Privacy Policy

This policy explains how Digisensus.com ("Digisensus", "we") handles personal data in the Digisensus online services used by the Digisensus Recorder app ("the Service"). Contact us about anything in it at [email protected].

The short version

• The App records on your Mac. We only receive what you choose to send to the Service. If you use your own servers or another provider, recordings never reach us.
• When you use the Service, we transcribe or summarise your recording for you and don't keep it. The one exception: unless you turn the setting off, the speech recorded by your own microphone in a transcription paid with free credit may be kept to improve our speech recognition, as raw data for at most 30 days and in pseudonymised form for at most 12 months. Other participants' audio is never used for this.
• We don't sell data, don't use it for advertising, and don't identify people by their voice.
• Everything runs on servers in the European Union.

Who is responsible

• For your account, your use of credit, and training data (see below), we are the controller.
• For the content you send to be transcribed or summarised, you (or the organisation you act for) are the controller, and we are your processor under the Data Processing Terms. If you use the App only for yourself, for personal or household purposes, the GDPR may not apply to you, but it still applies to us.

What we process, why, and on what basis

• Account: email address, accepted Terms version and time, credit balance and history. Why: To run your account and bill requests. Basis: Contract, Art. 6(1)(b). Kept: While the account exists; billing records afterwards without your email address, as accounting law requires.
• Devices: device name, a one-way hash of each access token, last use, last IP address and one-way device identifier. Why: To keep your account secure and prevent abuse. Basis: Contract and legitimate interest in security, Art. 6(1)(b) and (f). Kept: Until you sign the device out, then up to 12 months.
• Usage: time, model, audio length or token count, cost of each request (not its content). Why: Billing, capacity planning, fixing faults. Basis: Contract and legitimate interest, Art. 6(1)(b) and (f). Kept: While the account exists; afterwards without your email address.
• Content of a request: audio, instructions, transcript, summary. Why: To give you the transcript or summary. Basis: Contract, Art. 6(1)(b) (for you); we act as processor for your content. Kept: Only while the request runs.
• Training data: speech recorded by your microphone and its transcript, in transcription requests paid with free credit while the training setting is on, with model, time and length. Why: To develop, train, test and improve Digisensus speech recognition models. Basis: Legitimate interest, Art. 6(1)(f), see below. Kept: Raw data at most 30 days, pseudonymised data at most 12 months from the request.
• Sign-in requests: email address, device name, IP address, one-way device identifier, invite code. Why: To sign you in and stop abuse. Basis: Contract and legitimate interest, Art. 6(1)(b) and (f). Kept: Up to 30 days.
• Referral program data (see below). Why: To prevent referral fraud. Basis: Legitimate interest, Art. 6(1)(f). Kept: 12 months after the referral is closed.
• Messages between you and us. Why: To answer you. Basis: Contract and legitimate interest. Kept: Up to 3 years after the last message.
• Marketing email preference. Why: To send product news if you asked for it. Basis: Consent, Art. 6(1)(a), or the law on existing customers. Kept: Until you unsubscribe.

Improving speech recognition

After you accept the Terms and confirm your email address, the App shows the setting "Help improve Digisensus speech recognition". It is on unless you turn it off, and you can turn it off at any time in the App's settings. While it is on, we may keep the speech recorded by your microphone, and its transcript, from transcriptions paid with free credit, as described in the Terms of Service. We may do so, but we don't have to, and we may keep only some requests.

