Spokenly can run entirely on your device: with a local Whisper or Parakeet model, its privacy policy says "Your audio recordings and transcriptions never leave your device", and its Local Only Mode blocks the app's network access apart from model downloads and localhost. It can also send audio to the cloud, either to Spokenly's own backend for Pro cloud models or straight to a provider you hold an API key with. So its safety for private or work information depends on which path you set up. It is a proprietary app, and we found no published security audit on its pages.
Publisher note: SuprFlow publishes this guide and is one of the products discussed. Competitor details come from each company's own pages, checked on 1 October 2026. Prices and features change; confirm before you buy.
This page reports what Spokenly's own pages say. It is not a security audit; we have not tested Spokenly's network traffic.
What do Spokenly's own pages say?
As of 1 October 2026. Spokenly's privacy policy and terms both show a last-updated date of 1 October 2026.
| Question | What Spokenly states | Source |
|---|---|---|
| Who operates it? | "Spokenly is operated by LLP Inviral Tech, a company registered in Kazakhstan" | Terms of Use |
| Local models | "Your audio recordings and transcriptions never leave your device." Local models are free, need no account and "Works completely offline". | Privacy policy, Pricing |
| Pro cloud models | Audio is sent "to our backend for transcription purposes. Our backend does not store these recordings." Pro lists GPT-4o Transcribe, Deepgram Nova 3 and Soniox. | Privacy policy, Pricing |
| Your own API keys | Audio and text go "directly from your device to the provider you selected. This data never passes through Spokenly's servers." | Privacy policy |
| AI text processing | When a Mode uses AI Instructions, "your transcription and any optional context for that Mode may be sent to AI services for text processing, even if transcription itself runs locally." | Privacy policy |
| Training | For Spokenly-managed cloud features, it says it prefers "options that do not use customer data for model training where available". | Privacy policy |
| Analytics | In-house analytics of "button clicks and page views inside app" that never include audio or transcribed content. Local Only Mode blocks them. | Privacy policy, Local Only Mode |
| Who owns transcripts? | "You retain ownership of any content you create using the service, including your transcriptions." | Terms of Use |
| Audits and certifications | None found on the pages we read. | — |
| Platforms | macOS, Windows, Linux and iOS. | Pricing |
| Is the source public? | "Spokenly is not open source and is not available on GitHub." It calls itself "a proprietary application". | Open source statement |
The privacy policy names the third parties it may use: OpenAI, Deepgram, Soniox, Cartesia, ElevenLabs, Groq, Mistral AI, xAI, Microsoft Azure, Google Cloud, AssemblyAI, OpenRouter, fal.ai and Cerebras for transcription and AI, plus RevenueCat, Polar and Apple for subscriptions and payments.
Which of the three paths keeps speech on your device?
| Path | Where audio goes | Who sets the rules |
|---|---|---|
| Local model (free) | Nowhere; it stays on the device | Spokenly's privacy policy |
| Your own API key (free to Spokenly) | Straight to the provider you chose | That provider's terms, under your account |
| Pro cloud model | Spokenly's backend, then its chosen provider | Spokenly's policy and its providers |
The local path is the one for sensitive material; the other two are ordinary cloud processing. With your own keys you pay and contract with the provider directly. Spokenly pricing has the plan details.
Text formatting is a separate path. Spokenly's Modes page says a Mode with AI Instructions can include context such as the clipboard, the focused app, text around the cursor or the active browser URL, processed by Spokenly's built-in option, OpenAI, Anthropic or an OpenAI-compatible provider with your own key. Local speech plus a cloud AI Instruction is not fully local.
What does Local Only Mode do?
Spokenly's Local Only Mode documentation says it disables "all network activity except local model downloads", which come from Spokenly's CDN or Hugging Face. It allows connections to localhost, so locally hosted models still work, and it blocks the minimal analytics the app otherwise sends. If an online model is selected, "transcription fails because network requests are blocked". Spokenly's Sync Settings page adds that Local Only Mode disables settings sync.
That failure shows you a cloud setting is still on. This is Spokenly's description; we have not audited it.
What do Spokenly's pages not say?
- That cloud processing is never used for training. The policy says Spokenly prefers no-training options "where available", and that its third-party providers "may process and/or store certain data for their own purposes". That is weaker than a no-training commitment. If your work needs one, ask Spokenly in writing. What a no-training agreement should cover lists the questions.
- Where cloud requests are processed. We found no data region or residency statement.
- Security detail. The policy mentions "industry-standard security measures" but names no encryption standard, audit, certification, DPA or BAA.
- How synced API keys are protected. Sync is a Pro feature and every category is off until you enable it. When on, the Sync Settings page says "Synced data is stored on Spokenly's servers. If you choose to sync API keys, they are stored there as well." It does not describe how they are protected. Transcription history and audio recordings are not synced.
How can you check for yourself?
- Download from the official site. Spokenly's docs point to spokenly.app/download. How to check a signed Mac release shows how to see who signed the app you installed.
- Review permissions. Open System Settings → Privacy & Security and note what Spokenly has been granted.
- Pick a local model and turn on Local Only Mode.
- Open each Mode you use. Note whether it has AI Instructions, which provider handles them, and which context it includes.
- Leave sync off, especially for API keys.
- Run a network test with fictional text. Turn off Wi-Fi and dictate "Call Dana Whitlow about invoice 4471 for 2,950 dollars on Friday" in each Mode. If it fails offline, something in that Mode uses the network.
When is Spokenly a reasonable choice, and when does something else fit better?
Spokenly is a strong choice when you want free, unlimited on-device dictation, one switch that blocks network access, or the same app on Mac, Windows, Linux and iPhone.
Something else may fit better if:
- Your reviewer needs vendor audit reports or a BAA. Wispr Flow's security overview lists a SOC 2 Type II report and a BAA, though it processes audio in the cloud. See Is Wispr Flow safe?.
- You need source code you can read. Spokenly is proprietary. Open-source apps such as Handy publish theirs; Does open-source dictation prove privacy? explains what that does and does not tell you.
- You want a written no-training commitment rather than a stated preference.
SuprFlow is another local-speech option, not a clear upgrade. It recognises speech on the Mac with a downloaded WhisperKit model, and its Default formatting is local cleanup of punctuation, capitalization and numbers. It is also proprietary, has no Local Only switch, and uses the network for licence checks, model downloads, updates and usage telemetry. It costs $125 once with no free tier, is Mac-only (Apple Silicon, macOS 14 or later), and has no audit report. SuprFlow vs Spokenly compares them.
Questions people ask
Is Spokenly legit?
Yes. It is operated by Inviral Tech, publishes a privacy policy, terms and docs, and its terms give a 14-day money-back guarantee on purchases made on spokenly.app.
Is Spokenly open source?
No. Spokenly's own page says it "is not open source and is not available on GitHub" and calls it a proprietary application.
Does Spokenly work offline?
Yes, with local Whisper or Parakeet models. Spokenly's pricing page says local models work "completely offline". Cloud models, your own API keys and cloud AI Instructions need a connection.
Does Spokenly store my recordings on its servers?
Its privacy policy says its backend does not store recordings sent to Pro cloud models. The one exception it describes is a ChatGPT plugin, which keeps an encoded audio copy and transcript that expire within an hour. Third-party providers may process or store data under their own terms.
Related guides
- Spokenly pricing
- Spokenly alternatives
- Wispr Flow vs Spokenly
- Superwhisper vs Spokenly
- SuprFlow vs Spokenly
- On-device dictation vs no internet