoids contention and dropped audio. For that one recording VoxChimp uses whatever is ready: a model already loaded, or Apple Speech (which shows "Using Apple Speech while Whisper is unavailable."), so your hotkey never blocks. This is expected. To record with the model you just chose, wait for it to finish loading first.

When you release the hotkey, VoxChimp finishes the transcript in a few quick steps shown on the recording bar: Finishing audio, Transcribing, Cleaning up, then Formatting. For a long take it adds a reassuring note such as "Working through it", "Almost there!", or "Wow, that was a long one!". Live dictation reuses what it already transcribed while you spoke, so this is usually fast, but a very long recording can still need a short moment to finish. If it seems stuck for much longer than usual, send us these details so we can help:

• your app version,
• the selected model,
• the live transcription mode,
• the recognition language, and
• whether a model was loading at the time.

Email [email protected] and we'll take a look.

No. VoxChimp dictation and transcription run on-device. Agent Mode sends text only after you explicitly ask a provider-backed chimp to process it. The Markdown help content on this page is public website copy and never includes licence keys, emails, or personal audio.

Dictation, transcription, translation, and speaker labelling all run on-device, so nothing leaves for those. Three opt-in features send text (never audio): Agent Mode sends your command plus any selected or clipboard text to the provider you chose, web search sends the query, and cloud voices send the text to be spoken. Licence activation sends your licence key and an anonymised device identifier, and app analytics are off unless you opt in. The complete ledger is at Private by Design.

VoxChimp includes a History view for reviewing and searching recent transcriptions, with details such as when the transcription happened, how many words it contained, and where it was pasted.

Insights summarise usage patterns such as transcription volume and productivity trends. Voice Health is informational and based on vocal pattern analysis over time; it is not medical advice and should not replace professional care.

Yes. VoxChimp offers a free trial with full access, then a Pro plan for continued use. See the pricing page for current rates.