Is This TTS API HIPAA Compliant for Healthcare Providers?

By VoxMedica editorial team · Updated 2026-11-19

This page explains how VoxMedica’s TTS API handles data and voice synthesis for healthcare startups. We clarify what we do and do not claim regarding data retention, security measures, and certifications. You will learn exactly how your input text is used, whether audio is stored, and how synthetic voices are identified.

Our service is designed for developers building patient education materials who need clear, reliable audio without enterprise complexity. We do not offer a HIPAA certification or SLA, placing responsibility for compliance on your integration. Read on to see how our straightforward API fits into your workflow and what support options are available.

Is This TTS API HIPAA Compliant for Healthcare Providers?

Frequently asked questions

Do you store patient data?

Input text is used only to produce the audio and is not stored after generation. We keep request metadata, such as time, character count, and status, strictly for billing purposes. This limited retention supports our accounting without retaining the actual patient content long-term.

Is my text used for training?

No. Your input text is used exclusively to generate the requested audio file and is not used for model training. The generated audio is returned in the response and is not stored, ensuring your specific patient data remains private to your immediate use case.

Is this HIPAA compliant?

We do not offer a HIPAA compliance guarantee. There is no official certification, and we do not provide a Business Associate Agreement (BAA). While we do not store input text, you should verify if our standard service meets your specific regulatory requirements for protected health information.

Do you have a BAA?

No. We do not offer a Business Associate Agreement (BAA). Our service is designed for general use without enterprise-grade contracts. If you require a BAA for HIPAA compliance, you may need to handle that aspect internally, as it is not part of our standard offering.

Where is data stored?

We do not specify exact data center locations. Input text is not stored after generation. Request metadata is kept for billing. We do not offer on-premise installation or specific data-center choices, so your data resides in our cloud infrastructure without guaranteed geographic location details.

Can I delete my data?

Input text is not stored after generation, so it is effectively deleted upon delivery. Request metadata (time, characters, status) is kept for billing purposes. We do not offer a specific data deletion portal, but since content isn't retained, your primary data is already removed.

Are voices synthetic?

Yes. Every generated file carries an inaudible watermark that marks it as synthetic speech. The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain audiobook readings, but the output is generated via our text-to-speech engine, not a direct playback of original recordings.

Is there a watermark?

Yes. Every generated audio file carries an inaudible watermark that marks it as synthetic speech. This allows you to identify that the audio was produced by our text-to-speech API, distinguishing it from human-recorded content or other synthetic sources without affecting the listening experience.

What licenses do you use?

Our voices are based on licensed recordings from LibriTTS-R (CC BY 4.0), derived from LibriVox public-domain audiobook readings. This foundation ensures we use legally sourced material for our voice presets, providing a clear licensing background for the synthetic voices used in your patient education materials.

Is there an SLA?

No. We do not offer a Service Level Agreement (SLA). There is no uptime guarantee, and no enterprise contract is available. Our service is designed for straightforward integration without the complexity of guaranteed performance metrics, making it suitable for startups prioritizing simplicity over strict reliability guarantees.

Is there an uptime guarantee?

No. We do not provide an uptime guarantee. Requests are processed one after another, so under load, a request may wait in a queue. Typical speed is roughly real time, but we do not guarantee specific availability windows or performance thresholds for your integration.

Do you share data with others?

Input text is used only to produce the audio and is not used for training or shared externally. Generated audio is returned in the response and not stored. Request metadata is kept for billing. We do not share your patient data with third parties for other uses.

Can I use this for PHI?

You can use this for Protected Health Information (PHI), but we do not guarantee HIPAA compliance. We do not offer a BAA or specific certifications. Since input text is not stored, the risk of data retention is low, but you must assess if this meets your regulatory needs.

How do I contact support?

Support is by email only. There is no phone number or live chat. We do not offer 24/7 support. For issues with your API key or billing, please send an email to our support address for assistance.

Data Usage Policy

Sound wave illustration for: Data Usage Policy

When you send text to our API, we use it strictly to generate the audio file you receive. That input text is not ingested into our training datasets, ensuring your patient education content remains distinct from the model’s general knowledge base.

Text not used for training

Our engine processes your input solely for synthesis. Unlike some large language models, we do not retain the semantic content of your texts to improve future voice generation, keeping your specific medical phrasing private.

