Jutusta · speech platform
ETEN
Voice
Speech to Text
Speech to Text
  • 01 From file Upload an audio or video file.
  • 02 Live Speak into the mic — text appears instantly.
  • 03 Meetings Teams and Meet meeting recordings, with Zoom coming soon.
Translate
Translate
  • 01 Text translation Speak and read the translation.
  • 02 Simultaneous Hear the translation in a synthetic voice.
Integrations
Integrations
  • 01 For AI agents Connect Claude or another AI assistant to Jutusta
  • 02 API REST API and docs for developers
Pricing

Privacy Policy

Last updated: 2026-11-08

This privacy policy describes what personal data the Jutusta service (jutusta.ee) collects, how it is used, and how you can exercise your rights. The policy is drafted in accordance with the General Data Protection Regulation (GDPR, EU 2016/679) and the Estonian Personal Data Protection Act.

1. Data controller

The data controller is Häälmeedia OÜ, registry code 17602750, address Tartu mnt 80, Tallinn 10112, Estonia. We are the controller for accounts used for personal purposes and the content added to them, and for the data we process for our own purposes (for example account management, billing, security and accounting).

For company accounts and other business use (see section 1 of the Terms of Use), the controller of the personal data added to the Service is the customer — the company or institution — and we are the processor. That processing is governed by the data processing agreement between us and the customer and by the customer's own privacy notice; in that case this policy applies only to the data we process for our own purposes.

For data protection inquiries, contact privacy [ät] jutusta [punkt] ee .

2. What data we collect

2.1 Account data

If you create an account with an email address, we store your name, email address and a hash of your password (not the password itself). When you sign in via Google or Microsoft, we receive the following from the sign-in provider: first and last name, email address, profile picture URL, and the provider's internal user identifier. We also store the OAuth tokens issued by the provider (access token, refresh token, id token) to keep your session active and re-authenticate as needed.

2.2 Audio and transcription data

When you use the service, we store:

  • Uploaded and recorded audio and video files — files you send for transcription or noise removal (the web app, the API, the Teams and Meet integrations);
  • Transcription results — text, timestamps, speaker diarisation, job metadata;
  • Voice clone source material — if you create voice clones, we store the source audio and user-provided text prompts;
  • Real-time speech-recognition audio — streaming audio is forwarded to our processing server for transcription; we do not persist the stream itself.
  • Saved speaker voiceprints — if you choose to save a speaker in a transcription, we store a numeric signature of their voice (a voiceprint) so the same speaker can be recognised in your future transcriptions. A voiceprint is biometric personal data; it may be saved only with the explicit consent of that speaker themselves (section 5.1 of the Terms of Use). We use it solely to label speakers in your own transcriptions and never share it. You can permanently delete any voiceprint at any time in your account settings.

2.3 Technical data

Session data (session token, session creation and last-activity timestamps, IP address, user agent) is stored in our database to keep your session secure. At the Cloudflare infrastructure layer, request IP addresses and user agents are logged and used for error detection and abuse prevention. Logs are stored in our R2 bucket in the Western Europe region and auto-expire after 30 days.

2.4 Browser local storage

Preferences such as transcription language, voice selection, and UI layout are kept in your browser's localStorage. This data never leaves your browser and never reaches our servers.

2.5 API usage data

When you create API keys, we store only the SHA-256 hash of the key (never the key itself) along with the key's name, creation timestamp and last-used timestamp. API request metadata (request type, timestamp, status) is retained for service monitoring.

When you authorize a third-party application or AI agent to access your account through our OAuth/OIDC flow, we store the registered client (its name and icon), the consent you grant, and the access and refresh tokens issued to it, so it can act on your behalf until you revoke it. You can review and revoke connected applications at any time in your account settings.

2.6 The "Jutusta Dictation & Captions" browser extension

Microphone audio. The extension opens the microphone only when you start dictation, captions or a recording yourself (keyboard shortcut, toolbar button or context menu). In online mode the audio streams to our speech recognition server for transcription; the stream itself is not stored. In local-model mode the audio never leaves your computer.

Tab audio and video. While captioning or recording, we read the audio of the tab you started it on, and for a recording its video as well. Recordings are written to your browser's own storage (OPFS), not to our servers. Only the audio reaches our servers, and only when you press "Transcribe" in the library. Recording never starts on its own.

Meeting detection. On meet.google.com, teams.microsoft.com, teams.live.com and *.zoom.us the extension checks whether a call is open so it can offer a name for the recording. That information is not sent anywhere; it stays in the browser.