We rely on our legitimate interest in making speech recognition more accurate, especially for smaller languages such as Lithuanian, which is also what pays for the free tier. We have weighed this interest against your rights, and we protect them as follows:

• We use only the speech recorded by your microphone, never the audio of the other participants in your conversations.
• Only requests paid with free credit are used. Summaries, purchased credit, your own servers and other providers are never used.
• Raw data is kept for at most 30 days, only to pseudonymise it, and then deleted.
• Pseudonymised data has names, numbers, addresses and other identifying details removed or masked. It is linked only to a random ID; the key to your account is kept separately. It is kept for at most 12 months from the request. A voice can still be recognisable, so we treat it as personal data, and your rights still apply to it.
• We train models only on pseudonymised data. A speech recognition model learns to turn speech into text: it does not store your recordings, does not create a profile of your voice, and is not used to recognise who is speaking.
• Training data is stored in the EU, separately from account details. Only the people who work on our models, bound by confidentiality, can access it, to pseudonymise it and to correct transcripts.
• We never use training data for advertising or to make decisions about you.
• We may make trained models available to others, including as open-source models, but never the training data itself.

To stop it, turn the setting off: none of your later requests is kept. To have what we already hold deleted, write to [email protected] from your account's email address. Your free credit is not affected. A model already trained cannot be undone, but deleted data is not used again.

If you were in a recorded conversation

Someone who uses Digisensus Recorder may have recorded a conversation you took part in and sent it to us to transcribe. That person is responsible for recording it lawfully and for telling you about it. We process the recording only to return the transcript to them, and we do not keep your voice for training. We do not know who you are: we receive only the audio.

If you have questions, or want to exercise your rights, write to [email protected].

Messages we send

• Service messages (sign-in links, security notices, changes to our Terms, prices and the Service) are sent by email and in the App, because they are part of the Service.
• Messages in the App about new features and releases are shown to every copy of the App. The App fetches them from our server without identifying your account. You can turn their notifications off in macOS settings.
• Product news by email is sent only if you agreed to it, or as the law allows for existing customers, and every such email has an unsubscribe link.

Referral program

To prevent abuse of the referral program, we process the IP address and a one-way device identifier of the person who invites and of the new user when they accept the Referral Program Terms, sign up and use Digisensus, together with the new user's email address domain and the times of sign-up and use. The app derives the device identifier from the Mac's hardware identifier and makes it one-way on the Mac; we never receive the hardware identifier itself, and we store only a keyed hash of what the app sends. Automated checks may hold a referral for review; a person at Digisensus then decides. We process this data on the basis of our legitimate interest in preventing fraud. The IP addresses and device identifiers kept for a referral are deleted 12 months after the referral is closed. The person who invited you sees only a masked form of your email address (like j***@example.com) and the state of the referral.

Who we share data with

We don't sell personal data or share it for advertising. We use these service providers, who process data only on our instructions:

• Cloudflare, Inc., Google.

We may also disclose data when the law requires it, or to protect our rights in a legal claim.

Transfers outside the EU

The Service runs in the EU. Cloudflare and Google may process data in the United States. Those transfers are covered by the EU-US Data Privacy Framework and the European Commission's standard contractual clauses.

Your rights

You have the right to access your data, correct it, delete it, restrict its use, receive it in a portable format, and object to processing based on our legitimate interest, including model training. Where we rely on your consent, you can withdraw it at any time. Write to [email protected]; we answer within one month. We may ask you to confirm that the request comes from you, for example from your account's email address.

You can also complain to the State Data Protection Inspectorate of Lithuania (vdai.lrv.lt), or to the data protection authority where you live or work.

We do not make decisions about you based solely on automated processing that have legal or similarly significant effects.

Deleting your account

Choose "Delete account" in the App's settings. Your email address and devices are removed at once. Billing records are kept without your email address. Your remaining credit is kept for 30 days under a keyed hash of the address (not the address itself), so that signing up again with the same address carries it over. Training data is kept for its normal period (at most 30 days raw, 12 months pseudonymised) unless you ask us to delete it too: write to [email protected] before you delete the account, from its email address.

Security

We use encryption in transit, store access tokens and device identifiers only as one-way hashes, limit access to people who need it, pseudonymise training data, and keep it apart from account details.

Children

The Service is not for people under 18.

Changes

We publish every version of this policy at /privacy, with earlier versions still readable, and tell you about important changes in the App or by email.