The clearest difference is where speech recognition runs. SuprFlow uses a downloaded WhisperKit model on an Apple Silicon Mac. Aqua Voice says its Avalon model processes audio in the cloud and that it has no offline mode. Aqua's cloud design brings a free trial, real-time and cross-platform features, and documented team controls. SuprFlow's local speech path is useful when a firm cannot send dictated audio to a provider. Neither choice removes the need to inspect text formatting, retained copies, account data, and the destination app.
| Decision | SuprFlow | Aqua Voice |
|---|---|---|
| Speech recognition | On-device WhisperKit model in the Mac app. | Audio processed in the cloud by Avalon; official Mac page says no offline mode. |
| Optional rewrite | Default cleanup is local. Cloud API formatting sends transcript text and instructions to a chosen provider; Ollama can be local if its endpoint is local. | Cloud service handles dictation and editing features. Verify exactly what each plan and privacy setting sends and retains. |
| Training and retention | SuprFlow 1.0.0 does not show a content-training submission on the local path. Hosted formatter terms belong to the chosen provider. Its public privacy policy now distinguishes the local and hosted paths; verify the signed release. | Aqua's terms say it will not use content for model training, or allow a third party to do so, without explicit agreement. Its privacy policy says transcript retention depends on Privacy Mode and technical/session metadata may still be collected. |
| Mac support | Current requirements say macOS 14+ on Apple Silicon. | Current Mac page says its download supports Apple Silicon and Intel; the same subscription also covers other platforms it lists. |
| Teams | No centrally enforced formatter or privacy control was confirmed in SuprFlow 1.0.0. Ask for deployment evidence. | Aqua's Mac page lists Team controls including enforced Privacy Mode and a Business tier with SSO/SCIM and zero-data-retention offering. Ask for the current agreement and scope. |
| Offer | SuprFlow is a one-time $125 personal purchase. | Aqua's Mac page lists a 1,000-word free trial, Pro from $8/month billed annually, Max, Team, and custom Business terms. Check the live checkout rather than comparing list prices alone. |
If your rule is “audio cannot leave the laptop”
Aqua's current Mac FAQ answers this plainly: audio is processed in its cloud service. Its Privacy Mode and contractual no-training terms address retention and model improvement, but do not turn the recognition step into on-device processing. If your organization's rule prohibits external audio processing regardless of retention, start with SuprFlow's local speech setting or another verified local app. Test the signed build with synthetic speech and capture network behavior. SuprFlow's current homepage and privacy policy distinguish local recognition from optional hosted formatting and other network traffic.
If the rule instead allows a contracted processor with zero retention and team administration, Aqua deserves a serious evaluation. Its documented Privacy Mode changes transcript collection, and its Team and Business plans provide controls SuprFlow has not demonstrated publicly. A procurement team should request the current data-processing terms, security report, subprocessor list, residency details, and evidence that the chosen plan enforces the required setting. A public SOC 2 claim is a prompt to inspect scope, not a blanket approval for client data.
If your main concern is AI training
Separate processing, retention, and training. Aqua says it processes audio in the cloud. Its terms contain an explicit conditional no-training commitment absent consent; its privacy policy says transcript storage changes with Privacy Mode. SuprFlow recognizes audio locally but its optional hosted formatter sends transcript text to the selected provider, whose own terms decide retention and training. Release telemetry and license traffic are separate. Do not call either app “never sends data” based on a training statement.
A fair trial with your own vocabulary
Use the same Mac, microphone, destination app, and fictional 90-word passage. Include a surname, acronym, technical term, currency amount, date, and spoken correction. Record each app's version, mode, connection, time to usable text, and material corrections. Test Aqua's cloud mode against the SuprFlow local mode as a workflow comparison, while noting they use different processing locations. If SuprFlow hosted formatting is in scope, test it separately and name the provider. Repeat in the real app where staff write; reliable insertion can matter as much as transcript quality.
Questions people ask
Does Aqua Voice work offline?
Its current Mac FAQ says no. The Avalon model processes audio in the cloud. Confirm this for the exact current version and plan before making a deployment decision.
Does Aqua use dictation to train AI?
Its terms say content is not used to train Aqua or third-party AI models unless the customer explicitly agrees. Its privacy policy separately describes Privacy Mode and possible transcript retention when that mode is disabled. Read both documents and the applicable enterprise agreement.
Is SuprFlow automatically suitable for banking or legal work because its speech model is local?
No. The institution still needs a signed-build review, allowed formatter and endpoint, history and clipboard policy, telemetry payload review, account and update traffic, and approved destination storage.
Related guides
- Where does my dictation go?
- Does SuprFlow use dictation for AI training?
- SuprFlow vs Wispr Flow
- Private dictation for financial firms