Selected text. Read-aloud sends the text you selected to our text-to-speech API. No other page content is read or transmitted.

Data kept in the browser. The extension's chrome.storage.local holds your preferences (language, engine, interface language) and your sign-in tokens. We do not use chrome.storage.sync, so nothing is synchronised to your Google account. Signing out deletes the tokens; deleting a recording in the library removes it permanently from your computer.

Third parties. If you download the optional Estonian speech model it comes from huggingface.co (Hugging Face, Inc.); the download exposes only ordinary request information (IP address, user agent) to them. The extension bundles no analytics or crash-reporting service.

2.7 Confirming that you represent a company

When you confirm with your ID-card that you represent a company on a company account, we keep as evidence your first and last name, personal identification code and country, the ID-card's authentication certificate, the signature over the statement you confirmed (the company, the role you entered and the version of the data processing agreement), the certificate's validity response, and the time, IP address and user agent of the confirmation. The name and personal code come from the ID-card certificate. If you confirm with Mobile-ID, we pass the phone number and personal code you enter to SK ID Solutions AS. We do not store the phone number, and the personal code of an abandoned attempt is deleted within two days.

We use this data only to check that you act for the company, and so that we can later show who confirmed on the company's behalf. It is not a way to sign in. Jutusta administrators see your name and the last four digits of your personal code; other users of the company do not. We check the certificate's validity with the validity service of SK ID Solutions AS, which issues ID-card certificates.

The legal basis is legitimate interest (Art. 6(1)(f)): both the company and we have an interest in no one acting for a company without the right to. We keep the data while the company account is open, and after it is closed for the limitation period for claims. The confirmation stays even if you delete your user account, because it is the company's evidence.

2.8 People in recordings

When a user uploads or records a meeting, call or other conversation (including through the Microsoft Teams or Google Meet integration), we also process data about the people who take part in it or are mentioned in it: their voice and what they say, the text of the transcript, translation and summary, and the speaker separation, and, if the user chooses, the speaker's name or voiceprint. We receive this data from the user, not from the person. The user must inform participants about the processing in advance (section 4 of the Terms of Use).

For personal use we are the controller of this data and process it to provide the service to the user on the basis of legitimate interests (Art. 6(1)(f)); a voiceprint only with the speaker's own consent. For business use the controller is the user's company or institution and we process the data on its behalf; in that case, contact it first to exercise your rights. The data is kept for the periods in section 7. You have the rights described in section 8, including the right to object to the processing.

3. Legal bases for processing

We process personal data on the following GDPR Article 6 grounds. For saved speaker voiceprints as special-category biometric data, we also rely on GDPR Article 9(2)(a). This section covers processing for which we are the controller; for business use, the customer determines the legal basis.

  • Performance of a contract (Art. 6(1)(b)) — account creation and management, providing transcription and speech synthesis services, serving API requests;
  • Explicit consent (Art. 6(1)(a); for voiceprints also Art. 9(2)(a)) — saving a voiceprint and creating a voice clone only with the consent of the person whose voice it is. Consent can be withdrawn at any time; voiceprints and voice clones can be permanently deleted at any time in account settings;
  • Legitimate interests (Art. 6(1)(f)) — ensuring security, preventing abuse, retaining error logs, confirming company representatives (section 2.7) and, for personal use, processing data about people in recordings (section 2.8). Our and the user's interest is in stable, secure and useful service operation;
  • Legal obligations (Art. 6(1)(c)) — retention duties imposed by accounting and tax law (e.g., retaining invoices for 7 years).

We do not currently carry out other non-essential processing, such as analytics, on the basis of consent. Should that change in the future, we will request consent separately.

4. Purposes of processing

  • providing the service (transcription, speech synthesis, speech recognition, voice clones, and recognising saved speakers with consent);
  • account management and authentication;
  • checking company representatives and proving that a data processing agreement was concluded;
  • ensuring security and preventing abuse;
  • improving service quality (error log analysis);
  • complying with legal obligations.

We do not use your content (uploaded audio, transcripts, voice clones, or saved voiceprints) to train models.