Audio not stored

The generated audio file is returned directly in the HTTP response and is not archived on our servers. Once you download the MP3, WAV, or OPUS file, it exists only on your infrastructure, eliminating storage risks for sensitive patient audio.

Metadata kept for billing

We retain minimal request metadata, including timestamp, character count, and status, strictly to calculate your usage and process your USDT top-ups. This ensures accurate billing without retaining the actual audio or text content.

HIPAA Considerations

Sound wave illustration for: HIPAA Considerations

We do not claim HIPAA certification, nor do we offer a Business Associate Agreement (BAA). This API is designed as a straightforward tool for patient education materials, leaving compliance decisions to the startup’s own risk assessment.

We provide no uptime SLA guarantee. Our service is built for reliability, but healthcare data handling remains your responsibility when integrating audio outputs into your specific workflow.

No certification claimed

VoxMedica does not hold a formal HIPAA certification, nor do we provide a signed Business Associate Agreement (BAA). We operate as a general-purpose TTS API, allowing healthcare startups to assess our data handling practices against their own regulatory needs.

  • No official HIPAA certification
  • No standard BAA included
  • General-purpose infrastructure

No SLA guarantee

We do not offer a Service Level Agreement (SLA) guaranteeing specific uptime percentages. While our infrastructure is stable, we do not contractually promise availability windows, distinguishing us from larger enterprise-grade healthcare vendors.

Your responsibility

As the developer, you are responsible for determining if our service meets your specific HIPAA requirements. You must decide if using our API for patient education materials fits your compliance strategy, as we do not manage your overall data governance.

  • Startup-driven compliance decisions
  • No enterprise contract
  • Your governance responsibility

Voice Synthesis Nature

Sound wave illustration for: Voice Synthesis Nature

Our service generates synthetic speech using licensed recordings. The eight available voices are based on the LibriTTS-R dataset, which includes public-domain audiobook readings from LibriVox under a CC BY 4.0 license.

Each generated audio file contains an inaudible watermark that marks it as synthetic speech. This ensures transparency, clearly distinguishing machine-generated audio from real human recordings for patient education materials.

Synthetic voices

We provide eight preset voices, such as Aria and Felix, derived from licensed recordings. These voices are not clones of specific celebrities or actors but are distinct synthetic personas designed for clarity in medical contexts.

Inaudible watermark

Every audio file produced by our API carries an inaudible watermark. This technical marker confirms the content is synthetic, aiding in audit trails and ensuring patients know they are listening to generated speech.

Not real human voices

Unlike voice cloning services, we do not upload custom voice samples from your staff. Our voices are static presets. This consistency ensures predictable audio quality without the variability of human recording sessions.

Data Retention

Sound wave illustration for: Data Retention

We retain minimal data to ensure accurate billing and service reliability. Specifically, we keep request metadata, including the timestamp, character count, and status code for each synthesis job. This information is used solely to verify that charges correspond to actual requests and to troubleshoot any delivery issues.

Request metadata kept

The system logs essential details like the time of request, input length, and response status. This metadata helps us maintain transparency in usage reports and supports quick resolution of any discrepancies in your account activity.

  • Timestamp of each request
  • Character count processed
  • Response status code

Billing records

Since we charge per character, keeping a record of each request ensures you are billed accurately. These records are retained to support your financial audits and to confirm that failed requests were refunded correctly.

No audio archive

Unlike many other services, we do not store the generated audio files after they are delivered to you. Once the audio is returned in the response, it is not archived on our servers, reducing storage costs and minimizing data exposure.

Third-Party Licenses

Sound wave illustration for: Third-Party Licenses

Our voices are generated from licensed recordings of the LibriTTS-R dataset, which draws from public-domain audiobook readings available on LibriVox. This foundation ensures we do not rely on proprietary, high-cost voice assets, keeping our infrastructure simple and transparent for healthcare startups.

The underlying audio data carries a CC BY 4.0 license, allowing us to process and synthesize the content legally. By using these established public-domain readings, we maintain clear lineage for the synthetic voices used in your patient education materials.

LibriTTS-R source

The LibriTTS-R dataset provides the raw vocal recordings that power our eight voice presets. These recordings serve as the foundational material for our synthesis engine, offering a diverse range of vocal characteristics without requiring custom studio sessions for each voice.

