Privacy Policy
How Lyrictomelody AI protects your privacy
Your lyrics and generated takes stay private by default
Lyrictomelody AI provides AI-assisted lyric-to-melody generation for songwriters and teams. We only collect the minimum data needed to authenticate users, manage credits, and operate the service.
Last updated: October 12, 2026
Private by default
Your lyrics, prompts, and outputs are private to your account; downloads require you to be signed in. Files we hand to AI providers for processing are fetched through unguessable, unlisted storage links used only for that purpose.
Retention controls
We keep files only as long as needed to provide downloads (e.g., 7 days on Starter, longer on paid plans). You can delete projects and files anytime from your dashboard.
User control
You may request deletion of your profile data or billing records at any time.
Information we process
- Account data: email, encrypted auth ID, and profile credit balance.
- Billing data: customer ID, payment history, and charge metadata.
- Contact form data: name, email address, message, and the IP address provided by Cloudflare when you submit a support request. Contact requests are stored in Airtable so our team can respond and prevent abuse.
- Operational data: lyric length, generation duration, selected settings, and error codes for debugging.
- Content you submit: lyrics, prompts, MIDI, and audio you upload (including recordings, beats, stems, and voice samples) and the outputs generated from them. A beat you upload to AI Vocal Hooks for Beatmakers stays in your storage under the same retention rules and is used only to build your demo mix and aligned vocal; it is not sent to the AI vocal provider.
- Rights-confirmation records: when you confirm you have the rights to an upload or permission to use a voice, we record your account, the time, the version of the confirmation wording, the tool, and the file paths it covered. We keep these records for as long as your account exists and afterwards as needed to handle complaints and disputes.
- Cookie choices: your analytics-cookie preference is stored in your browser (see Cookies & analytics).
- Marketing email records: which marketing emails we queued or sent to you, when, a copy of what was sent, whether SES reported it delivered, opened, clicked, bounced or complained about, the checkouts you started and whether they were completed, and your unsubscribe choice (see Marketing emails).
How we use the data
- Authenticate users and allocate credits.
- Process payments and send invoices through our payment processor.
- Improve melody generation quality and guard against abuse/fraud.
Projects & confidentiality
- All requests are encrypted in transit (TLS).
- Inputs and outputs may be temporarily stored to complete processing and provide downloads during the retention period.
- We sign Data Processing Agreements (DPAs) upon request for teams with compliance requirements.
Voice data (training, conversion and reference clips)
- A person's voice can identify them. In voice training, Voice Reference conversion, MIDI to Vocals and Change Lyrics, the recordings you submit are used only to perform the task you requested. We do not use them to build voices for, or offer them to, other users.
- You must have the voice owner's permission for any voice you submit; you confirm this before each such job (see the Terms). If you believe your voice was used without your permission, follow the complaint procedure in the Terms or email us and we will investigate and, where appropriate, remove the model, clip, and outputs.
- Uploaded training audio is processed to build your voice model and is deleted per the retention rules above once training completes.
- The resulting voice model (a set of parameters, not the original recording) is retained until you request its deletion, independent of the source audio's own retention period.
- Trained voice models are private to your account by default; we do not currently offer a way to share a trained voice model with other users.
- Song With My Voice: the short recording you upload is used only for that one song. Your lyrics and style choices are sent to MiniMax (through Replicate) to write the song. Your recording and the generated song are then sent, through short-lived signed links, to our voice-conversion service on Modal, which returns the converted vocal and the instrumental; we mix them into the finished song. We delete our copies of your recording and of the intermediate files as soon as the song is finished or fails, and we keep only the finished song for the retention period. We do not save your voice for reuse. Modal currently keeps the converted vocal and instrumental in its output storage until they are cleared; we delete them on request (email us), and we do not promise a fixed deletion time yet. Replicate removes the generated files of API requests about an hour after they finish and keeps its own request records under its policy.
- We do not use your private uploads (lyrics, audio, or voice recordings) to train our own general-purpose models unless we obtain your separate, explicit permission.
Third parties & AI processing providers
We rely on vetted service providers (processors) that follow industry privacy standards:
- Cloudflare: hosting, delivery network, database (D1) and file storage (R2).
- Google: sign-in (Google OAuth / One Tap) and Google Analytics.
- Whop: new payments and subscriptions. Waffo and Creem may continue processing subscriptions originally purchased through those providers.
- Inngest: background job orchestration (job identifiers and status, not file contents).
- Amazon Web Services (Amazon SES, Sydney region): sending our marketing emails and reporting what happened to them (see Marketing emails).
- Airtable: contact-request storage and support operations.
- Microsoft Clarity and Ahrefs Web Analytics: product analytics (see Cookies & analytics).
To generate your outputs, the specific lyrics, prompts, audio, or voice recordings you submit for a given task are sent to the AI model provider(s) that power that task -- currently Replicate, fal, Modal (which runs the SoulX-Singer models we host), MiniMax, Google DeepMind, Mureka (through fal), Meta, ElevenLabs, and OpenAI (running models that include open-source RVC and Demucs). See the Terms of Service for the current list of which provider powers which feature. These providers process your submission to generate the requested output and, except as described next, do not use it to train their own general-purpose models under our agreements with them (except where a provider is itself the operator of an open-source model we self-host via Replicate).
Mureka is the exception. Its API terms allow it to use personal data to provide its services and to improve its algorithms. For Finish This Melody, Melody to Song, Reference Audio to Song, Lyrics to Backing Track, the AI Instrumental Generator, and Text to Singing Voice in languages other than English, what you submit (the lyrics, any melody file, the 30-second reference clip you choose in Reference Audio to Song, or the music description you type) is therefore processed under Mureka's terms, on servers in the United States. The submission is passed through fal's storage with a short expiry, and fal keeps request records for up to 30 days. Text to Singing Voice does not use or upload your voice recording in this mode; the song Mureka makes is sent to Demucs (via fal) to isolate the vocal, and we delete our copy of any uploaded melody file and of the intermediate audio when your job ends, keeping only the finished vocal for the retention period. For Lyrics to Backing Track we also use Demucs to separate the song, and keep the finished instrumental, guide vocals and full song for the retention period. A reference clip you submit stays in your storage under the same retention rules as your other uploads. Song With My Voice no longer uses Mureka. Voices that were created with an earlier version of Song With My Voice are still held by Mureka and we cannot delete them ourselves; to have one removed, contact us and we will ask Mureka to delete it.
AI DJ Drop works differently: the name, the words and the style you type are sent to fal, which runs ByteDance's Seed Audio 1.0 model to make the clip, and fal keeps request records for up to 30 days. We have not been able to confirm from the model provider whether it keeps or reuses submitted text, so do not enter anything you would not want a third-party AI provider to process. DJ Drop uses built-in voices only and never receives a voice recording from you. The finished clip is kept in your storage under the same retention rules as your other results.
Marketing emails
We send marketing emails (a welcome email shortly after you sign up, and a reminder an hour and a day later if you start a checkout and do not finish it) in two cases only: you ticked the optional box at sign-up (it is never pre-ticked, and signing up works without it), or you signed up or started the checkout from the United States, where the law lets us email you as long as every email has a clear way to opt out. Everywhere else, and wherever we cannot tell your country, we send nothing unless you ticked the box. We do not buy lists or email people who have not signed up. We keep a record of the basis we relied on (when and where you ticked the box, or the country rule that applied) and send nothing once you unsubscribe. Emails about your account itself, such as confirming your address or resetting your password, are not marketing emails and are not affected by unsubscribing.
- Every marketing email has an unsubscribe link and supports one-click unsubscribe in mail apps. Unsubscribing stops them straight away.
- We send them through Amazon SES. SES tells us whether each email was delivered, bounced, or was marked as spam, and, using a small image and link tracking in the email, whether it was opened and which links were clicked. Opens are only approximate, because some mail apps open emails on your behalf. If you would rather not be tracked this way, unsubscribe from marketing emails or turn off image loading in your mail app.
- We keep a record of each marketing email (including a copy of what was sent), the events above, and the checkouts you started, for as long as your account exists, and erase them when we erase your account data. An address that bounced, complained or unsubscribed is kept on a suppression list (the address only, even after your account is deleted) so that it is never emailed again.
- We use these records only to run and improve these emails and to stop sending to people who should not receive them.
Cookies & analytics
We use essential cookies to keep you signed in and remember basic preferences (like light/dark theme). With your permission, we also use analytics: Google Analytics (aggregate usage) and Microsoft Clarity (which can record how you interact with a page, such as clicks and scrolling, to help us find usability problems). Ahrefs Web Analytics is cookie-free.
If you are in the EEA, the UK, or California, analytics cookies stay off until you accept them in the cookie banner. Elsewhere analytics are on by default and you can switch them off at any time. We honor the Global Privacy Control signal. Use Cookie settings or Do Not Sell or Share My Personal Information in the page footer to change your choice. We do not sell your personal information or use advertising cookies. These tools do not receive the content of your lyrics, audio, or voice files.
International data transfers
Because our infrastructure and AI processing providers operate in multiple countries, your data may be processed outside your country of residence, including in jurisdictions that may not have the same data protection laws as your own. Where required, we rely on appropriate safeguards (such as standard contractual clauses) for these transfers.
Children's privacy
Lyrictomelody AI is intended for users 18 and older (see our Terms of Service). We do not knowingly collect personal data from children. If you believe a child has provided us data, contact us and we will delete it.
Your rights
Depending on where you live, you may have the right to:
- Access or correct your profile information.
- Delete your account. Access is disabled immediately; to have your stored files, voice models and personal data permanently erased, email us after deleting (we acknowledge within 7 days and complete erasure within 30 days). We keep only what the law requires, such as invoices and the records of rights confirmations needed to handle disputes.
- Receive a machine-readable export of your account data (use "Download my data" in your profile settings, or email us).
- Object to, or request we restrict, certain processing of your data.
- Withdraw previously given consent (e.g., for optional data uses) at any time.
Email support@lyrictomelodyai.com for any privacy request. We respond within 7 days.
Who operates this service
Lyrictomelody AI is operated by a small, product-focused team based in Shanghai, China. For formal privacy inquiries, data protection requests, or to reach our data controller, email support@lyrictomelodyai.com.
Our mailing address for correspondence: Tom Red, 444 Alaska Avenue, Torrance, CA 90503, USA.