5. Sub-processors and third parties

To provide the service we use the following sub-processors:

  • Cloudflare, Inc. — infrastructure services (Workers, R2 object storage, D1 database, Durable Objects). R2 and D1 are locked to EU jurisdiction, meaning data at rest is stored in European Union servers. Workers serverless execution runs at the Cloudflare PoP closest to the user — for Estonian and EU users that means EU PoPs (Tallinn, Helsinki, Stockholm);
  • Google LLC — authentication via OAuth (Google Sign-In);
  • Microsoft Ireland Operations Ltd. — if you choose the optional "maximum accuracy" summary engine for Teams meetings, your meeting transcripts are processed by Microsoft Azure (Azure OpenAI service) within the European Union. Microsoft does not use this content to train AI models and does not share it with OpenAI. The content is not stored, except that content automatically flagged as potential service abuse may be retained for up to 30 days and reviewed only within the European Economic Area. This engine is used only if you select it during Teams setup; otherwise summaries are generated on our own servers in Estonia;
  • Google Cloud EMEA Ltd. — optional Google Gemini engines (Gemini model on the Vertex AI platform in data centers within the European Union), used only at your own choice. For live speech translation, Gemini is used only if you choose it as the translation engine and give separate consent for it; it receives the speech-recognition text to translate, not the audio. Without that consent, translation runs on the model on our own servers in Estonia. If you choose the optional Google Gemini summary engine for Google Meet meetings, the same model also processes your meeting transcripts; that engine is used only if you select it during Meet setup, otherwise summaries are generated on our own servers in Estonia. Google commits that the AI processing itself also stays in the EU. Google does not use this content to train AI models. The content is not stored, except that content automatically flagged as potential service abuse may be retained for up to 30 days;
  • Plus Five Five, Inc. (Resend) — sending email: account and service emails (for example email verification, password reset, invoices) and meeting transcripts and summaries when they are delivered to you by email;
  • Häälmeedia OÜ on-premises hardware (Estonia) — speech recognition and synthesis AI models run on our own hardware in Estonia. This is not a sub-processor in the legal sense, as it is the same controller's infrastructure.

For the full list with legal bases, see the Sub-processors page.

Third-party applications you authorize. If you connect a third-party application or AI agent to your account (via our OAuth/OIDC flow), we disclose to it the identity details covered by the access you approve — your name and profile picture, your email address — and we process the content you direct it to create or transcribe. Such a provider is an independent controller for what it does with that data, not our sub-processor; its handling of your data is governed by its own privacy policy, not this one. Only authorize applications you trust, and revoke access from your account settings when you no longer use them.

6. Data location and international transfers

  • Object storage (R2) — audio files and voice clone source material are stored in Cloudflare R2, configured with EU jurisdiction. Data at rest remains in EU servers;
  • Database (D1) — user, session, transcription and account data is stored in Cloudflare D1 with EU jurisdiction;
  • Workers infrastructure — requests are served from the Cloudflare PoP closest to the user. For our Estonian and EU user base this means EU PoPs (Tallinn, Helsinki, Stockholm). Persistent data remains in the EU regardless;
  • Error logs — error logs are stored in our R2 bucket in the Western Europe region and auto-expire after 30 days. Logs contain only technical error information (HTTP status, error, request path, IP, user agent), not transcription or audio content;
  • Speech recognition and synthesis — speech recognition and synthesis AI models run on our own hardware in Estonia;
  • Optional Azure summary engine — if you have selected the optional "maximum accuracy" summary engine for Teams meetings, meeting transcripts are processed by Microsoft Azure within the European Union (EU Data Boundary, Sweden). Content automatically flagged as potential abuse is reviewed only within the European Economic Area;
  • Optional Gemini translation and summary engine — if you have chosen the Gemini translation engine for live speech translation and consented to it, or selected the optional Google Gemini summary engine for Google Meet meetings, the relevant text is processed on Google Cloud's Vertex AI platform in data centers within the European Union; Google commits that the AI processing itself also stays in the EU;
  • Google OAuth — Google LLC may transfer authentication data (your Google account identifier, name, email, profile picture) to the United States. The legal basis for this transfer is the Standard Contractual Clauses (SCC) and the EU–U.S. Data Privacy Framework, to which Google has self-certified.
  • Microsoft sign-in — Microsoft may transfer authentication data to the United States. The legal basis for this transfer is the Standard Contractual Clauses (SCC) and the EU–U.S. Data Privacy Framework, to which Microsoft has self-certified.
  • Email — Resend sends email through servers in the United States. The legal basis for this transfer is the Standard Contractual Clauses (SCC).

