Text-to-Speech API for Automated Voice Assistants

By Vocalix editorial team · Updated 2026-11-18

Vocalix delivers a reliable cloud API designed for large engineering teams building voice assistants and conversational bots. Send plain text via JSON and receive synthesized audio files. While requests are processed sequentially and may queue under load, generation speed matches real-time audio (25 characters per second), ensuring your automated voice interactions meet strict volume and performance requirements.

This page outlines the technical specifications, pricing model, and integration workflow for our text-to-speech engine. Whether you are powering IVR systems or chatbot responses, you will find the concrete details needed to deploy our API using standard HTTP clients or OpenAI-compatible SDKs.

Text-to-Speech API for Automated Voice Assistants

Frequently asked questions

What is a voice generator API?

A voice generator API is a cloud text-to-speech service where you send text and receive an audio file. The endpoint accepts JSON bodies similar to the OpenAI speech endpoint, allowing standard SDKs to work by adjusting the base URL. This enables automated voice interactions for chatbots and voice assistants, eliminating the need for manual audio file creation and management.

How is this different from a consumer voice generator?

This is an engineering-focused API designed for large teams and high-volume automated interactions. It offers programmatic control over voices, formats, and speed via a REST API. Unlike consumer apps, it provides direct access to raw audio files, supports multiple languages, and integrates directly into your backend infrastructure for scalable voice synthesis.

Which voice is best for a customer service bot?

Milo offers a friendly, conversational tone, while Sage provides calm, instructional guidance. Neither is objectively superior; suitability depends on your specific interaction style. You can test all eight presets to determine which voice best fits your bot’s personality. The API allows you to switch voices dynamically per request to match your requirements.

Does the API support streaming audio?

No, the API does not support streaming, WebSocket, or real-time partial audio. It returns complete audio files upon request completion. Requests are processed sequentially, so latency depends on queue depth. For real-time needs, you must handle buffering on the client side or manage the full file retrieval workflow.

What happens if the synthesis fails?

If synthesis fails, the charge is returned in full. You are only billed for successful requests. This ensures you do not pay for wasted compute resources. You can retry failed requests or adjust parameters like speed or voice to resolve issues without incurring additional costs for the failed attempt.

Can I use this for IVR systems?

Yes, the API is suitable for IVR systems. You can generate audio files for menu options and prompts using voices like atlas for neutral news-style delivery or felix for energetic prompts. The API supports standard audio formats like mp3 and wav, which are widely compatible with IVR platforms and telephony systems.

How fast is the audio generation?

Generation takes roughly real time, about 25 characters per second of audio. Short phrases may take a few seconds. Requests are processed sequentially, so under load, a request may wait in a queue. This latency is typical for batch processing and is suitable for non-real-time voice assistant interactions.

Which languages are supported?

We support 23 languages: Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. Any voice can speak any supported language. If the language is not specified, it is guessed from the script, defaulting to English for Latin text.

Is there a limit on text length?

Each request supports up to 4096 characters. For longer texts, you must split the input into parts of this size and join the resulting audio files. Spaces and punctuation count toward the character limit. This ensures consistent processing times and manageable API response sizes for each individual synthesis request.

Do I need to pay for each voice change?

No, changing the voice does not incur extra costs. You pay based on the character count of the input text, regardless of the voice selected. All eight voices use the same engine and pricing structure. You can freely switch between voices like aria, nova, or atlas without affecting your billing rate.

How do I handle long texts for a bot?

Split long texts into chunks of up to 4096 characters per request. Generate audio for each chunk separately, then join the audio files on your end. This approach ensures compatibility with the API limits and allows you to manage latency by processing chunks in parallel if your infrastructure supports it.

Is the audio watermarked?

Yes. Every generated file carries an inaudible watermark that marks it as synthetic speech. This watermark is embedded in the audio data and is not audible to listeners. It serves to identify the content as AI-generated, which is useful for transparency and compliance in voice assistant and bot deployments.

Can I clone my own voice for the bot?

No, we do not offer custom voice cloning or uploading your own voice. You must choose from the eight available presets. The voices are based on licensed recordings from LibriTTS-R and LibriVox. If you need a specific voice, you must select the closest matching preset from our available options.

How do I top up my account?

Top up using USDT on the TRON network (TRC-20). The minimum deposit is $5, and funds are credited automatically within minutes. We do not accept cards, PayPal, or invoices. Your balance does not expire, and you can manage your account and view usage details directly on the dashboard after signing in.

First request

curl

curl https://api.playu77.com/v1/audio/speech \
  -H "Authorization: Bearer $TTS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"tts-1","voice":"aria","input":"Hello from the API.","response_format":"mp3"}' \
  --output hello.mp3

How it works

Sound wave illustration for: How it works

The Vocalix API converts text into speech audio via a standard REST endpoint. You send a JSON payload with your text and voice selection, and receive an audio file. This integration is compatible with existing OpenAI SDKs, requiring only a base URL change and your API key.