  • Based on standard public datasets
  • Used for all eight voice presets

CC BY 4.0 license

The source recordings are distributed under the Creative Commons Attribution 4.0 International license. This means the original readings are free to use and adapt, provided appropriate credit is given to the original authors and readers, ensuring our API remains legally robust for commercial use.

  • Standard open license
  • Clear usage rights

Public domain readings

Many of the source texts come from LibriVox, a volunteer project providing public-domain audiobooks. This ensures that the underlying literary content and many of the voice samples are free from restrictive copyright claims, simplifying compliance for your healthcare applications.

  • Volunteer-recorded content
  • Minimal copyright restrictions

Security Measures

Sound wave illustration for: Security Measures

We use HTTPS encryption for all API communications, ensuring your patient data remains secure during transmission. Our API key authentication uses a simple Bearer token scheme, making integration straightforward for healthcare developers without requiring complex enterprise-grade security setups.

HTTPS encryption

All requests to https://api.9a9993.info/v1 are secured via HTTPS. This ensures that the text you send and the audio files you receive are encrypted in transit, protecting sensitive patient information from interception.

Bearer token auth

We use standard Bearer token authentication. You pass your unique API key in the Authorization: Bearer header. This method is simple, widely supported by HTTP libraries, and aligns with how many modern APIs handle security.

  • One active key per account
  • Revoke old keys instantly
  • Standard HTTP header format

No third-party sharing

Your input text is used solely to generate the audio and is not used for training our models. We do not share your data with third parties. Request metadata is retained only for billing purposes.

Security Measures

Patient Consent

Sound wave illustration for: Patient Consent

When using synthesized voices for patient education materials, clear disclosure helps maintain trust. Since the voices are synthetic, informing users that they are listening to AI-generated audio is a straightforward step toward transparency in your healthcare communications.

We provide tools to help you integrate these audio files seamlessly, ensuring your patient guides remain accessible and compliant with general transparency standards.

Synthetic voice disclosure

Every audio file generated by our API carries an inaudible watermark marking it as synthetic speech. This technical feature supports your disclosure strategy by providing verifiable proof that the voice is not a real human recording.

You can explicitly state in your patient guides that the narration is AI-generated. This clarity helps patients understand the nature of the audio without needing complex legal jargon.

  • Inaudible watermark in every file
  • Clear distinction from human recordings

Transparency in guides

Include a simple note in your digital or printed patient materials stating that the audio is synthesized. This approach aligns with best practices for digital health communication.

Our API allows you to generate consistent audio for various materials, making it easier to maintain a uniform disclosure message across all your content.

Legal requirements

While we do not offer a HIPAA certification, your responsibility is to ensure the chosen technology meets your specific regulatory needs. Using synthetic voices with proper disclosure is generally accepted for educational purposes.

Consult your legal team to confirm that synthetic narration meets your jurisdiction's requirements for patient communication. We provide the data to support your compliance documentation.

Support Contact

Sound wave illustration for: Support Contact

Support is handled entirely via email, ensuring a clear written record of your inquiries and our responses. This approach keeps communication focused and documented for your team’s reference.

We do not offer live chat or phone support, allowing us to maintain a straightforward, low-overhead service structure ideal for startups without the complexity of enterprise-grade contact centers.

Email support only

All support requests are managed through our dedicated email channel. This ensures that every question, whether about API keys or billing, receives a documented response that you can easily search and reference later.

There is no phone number to call, and no live chat widget on the page, keeping our operations simple and cost-effective.

  • No phone number
  • No live chat

No live chat

We prioritize asynchronous communication to allow our support team to provide thoughtful, accurate answers without the pressure of real-time interaction. This reduces miscommunication and ensures detailed explanations for technical questions.

Response times

We aim to respond to all support emails within one business day. While we strive for speed, complex technical queries may take slightly longer as we verify details against your account data.

Comparison with Azure

Sound wave illustration for: Comparison with Azure

Azure offers HIPAA compliance options with Business Associate Agreements (BAAs) and robust infrastructure, making it a standard for large healthcare institutions. However, this comes with complex contracts, higher minimums, and enterprise-grade overhead that many startups find unnecessary for simple text-to-speech needs.

Our approach strips away the enterprise complexity. We provide a straightforward API that mimics familiar structures without requiring a BAAs for basic usage. This simplicity allows healthcare startups to integrate patient education audio quickly, focusing on development rather than negotiating data processing agreements for every minor feature.

Azure HIPAA compliance

Azure maintains strict HIPAA compliance with documented safeguards and offers Business Associate Agreements (BAAs) to cover protected health information. It provides extensive infrastructure controls, including dedicated data centers and detailed audit logs, suitable for large-scale hospital systems managing vast amounts of patient data.

  • Full HIPAA compliance certification
  • Business Associate Agreements (BAAs) available
  • Extensive infrastructure and data center controls
  • Complex contract structures for enterprise use

Our simpler approach

We focus on a clean API interface for text-to-speech without the overhead of enterprise contracts. Our service is designed for developers who need reliable audio outputs for patient guides without the complexity of managing extensive data processing agreements or navigating rigid enterprise licensing terms.

  • Straightforward API integration
  • No BAAs required for basic usage
  • No minimum spend or complex contracts
  • Lower barrier to entry for startups

Startups vs Enterprise

Startups typically need quick integration and predictable costs rather than the full suite of enterprise guarantees. Our service offers a cost-effective solution for generating patient education materials, allowing teams to scale usage based on actual demand without committing to long-term enterprise agreements or high minimums.

  • Predictable per-character pricing
  • Quick integration for development teams
  • No long-term enterprise commitments
  • Ideal for scaling patient education content

Technical Details

Sound wave illustration for: Technical Details

Our API provides transparent technical signals to help developers debug integration issues quickly. We track request metadata for billing and performance monitoring without storing the generated audio files.

Understanding these technical details ensures you can accurately count characters and handle responses correctly in your patient education workflows.

This section covers the mechanics of how your requests are processed and logged.

We focus on clarity and simplicity for startups.

You get the data you need without enterprise bloat.

Request logging

We log request time, character count, and status for billing. Audio files are returned in the response and not stored.

This minimal logging supports troubleshooting without cluttering your data.

  • Logs kept for billing purposes only

Character counting

Every character, including spaces and punctuation, counts toward your usage. The limit is 4096 characters per request.

Accurate counting ensures predictable costs for your patient guides.

  • Spaces count as characters
  • Max 4096 chars per request

Status codes

Standard HTTP status codes indicate success or failure. Failed requests are free, so you won't be charged for errors.

Use these codes to automate your retry logic.

  • Failed requests are free
  • Standard HTTP responses

Limitations

Sound wave illustration for: Limitations

Our service is a cloud API, meaning you cannot install the TTS engine directly on your own servers. You send requests to our endpoint and receive audio files back. This keeps setup simple but means you depend on our infrastructure for uptime and processing.

No on-premise install

We do not offer an on-premise install option. All processing happens in our cloud environment. This simplifies your stack but requires an internet connection for every synthesis request.

No data center choice

We do not offer specific data-center choices. Your data is processed in our unified cloud infrastructure. This reduces complexity but means you cannot dictate the physical location of your audio generation for specific regulatory jurisdictions beyond general compliance.

No enterprise contract

We do not provide a formal enterprise contract with custom SLAs or dedicated support lines. We operate on a straightforward pay-per-use model. This is ideal for startups that need flexibility without the overhead of long-term legal agreements or minimum spend commitments.

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. This consistency helps you predict costs for large volumes of patient guides or appointment reminders without tracking tiered usage.

Avatar of the voice aria
ariafemale
Avatar of the voice nova
novafemale
Avatar of the voice luna
lunafemale
Avatar of the voice sage
sagefemale
Avatar of the voice felix
felixmale
Avatar of the voice milo
milomale
Avatar of the voice atlas
atlasmale
Avatar of the voice orion
orionmale

23 languages

Any voice speaks any of these languages.

Sources and further reading

  1. Wikipedia
  2. LibriTTS-R paper (arXiv 2305.18802)
  3. Creative Commons Attribution 4.0
  4. TRON developer hub: TRC-20
  5. LibriVox public-domain audiobooks
  6. openai-python on GitHub
  7. ISO 639 language codes
  8. Node.js documentation
  9. LibriTTS-R corpus (OpenSLR 141)
  10. curl documentation
  11. W3C WAI: making audio and video accessible
  12. FFmpeg documentation

More on this site