How Does Our TTS API Convert Text to Audio?

By LuminaRead editorial team · Updated 2026-11-18

This page explains how LuminaRead’s API converts text into natural, accessible speech for users with dyslexia or visual impairments. We break down the technical workflow so product teams understand exactly how their text becomes audio, ensuring reliable integration for inclusive digital experiences.

You will learn about input processing, voice selection, and latency. We also clarify our per-character billing, language support, and data privacy practices. Read on to see how simple, compliant audio generation fits into your development workflow without infrastructure overhead.

Frequently asked questions

What is the difference between TTS and speech synthesis?

Text-to-speech and speech synthesis are effectively the same technology. LuminaRead uses a cloud API to convert your text input into natural audio files. This process allows product teams to generate accessible speech for users with dyslexia or visual impairments on demand, rather than relying on static pre-recorded files.

Is a TTS API better than a pre-recorded audio file?

An API offers dynamic generation for varying content lengths without storing massive audio libraries. You send text and receive an audio file instantly. This is ideal for updating digital experiences frequently. Pre-recorded files are static; our API adapts to your specific text, ensuring accurate pronunciation and consistent tone.

Can TTS handle different accents?

Our API supports 23 languages, including English, Japanese, and Spanish. While voices are preset, they speak all supported languages naturally. For non-English Latin-script text, you can set the language parameter to ensure correct accent handling. We do not offer custom voice cloning or specific regional accent controls beyond the preset voices.

Does the API support background music?

No, the API generates pure speech audio only. It does not support adding background music, sound effects, or audio mixing. You receive a clean audio file containing only the synthesized voice. If you need background music, you must add it using your own audio editing tools after generation.

How accurate is Japanese pronunciation?

Japanese is one of our 23 supported languages. The model is trained on licensed recordings from LibriTTS-R and LibriVox, ensuring high naturalness. The voice presets handle Japanese pronunciation accurately without requiring special markup. You simply include Japanese text in your request to generate clear, natural-sounding speech for your users.

Can I control the pitch of the voice?

No, the API does not support pitch control. You can adjust the speech speed between 0.25 and 4.0, but pitch remains fixed per voice preset. Each of the eight voices has a specific tonal quality. If you need pitch variation, you would need to process the generated audio file separately.

Is the API suitable for live presentations?

The API is not designed for real-time live presentations. Requests are processed sequentially, which may cause delays under load. While short phrases generate quickly, longer texts take time proportional to their length. For live use, pre-generating audio files is more reliable than calling the API during a presentation.

Do I need internet access to use the API?

Yes, you need an internet connection because LuminaRead is a cloud API. You send text to our endpoint and receive an audio file via HTTP. There is no local processing or offline mode. Your application must maintain connectivity to submit requests and receive the generated audio responses reliably.

Can I use the API offline?

No, the API requires an active internet connection. All processing happens on our cloud servers. We do not offer an offline SDK or local installation. You must send your text to https://api.playmx52.com/v1 and receive the audio file over the network. Offline usage is not supported.

How does the API handle punctuation?

Punctuation is treated as part of the character count and influences natural speech pauses. The API interprets standard punctuation to create realistic rhythm and intonation. Spaces and punctuation count toward your 4096-character limit per request. This ensures the generated audio sounds natural without requiring special markup like SSML.

Is the audio format standard across providers?

We support standard formats: mp3, wav, opus, flac, and pcm. The request structure matches the OpenAI speech endpoint, making it easy for developers to integrate. You choose the format in your JSON body. This compatibility allows you to use existing tools and workflows familiar with standard audio generation APIs.

Can I use the API for voiceovers?

Yes, the API is excellent for voiceovers. You can generate natural-sounding speech for videos, e-learning, or digital content. Choose from eight distinct voice presets to match your brand tone. The generated audio files are ready for immediate use in your production pipeline, supporting various languages and speeds.

Does the API support multiple speakers?

No, each request generates audio from a single voice preset. You cannot combine multiple voices in one request. To create a dialogue, you must send separate requests for each speaker and concatenate the resulting audio files. This gives you full control over each character's voice and pacing.

How do I test the API before buying?

Sign in with Google to receive a free trial with $0.50 credit, valid for 7 days. This allows about 50,000 characters of testing. You can use the provided curl example on the /get-api-key/ page to generate audio. Failed requests are free, so you can test error handling without cost.

Understanding TTS Technology

Sound wave illustration for: Understanding TTS Technology

LuminaRead’s API transforms written text into natural speech through a streamlined cloud process. Developers send a JSON payload containing the text, model, and voice selection to our endpoint. The system processes up to 4,096 characters per request, ensuring consistent output quality for diverse content types.

The generated audio is delivered directly in the response, supporting formats like MP3, WAV, and FLAC. This immediate delivery allows product teams to integrate audio playback seamlessly into their applications without managing complex infrastructure or file storage systems.

Text Input Processing

Text Input Processing

When your application sends a request, the API accepts the text input and validates it against character limits. The system automatically detects the language unless you specify an ISO code, ensuring accurate pronunciation for the selected voice preset.

Audio File Generation

Our engine synthesizes speech using licensed recordings from LibriVox. Each generated file includes an inaudible watermark to identify it as synthetic speech. You receive the audio file directly in the API response, ready for immediate playback.

Cloud-Based Delivery

Since LuminaRead is a cloud API, there is no need for local infrastructure. You simply adjust your base URL in the OpenAI SDK or use standard HTTP requests. The service handles scaling, so you can support many users without server management.

How APIs Simplify Integration

Sound wave illustration for: How APIs Simplify Integration

APIs abstract the complex neural inference and audio encoding layers, allowing product teams to focus on user experience rather than server management. You simply send a JSON payload to our cloud endpoint, and we handle the heavy lifting of converting text into natural speech.

This standardized approach means you don’t need dedicated infrastructure. By using familiar HTTP requests, your team can integrate audio generation directly into existing workflows without building custom backend services or managing model versions.

No Infrastructure Needed

Our cloud API manages all compute resources. You don’t need to provision GPUs or manage storage for audio files. Just include your API key in the request header, and we return the audio file instantly.

This reduces your engineering overhead significantly. Your team can deploy faster, knowing that audio generation scales automatically without you worrying about server capacity or model updates.

Standardized Endpoints

We use a POST endpoint compatible with the OpenAI speech format. This means you can use existing OpenAI SDKs by simply changing the base_url to our API domain.

This compatibility minimizes learning curves. Your developers can use familiar libraries and tools, ensuring a smoother integration process for your product team.

Standardization also means consistent request shapes across different models, making it easier to switch between tts-1 and tts-1-hd if needed.

Easy Scaling

As your user base grows, our API handles the load. You don’t need to manually scale servers or optimize database queries for audio storage.

Rate limits are clearly defined (60 requests per minute per key), allowing you to plan your capacity accurately. For higher volume, you can issue additional keys to distribute the load.

This predictability helps product teams maintain consistent performance, even during peak usage times, without sudden infrastructure surprises.

Optional: SDK Compatibility

  • Works with official OpenAI SDKs via base_url change
  • Supports standard HTTP libraries (curl, requests, axios)
  • No custom SDK required

Role in Accessibility

Sound wave illustration for: Role in Accessibility

LuminaRead helps product teams build accessible digital experiences by turning text into natural speech. Our API supports users with dyslexia and visual impairments through clear, consistent audio generation.

By converting content to audio, we enable inclusive design that meets diverse user needs without requiring complex infrastructure or additional development effort.

Dyslexia Support

Our API provides clear, natural speech that helps users with dyslexia process text more easily. By converting written content into audio, we reduce the cognitive load of reading complex text.

Users can listen to articles, instructions, or product descriptions at their own pace, improving comprehension and reducing fatigue.

  • Natural speech reduces reading fatigue
  • Adjustable speed for individual needs
  • Consistent voice quality across all content

Visual Impairment Aid

For users with visual impairments, our API offers an alternative to screen readers by providing human-like audio. This enhances accessibility for those who prefer natural speech over robotic voices.

Our API supports multiple languages, ensuring that users from diverse backgrounds can access content in their preferred language.

  • Natural speech over robotic voices
  • Support for 23 languages
  • Consistent audio quality

Inclusive Design

LuminaRead enables inclusive design by providing a simple API that integrates seamlessly into existing applications. This allows product teams to create accessible experiences without significant technical overhead.

By offering clear, consistent audio generation, we help ensure that all users, regardless of their abilities, can access and engage with digital content.

  • Simple API integration
  • Clear, consistent audio
  • Support for diverse user needs

Request and Response Flow

Sound wave illustration for: Request and Response Flow

Sending a request is straightforward: you POST a JSON payload to the speech endpoint with your authentication key. The body includes the text input, chosen voice, and desired output format.

JSON Payload

Include the model ("tts-1" or "tts-1-hd"), input text (up to 4096 characters), voice preset, response_format, and optional speed and language parameters. Use the OpenAI SDK by pointing base_url to our API endpoint.

  • Auth via Authorization: Bearer tts-...
  • Text limit: 4096 characters per request
  • Supported formats: mp3, wav, opus, flac, pcm

Audio File Return

The API returns the generated audio file directly in the response body. For texts exceeding the character limit, you split the input, receive multiple audio chunks, and join them client-side.

Status Codes

A 200 OK response indicates success with the audio data. Use 400 or 429 codes to handle invalid inputs or rate limits (60 requests per minute). Failed synthesis charges are refunded automatically.

Latency and Performance

Sound wave illustration for: Latency and Performance

Our TTS API converts text to audio in the cloud, delivering synthesized speech via standard HTTP responses. While short phrases generate in a few seconds, longer texts process at roughly real-time speed, generating approximately 25 characters of text per second of audio.

Real-Time Generation

For extended content, the API processes requests sequentially, meaning generation time closely matches the audio duration. A 1,000-character input typically yields about 40 seconds of audio, created in near real-time without requiring complex infrastructure on your end.

Queue Processing

Requests are processed one after another, so high load may introduce a brief queue wait. With a limit of 60 requests per minute, most standard integrations remain responsive, though burst traffic might delay synthesis slightly.

Optimization Tips

Split texts into parts of up to 4,096 characters and join the resulting audio files for smoother playback. Keep request volumes under 60 per minute to avoid queue delays, and ensure your app handles HTTP status codes gracefully.

  • Split input at 4,096 characters
  • Monitor rate limits (60 req/min)
  • Handle HTTP status codes

Voice Selection Impact

Sound wave illustration for: Voice Selection Impact

The selected voice directly influences how effectively text is converted into natural speech. In accessibility contexts, clarity is paramount; voices like aria provide a warm, clear narrator tone, while sage offers calm, instructional tones ideal for long-form content. Each of the eight voices has a specific tonal quality that affects listener comprehension and fatigue.

Clarity for Listeners

Clear articulation ensures that users with dyslexia or visual impairments can process information without strain. Our voices are designed for natural speech patterns, avoiding robotic artifacts that can cause listener fatigue during extended reading sessions.

Emotional Tone

While we do not offer emotion tags, each voice carries a distinct character. Felix brings energy to promos, while Atlas remains neutral for news. This consistency helps users form a reliable auditory relationship with the content, enhancing engagement without requiring custom voice design.

Consistency

All voices are generated from licensed recordings, ensuring consistent, high-quality audio. Whether using nova for upbeat content or orion for documentaries, the tone remains stable across requests, providing a uniform experience for your end users.

Language Support

Sound wave illustration for: Language Support

The LuminaRead API supports 23 languages, including English, Spanish, French, German, and Japanese, allowing your product to reach a global audience without managing separate language models. Each voice can speak any of these supported languages, ensuring consistent audio quality across your user base.

While the API can guess the language from the text script, it defaults to English for Latin characters. For accurate pronunciation of non-English text in Latin script, you must explicitly set the language parameter. Languages like Vietnamese, Thai, and Ukrainian are not supported.

Multi-Language API

Our API handles 23 languages natively, from Arabic to Swahili. Every voice preset can speak any supported language, ensuring uniform audio characteristics regardless of the language selected. This simplifies your integration, as you do not need to manage multiple voice models or language-specific endpoints.

  • Supported: Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Turkish

Auto-Detection Limits

If you omit the language field, the API guesses based on the script. Latin script text defaults to English, which may cause mispronunciation for other Latin-based languages. Detection is not perfect for mixed-language text or less common scripts.

Explicit Language Setting

For non-English Latin-script text, always set the language parameter to ensure correct pronunciation. This is crucial for maintaining clarity for users with dyslexia or visual impairments who rely on accurate auditory cues. Unsupported languages like Vietnamese or Ukrainian will not render correctly.

Cost Structure

Sound wave illustration for: Cost Structure

LuminaRead charges $10 per 1 million characters, which equals $0.01 per 1,000 characters. Spaces and punctuation count as characters. Failed requests are free, with the charge returned in full if synthesis fails.

Per-Character Billing

Billing is transparent and usage-based. You only pay for what you generate, with no hidden costs for model selection or voice choice. Both tts-1 and tts-1-hd use the same engine and price.

No Setup Fees

There are no monthly plans, seats, or minimum spends. Start with a $0.50 free trial (valid for 7 days) when signing in with Google. Accounts created with email and password start without trial credit.

Volume Discounts

The per-character rate remains consistent regardless of volume. Top-ups are made via USDT on the TRON network (TRC-20) with a minimum of $5. Balance does not expire.

Data Privacy

Sound wave illustration for: Data Privacy

LuminaRead prioritizes your data integrity. Input text is used exclusively for the immediate generation of audio and is never repurposed for training our models. This ensures that your content remains proprietary and distinct from the training data used to refine the underlying engine.

No Data Training

Generated audio files are returned directly in the API response and are not stored on our servers after delivery. Only request metadata—such as timestamp, character count, and status—is retained for billing accuracy. Your actual voice content remains transient.

Temporary Processing

Text is processed in memory to produce the audio file. Once the audio is transmitted to your client, the input text is not archived for long-term storage or analysis. This minimizes the attack surface for potential data leaks.

Secure Transmission

All communications between your application and our API endpoint occur over HTTPS. Your API key and text payloads are encrypted in transit, ensuring that sensitive content remains secure from endpoint to endpoint.

Common Use Cases

Sound wave illustration for: Common Use Cases

Our API turns text into natural speech, enabling product teams to build accessible experiences for users with dyslexia and visual impairments. By sending a JSON payload, you receive audio files instantly, integrating seamlessly into existing digital workflows without managing infrastructure.

This reliability supports diverse applications, from educational tools to content platforms, ensuring that text content is accessible through high-quality synthesized voice.

Educational Apps

Educational platforms use our API to convert reading materials into audio, supporting students with dyslexia who benefit from auditory learning. The clear, natural voices reduce cognitive load, allowing learners to focus on comprehension rather than decoding text.

Developers can specify languages like English, Spanish, or French, ensuring content is accessible to a global student base. The consistent voice quality maintains engagement across long lessons.

  • Supports students with dyslexia through auditory learning
  • Clear voices reduce cognitive load
  • Multi-language support for global classrooms

News Aggregators

News platforms integrate our TTS API to offer audio versions of articles, catering to users with visual impairments or those on the go. The neutral, clear tones of voices like Atlas ensure factual delivery without distraction.

With real-time generation and reliable latency, news aggregators can provide up-to-date audio content without significant delay, enhancing user accessibility and engagement.

  • Audio versions of articles for visual impairments
  • Neutral tones for factual delivery
  • Real-time generation for up-to-date content

Audiobook Platforms

Audiobook services leverage our API to convert text-based books into audio files, expanding accessibility for readers with visual challenges. The variety of voices allows for engaging narrations, enhancing the listening experience.

By using the API, platforms can offer a wide range of voices and languages, ensuring diverse content is available to all users, regardless of their reading preferences or abilities.

  • Converts text books into audio files
  • Expands accessibility for visual challenges
  • Wide range of voices and languages

Choosing the Right API

Sound wave illustration for: Choosing the Right API

Selecting the right TTS API hinges on matching your technical stack, audio requirements, and budget. Since LuminaRead follows the OpenAI request shape, existing OpenAI SDKs work seamlessly by simply updating the base URL, reducing integration friction for product teams already familiar with that ecosystem.

SDK Compatibility

Our API uses the same request structure as the OpenAI speech endpoint. This allows you to use official OpenAI SDKs by changing the base_url to https://api.playmx52.com/v1. This compatibility ensures that your existing code handles authentication and payload formatting correctly without significant refactoring.

  • Official OpenAI SDKs work with base URL adjustment
  • Standard JSON payload structure
  • No custom client libraries required

Voice Quality

Choose from eight presets tailored for clarity and engagement. For accessibility, aria offers a warm, clear narrator voice, while sage provides calm, instructional tones ideal for long-form content. All voices are generated from licensed recordings, ensuring consistent, high-quality audio that meets accessibility standards for dyslexia and visual impairments.

  • 8 distinct voice presets
  • Optimized for clarity and comprehension
  • Licensed source recordings

Pricing Model

Billing is strictly per character at $0.01 per 1,000 characters, with no setup fees or monthly plans. Failed requests are free, and you only pay for successful synthesis. This transparent model allows product teams to accurately predict costs based on user volume and text length without hidden fees.

  • $0.01 per 1,000 characters
  • No monthly plans or seats
  • Failed requests are free

Listen to the 8 voices

Each sample below was generated by this API from the same sentence.

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. OpenAI API reference: create speech
  4. FFmpeg documentation
  5. MDN: Web Speech API
  6. Creative Commons Attribution 4.0
  7. Node.js documentation
  8. W3C WAI: making audio and video accessible
  9. Tether (USDT)
  10. openai-python on GitHub
  11. TRON developer hub: TRC-20
  12. ISO 639 language codes

More on this site