Send JSON, get audio

POST your text to the speech endpoint. The engine processes the request and returns the audio file in your chosen format (mp3, wav, opus, flac, or pcm). Texts longer than 4096 characters should be split and joined client-side.

OpenAI SDK compatible

Use the official OpenAI SDKs by pointing the base URL to https://api.playu77.com/v1. Pass your Bearer token in the Authorization header. The request shape matches the standard speech endpoint, allowing immediate integration into existing codebases.

Roughly real-time synthesis

Generation speed is roughly real-time, processing approximately 25 characters of text per second of audio. Short phrases may take a few seconds. Requests are processed sequentially, so under load, a request may wait in a queue before synthesis begins.

Voice presets for bots

Sound wave illustration for: Voice presets for bots

Vocalix provides eight distinct voice presets designed for consistent audio output in automated systems. Each voice is generated from licensed recordings and carries an inaudible watermark to identify synthetic speech, ensuring clear attribution without requiring SSML or custom pronunciation tags.

Select voices based on your application’s tone requirements. All voices support 23 languages and can be controlled via standard API parameters like speed and response format.

Female voices: Aria, Nova

Aria offers a warm, clear narrator style suitable for instructional content or guided interactions. Nova provides a bright, upbeat tone ideal for promotional messages or engaging conversational flows.

  • Aria: warm, clear narrator
  • Nova: bright, upbeat

Male voices: Felix, Milo

Felix delivers an energetic, promo-oriented delivery for time-sensitive notifications or advertisements. Milo offers a friendly, conversational tone well-suited for customer service bots or casual dialogue systems.

  • Felix: energetic, promo
  • Milo: friendly, conversational

Neutral options: Atlas, Orion

Atlas provides a neutral, news-style delivery for clear information dissemination. Orion offers a deep, documentary tone appropriate for long-form audio content or authoritative voice assistants.

  • Atlas: neutral, news
  • Orion: deep, documentary
Voice presets for bots

Language support

Sound wave illustration for: Language support

The API supports 23 languages, allowing your voice assistant to interact with users globally. Every voice preset can speak any of the supported languages, ensuring consistent vocal identity across your bot’s interactions regardless of the user's locale.

Supported languages include Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, and Bengali are not currently supported.

23 supported languages

Our engine handles a diverse set of languages, enabling your conversational bot to serve a global audience. You can mix languages within a single session if needed, as each voice preset is capable of speaking any of the 23 supported languages.

  • 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 and manual

The API attempts to detect the language automatically based on the input text script. If detection is uncertain or you require precise control, you can explicitly set the language field using the ISO code in your request.

Latin script default

For text written in Latin script that is not English, the system defaults to English. To ensure accurate pronunciation for non-English Latin-script text (e.g., French or German), you must set the language parameter manually in your request.

Pricing model

Sound wave illustration for: Pricing model

The API charges based on input text length, where every character—including spaces and punctuation—counts toward the total. Failed requests are free, ensuring you only pay for successful synthesis. The standard rate is $10 per 1 million characters, or $0.01 per 1,000 characters.

Pay per character

Costs scale directly with usage. You receive immediate billing data for every request, allowing precise cost forecasting for high-volume bot deployments.

  • Spaces and punctuation count as characters
  • Failed requests are free
  • Real-time usage tracking

$10 per million characters

The flat rate applies to all models. Whether using tts-1 or tts-1-hd, the character cost remains identical, simplifying budget calculations for engineering teams.

  • Flat rate across all models
  • Predictable unit economics

No monthly minimums

Start with a $0.50 free trial credit valid for 7 days. Top-ups are processed via USDT (TRC-20) on the TRON network, with a $5 minimum deposit. Credits are added automatically within minutes after transfer confirmation, with no subscriptions or invoices required.

  • USDT (TRC-20) top-ups only
  • Minimum $5 deposit
  • Automatic credit within minutes

Technical specs

Sound wave illustration for: Technical specs

The API enforces a strict rate limit of 60 requests per minute per key, scaling from 10 per minute during the free trial. Requests are processed sequentially, so under load, synthesis may queue behind prior jobs. This sequential processing means naive retries can duplicate charges or cause latency spikes; clients should implement exponential backoff to manage concurrency safely.

60 requests per minute

Each API key supports a maximum throughput of 60 requests per minute. New accounts starting with the free trial are capped at 10 requests per minute until a top-up is processed. This limit applies to successful and failed requests, ensuring fair usage across large teams building voice assistants.

  • Standard limit: 60 requests/minute/key
  • Trial limit: 10 requests/minute/key
  • Applied per API key, not per IP

Multiple audio formats

Clients can select the output format via the response_format parameter. Supported formats include mp3, wav, opus, flac, and pcm. This flexibility allows integration with various voice assistant frameworks and IVR systems that have specific codec requirements, from lightweight streaming (opus) to high-fidelity playback (wav, flac).