7. Retention periods

Data typeRetention period
User profile (name, email, picture)Until account deletion; the account is closed if it has not been used for 24 months (we tell you 30 days in advance)
Session data (token, IP, user agent)7 days, refreshed on activity
Sign-in provider (Google, Microsoft) OAuth tokensUntil account deletion
API key hashesUntil key revocation
Uploaded and recorded audio and video filesWith their transcript, no longer than 12 months after last use; the audio can be deleted separately earlier
Transcripts, translations and summariesUntil you delete them, but no longer than 12 months after last use
Voice clones (source + clone)Until you delete them, but no longer than 12 months after last use
Saved speaker voiceprintsUntil you delete them, but no longer than 12 months after last use
Synthesised speechUntil you delete it, but no longer than 12 months after last use
Webhook subscriptionsUntil you delete them or the account is closed
Error logs (R2, Western Europe)30 days (auto-deleted)
Intermediate results and logs on processing servers (speech recognition, speech synthesis)Intermediate results up to 24 hours after processing, logs up to 14 days
Database backups (restore points)30 days; deleted data disappears from backups within that time
Consent records (time, IP address, user agent)Until account deletion
Company representative confirmations (name, personal code, certificate, signature)Until the company account is closed, then for the limitation period for claims; kept even if the user account is deleted
Accounting records (invoices etc.)7 years (statutory)

"Use" means opening, editing or downloading content, or using it for speech synthesis or speaker recognition. Retention of content covered by a data processing agreement (a company account or a separately signed agreement) is governed by that agreement: audio and video files are deleted 12 months after processing has finished unless the agreement says otherwise, and other content is kept until the customer deletes it; the 12-months-unused rule does not apply to it.

8. Your rights

Under the GDPR, you have the following rights:

  • Right of access (Art. 15) — request what personal data we process about you;
  • Right to rectification (Art. 16) — request correction of inaccurate data;
  • Right to erasure (Art. 17) — request deletion of your data ("right to be forgotten");
  • Right to restrict processing (Art. 18);
  • Right to data portability (Art. 20) — request export of your transcription and voice-clone data in machine-readable form;
  • Right to object (Art. 21) — object to processing based on legitimate interests;
  • Right not to be subject to automated decision-making (Art. 22) — we do not make automated decisions that produce legal effects concerning you.

To exercise your rights, write to privacy [ät] jutusta [punkt] ee . We respond within 30 days.

If you believe your data has been processed unlawfully, you have the right to lodge a complaint with the Estonian Data Protection Inspectorate (Andmekaitse Inspektsioon).

9. Account deletion

You can request deletion of your account by writing to privacy [ät] jutusta [punkt] ee . An account is also closed and its content deleted automatically if you have not signed in or used the service through the API for 24 months and have no active monthly subscription; we tell you by email 30 days in advance. This does not apply to accounts linked to a company account. Upon deletion:

  • your user profile, OAuth tokens and sessions are removed from our database;
  • transcription results, voice clones, saved voiceprints and API key hashes are removed with the account;
  • uploaded audio files are removed within 30 days at the latest;
  • deleted data disappears from database backups within 30 days;
  • audit entries in error logs expire automatically within 30 days;
  • a company representative confirmation (section 2.7) stays as the company's evidence for the period in section 7;
  • accounting records (if any) are retained for the 7-year statutory period.

10. Children

The service may only be used by people who are at least 18 years old. We do not knowingly collect data from minors, and no voice clone or voiceprint may be created from a minor's voice. If we become aware that we have collected a minor's data, we will delete it promptly.

11. Security measures

We apply the following technical and organisational measures:

  • TLS 1.3 encryption for all traffic;
  • OAuth tokens stored securely in the database;
  • API keys stored only as SHA-256 hashes;
  • EU jurisdiction for Cloudflare R2 and D1;
  • Session cookie is HttpOnly, Secure, SameSite=Lax;
  • access roles are role-based and minimised;
  • regular dependency security review;
  • data is deleted automatically according to the periods in section 7;
  • a procedure for handling personal data breaches is in place: where required, we notify the Data Protection Inspectorate within 72 hours and the people and business customers affected without undue delay.

12. Changes to this policy

We may update this privacy policy from time to time. We will notify you of material changes by email at least 30 days before they take effect. For less significant changes, we will update the "Last updated" date above.

13. Contact

Häälmeedia OÜ
Tartu mnt 80
Tallinn 10112, Estonia
Registry code: 17602750
Email: privacy [ät] jutusta [punkt] ee

Terms · Privacy · Cookies · Withdrawal · DPA · Sub-processors · About · Blog
Jutusta · voice cloning, transcription, and translation · jutusta.ee
Häälmeedia OÜ · registry code 17602750 · Tartu mnt 80, 10112 Tallinn, Estonia · email: · phone: