Privacy Policy
How Dialect collects, uses, and protects your information, how voice and language are processed, and your rights to export or delete everything.
Last updated August 2026
Overview
This Privacy Policy explains how Dialect Inc. ("Dialect", "we", "us") handles the personal information you share with us when you use the Dialect service at trydialect.com. We have written it to be readable, not just legally complete.
Dialect is a language practice product. It is not a healthcare provider, not a covered entity under HIPAA, and it does not collect health data. The information we hold is used to run the service and for nothing else. We do not sell it and we do not use it for advertising.
What we collect
We collect only what we need to make Dialect work for you:
- Account information: your name, email address, and a secure credential used to sign in.
- Session data: the language and level you chose, the topic of the conversation, the turn-by-turn transcript of your practice conversation, the practice report generated after the session, and your daily usage count.
- Voice input: during a live conversation, your speech is recognized in your browser and converted to text. The text, not the audio, is what we send to our language model to generate the partner's next line. Audio recordings of your voice are off by default and only saved if you explicitly enable the recording feature in Settings.
- Usage information: basic information about how you use the product, such as which languages you practice most often, to help us improve Dialect. This information is not linked to your transcripts.
- Communications: messages you send through support forms or by email.
- Billing information: if you subscribe to Speaker or Fluent, your card details are handled directly by our payment processor and never pass through or are stored on Dialect's servers.
How voice is processed
Voice input is handled with care and with the minimum necessary footprint:
- Speech recognition runs in your browser using the Web Speech API or an equivalent browser-native capability. Your audio is processed on your device and by your browser's speech provider (typically Google or Apple depending on your browser). Dialect does not receive raw audio.
- The text transcript of each turn is sent from your browser to a Dialect server route, which calls the language model API to generate the AI partner's next response. Your API key is stored on the server and is never exposed to the client.
- The AI partner's spoken response is generated by the ElevenLabs voice API. The text of each partner turn is sent to ElevenLabs, and the audio is streamed back to your browser. Your ElevenLabs API key is stored on the server and is never exposed to the client.
- At the end of a session, the full transcript is sent to the language model API to generate your practice report. Only your session transcript is used for this, not any data from other users or sessions.
- Audio recordings of your own voice are off by default. If you enable recordings in Settings, the audio file is stored in your account and you can delete it at any time.
Neither the language model provider nor ElevenLabs is permitted to use data we send them for training or improving their public models. We have data processing agreements in place with each.
Subprocessors
We use a small number of third-party services to operate Dialect:
- OpenAI (or equivalent): generates the AI partner's responses during a conversation and your practice report after each session. Data sent: the turn-by-turn transcript of the current session.
- ElevenLabs: converts the AI partner's text responses into spoken audio. Data sent: the text of each partner turn.
- Stripe: processes payments for Speaker and Fluent subscriptions. Data sent: billing information you enter at checkout. We never see your card number.
- Neon (or equivalent cloud database provider): stores your account, conversation transcripts, and practice reports securely at rest.
We maintain a current list of subprocessors and will notify you materially before adding a new one that processes your session data. Questions about subprocessors can go to privacy@trydialect.com.
How we use your information
We use the information you share to:
- Run your practice conversations and generate your practice reports.
- Keep your account secure and your conversation history available to you.
- Send receipts and billing notifications if you are on a paid plan.
- Respond to support requests.
- Improve Dialect's features using aggregated, anonymized patterns. We do not use your individual transcripts to improve the service.
We do not use your data for advertising. We do not build a profile of you to sell or share with third parties.
Your rights
You have the following rights regarding your information:
- Export: you can download a copy of all your conversation transcripts and practice reports from Settings at any time.
- Deletion: you can delete individual conversations or your entire account, including all transcripts and reports, from Settings. Deletion is permanent.
- Correction: you can update your name and email address from Settings.
- Objection: you can contact us to ask questions or object to any processing described in this policy.
To exercise any of these rights, use the Settings page in the app or write to us at privacy@trydialect.com. We will respond within 30 days.
Data retention
We keep your account information and session data for as long as your account is active. If you delete your account, we remove your personal information and all conversation transcripts and practice reports within 30 days.
Some records, such as billing history, may be retained for longer where the law or a legitimate business need requires it. Anonymized usage aggregates are retained indefinitely but cannot be traced back to you.
Children
Dialect is not intended for anyone under the age of 13. In the European Economic Area and the United Kingdom, Dialect is not intended for anyone under 16. We do not knowingly collect personal information from children below those ages. If you believe a child has created an account, please write to us at privacy@trydialect.com and we will delete it promptly.
Security
We protect your data with encryption in transit (TLS) and at rest, strict access controls, and isolated database storage. API keys for the language model and ElevenLabs are stored server-side only and are never exposed to the browser.
To report a security concern, write to security@trydialect.com.
Changes to this policy
We will update this policy when our practices change in a meaningful way. We will notify you by email and by posting the updated policy with a new effective date. Continued use of Dialect after an update means you accept the revised policy.
Contact us
For questions about privacy, write to privacy@trydialect.com. For security matters, write to security@trydialect.com.
Dialect Inc., 80 White Street, Floor 2, New York, NY 10013, United States.