  • mp3, wav, opus, flac, pcm
  • Configurable per request
  • No format conversion required on client side

Retry-friendly design

While the API is designed for robust integration, requests are processed sequentially. This means a retry may wait behind previous jobs, potentially causing delays. Clients should handle timeouts gracefully and avoid duplicate submissions without idempotency keys, as the sequential queue can make naive retries risky for real-time voice responses.

Technical specs

Use cases

Sound wave illustration for: Use cases

Vocalix integrates directly into conversational workflows, converting text responses into audio files that voice assistants can play instantly. The API handles character limits by splitting long inputs, ensuring seamless playback for chatbots without complex client-side logic.

Chatbot voice responses

Send user messages as JSON payloads and receive MP3 or WAV files in the response body. This allows chatbots to deliver natural-sounding replies without relying on third-party streaming services or maintaining local audio assets.

  • Instant audio delivery via response body
  • Supports 23 languages with auto-detection
  • No custom SDK required

IVR systems

Generate dynamic phone menus and notification messages on demand. The API’s predictable pricing and lack of monthly minimums make it suitable for high-volume call centers that need varied audio content without fixed infrastructure costs.

  • Dynamic menu generation
  • No monthly minimums
  • Predictable per-character billing

Audio notifications

Convert system alerts or status updates into spoken audio for voice assistants. Since generated audio is returned in the response body and not stored, clients can play the file immediately or discard it, optimizing storage and bandwidth.

  • No audio storage required
  • Immediate playback capability
  • Efficient bandwidth usage

Developer experience

Sound wave illustration for: Developer experience

Vocalix provides a straightforward REST endpoint for integrating text-to-speech into voice assistants and chatbot workflows. The API accepts standard JSON payloads and returns audio files directly, requiring no complex middleware or custom client libraries to function.

Simple REST endpoint

Send a POST request to the /v1/audio/speech endpoint with your text, voice selection, and format preferences. The service processes the input and returns the synthesized audio file in the response body, allowing immediate playback or storage.

cURL ready

The endpoint is fully compatible with standard HTTP clients. You can generate audio instantly using a simple cURL command, making it easy to test integration without setting up a full development environment or installing specific packages.

No custom SDK needed

You do not need a proprietary SDK to use the API. The request format mirrors the OpenAI speech endpoint, so existing OpenAI SDKs work by simply changing the base URL. This allows teams to leverage familiar code structures and libraries.

Data and reliability

Sound wave illustration for: Data and reliability

Vocalix processes input text strictly to generate the requested audio file. We do not use your input text for model training or to improve the underlying engine, ensuring your conversational data remains private to your application's logic.

Text not used for training

Your text input is processed in real-time to produce the audio output. It is not retained for training purposes or used to refine the voice models, keeping your conversation data distinct from our training datasets.

No audio storage

Generated audio files are returned directly in the response and are not stored on our servers. Once the audio is delivered to your client, it is not kept by Vocalix, reducing your storage overhead and ensuring ephemeral audio delivery.

Billing metadata kept

We retain request metadata, including timestamp, character count, and status, solely for billing accuracy. This data allows you to verify charges against your usage and ensures transparent accounting for your automated voice infrastructure.

  • Request time and status
  • Character count per request
  • Billing verification data
Data and reliability

Get started

Sound wave illustration for: Get started

Access the Vocalix API by visiting our /get-api-key/ page. Create an account using your Google credentials or an email and password. Upon registration, you receive a unique API key and immediate access to your dashboard, where you can view your balance, usage statistics, and a ready-to-use cURL example.

Sign up for API key

Navigate to the /get-api-key/ page to create your account. You can register using Google or standard email authentication. Once created, your account is active and linked to your new API key, allowing you to start sending requests to https://api.playu77.com/v1 immediately.

Free trial credit

Sign in with Google to receive $0.50 in trial credit, valid for 7 days. This provides approximately 50,000 characters of synthesis. Accounts created with email and password do not include this trial credit. You can top up later using USDT on the TRON network.

  • Google sign-in: $0.50 credit (50k chars)
  • Validity: 7 days
  • Email accounts: No trial credit

Documentation link

Review the full technical documentation at /docs/. It details the request structure, available voices, and language codes. You can also check /pricing/ for clear character-based rates without monthly minimums.

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. Modern teams now rely on intelligent assistants to streamline workflows and boost overall productivity.

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. ISO 639 language codes
  3. openai-python on GitHub
  4. Node.js documentation
  5. Tether (USDT)
  6. Python documentation
  7. OpenAI API reference: create speech
  8. FFmpeg documentation
  9. LibriVox public-domain audiobooks
  10. openai-node on GitHub
  11. MDN: Web Speech API
  12. IETF RFC 7845: Ogg Opus

More on this site