Legal

Privacy Policy

Last updated: 3 October 2026

What we collect, and from which tools

Text-to-speech (Studio): the script text you submit, your voice and language selection, and the generated audio. Script text is sent to the TTS engine you select (Microsoft Edge or our self-hosted models) to produce audio, and is not stored longer than needed to generate and return the result.

Voice cloning (coming soon): voice cloning is currently paused. If it is re-launched, any reference audio you upload would be processed only to generate your requested speech and this policy will be updated before it goes live.

Transcription, conversion, merge, cut, denoise, voice-changer, video-to-audio tools: the audio or video file you upload is processed in memory or in a temporary location and deleted once the tool returns your result — we don't keep a library of files you've processed. Transcription runs on our own server (Whisper) and may also send short audio snippets to Google's speech recognition service; see Third parties we use below.

Video redub (Pro): the video you upload, the audio extracted from it, the transcript and translated text (including any edits you make on the review screen), your voice, language and option choices, and the generated dubbed video and audio. Audio is transcribed on our server (Whisper) and/or by Google's speech recognition service; the text is translated by Microsoft Azure Translator or, if that is unavailable, Google Cloud Translation; the new voice is generated by Microsoft Edge text-to-speech. Optional audio denoising and the AI separation that keeps the background music run on our own server and do not send your audio to a third party. How long each item is kept is described under Data retention.

Music generator: your style tags and chosen length are processed on our own server by a code-based generator to compose the track. No lyrics or audio are sent to a third-party AI service, and generated tracks are not kept after the temporary result expires.

Account and licensing: your email address if you request Pro access, submit a payment for verification, or contact support; a hashed (not raw) version of your IP address and a browser fingerprint, used only to enforce free-tier usage limits per device and to detect duplicate license activations.

How we use it

To generate the audio or text output you request, to enforce free-tier limits fairly across devices, to verify and deliver Pro license keys, to respond to support requests submitted through Contact, and to diagnose bugs when a tool fails. We do not sell your data, and we do not use your submitted text or audio to train models beyond what's needed to generate your own output.

Third parties we use

  • Microsoft Edge and self-hosted TTS/voice models — to generate speech and cloned voice audio.
  • Modal (GPU infrastructure) — to run voice cloning and music generation workloads, and, if enabled, GPU speech recognition for transcription and video redub.
  • Google — speech recognition for transcription and video redub (short audio snippets of your upload) and, as a fallback, Google Cloud Translation (text only). Google's own privacy policy applies to data it receives.
  • Microsoft Azure Translator — to translate transcript text for video redub (text only, no audio).
  • Resend — to deliver license key emails and support/contact notifications.
  • Freemius — to process Pro subscription payments (card/PayPal); Freemius handles your payment details directly, we never see full card numbers.
  • Adsterra and, where approved, Google AdSense — to serve advertising to free-tier visitors. These networks may use cookies or similar technology as described in their own privacy policies.

Data retention

Most tools process your upload only for the length of the request and delete it when your result is returned. These are the exceptions, all temporary:

  • Video redub review session: if you use the review step, your uploaded video, the transcript and translated lines, and (if you keep the background music) the separated music track are stored on our server so you can edit lines and generate again without uploading. They are deleted automatically 45 minutes after your last activity in that session; cleanup runs continuously, so please allow up to about an hour.
  • Redub results: the generated dubbed video and audio files are kept for about 10 minutes so you can download them, then deleted.
  • Connection-recovery copy: so that a dropped mobile connection does not lose a finished result, a copy of the result a tool returns to you (not your upload) is kept for up to 20 minutes under a random identifier tied to your browser session, then deleted. We use it for nothing else.

Free-tier usage counters (tied to hashed IP and fingerprint) are pruned after roughly 2 months. License records, request IDs, and support messages are kept as long as needed to provide and support the service, or to resolve a dispute.

Your choices

You can request a copy of, or deletion of, the data we hold about your license or contact history at any time — email support@voxcraft.site or use the contact form. We'll confirm and act on deletion requests within a reasonable time, except where records must be kept for fraud prevention or legal reasons.

Children's privacy

VoxCraft is not directed at children under 13, and we do not knowingly collect personal information from them.

Changes to this policy

If we materially change how we collect or use data, we'll update the date at the top of this page. Continued use of VoxCraft after a change means you accept the updated policy.

Audio tools processing

Browser tools (convert, merge, trim, denoise, normalize, volume, speed, fade, split, reverse, mono, loop, EQ, voice effects, video-to-audio, and transcription) process files you upload for the duration of the request. Output is returned to your browser. We do not use these uploads to train public AI models. Temporary processing files are deleted after the job completes.

When you use “Send to another tool,” the audio stays in your browser’s session storage for up to about 30 minutes so you can continue editing without re-uploading. That transfer data is not sent to our servers until you run the next tool.

Operator

VoxCraft is independently operated from Lahore, Punjab, Pakistan. Contact: support@voxcraft.site.

Sponsored Remove with Pro