TTS API for Voice Assistant and Chatbot Development

By VoiceStack editorial team · Updated 2026-11-18

VoiceStack provides a reliable text-to-speech API designed for product teams building voice interfaces, chatbots, and audio-driven applications. This page outlines how our developer-first infrastructure delivers natural-sounding synthetic speech with clear pricing and straightforward integration, enabling you to maintain consistent voice identity across user interactions without complex engineering overhead.

We focus on the technical specifics that matter for scalable products: standard audio formats, and transparent character-based billing. Whether you are developing a responsive chatbot or a voice assistant, you will find the concrete implementation details, voice library options, and API constraints needed to integrate speech synthesis into your existing workflow efficiently.

TTS API for Voice Assistant and Chatbot Development

Frequently asked questions

Is this TTS service truly free to start?

Yes. Signing in with Google provides $0.50 of free credit, valid for seven days, allowing approximately 50,000 characters of generation. Accounts created with email and password start without trial credit. You can top up using USDT on the TRON network to continue using the service without any monthly plans or minimum spend requirements.

Which voices work best for customer support chatbots?

For customer support, milo offers a friendly, conversational tone suitable for natural dialogue. Sage provides a calm, instructional voice ideal for guiding users through steps. Aria is a warm, clear narrator. These presets deliver natural-sounding synthetic speech designed to integrate smoothly into voice interfaces and chatbot applications without requiring custom training.

How does the API handle language detection?

The API guesses the language from the script if the language field is not set. Latin script text defaults to English. For non-English Latin-script text, you should explicitly set the language parameter. The service supports 23 languages, including Arabic, Chinese, French, German, Japanese, Spanish, and others, with any voice capable of speaking any supported language.

Can I use the API for commercial voice assistants?

Yes. The API is designed for product teams building voice interfaces, chatbots, and audio-driven applications. Input text is used only to produce audio and is not used for training. Generated audio is returned in the response and not stored. There are no usage restrictions preventing commercial deployment in voice assistant products.

What happens if a synthesis request fails?

Failed requests are free. If synthesis fails, the charge is returned in full. You are only billed for successful audio generation. This ensures you do not pay for errors or timeouts. The API processes requests sequentially, so under load, a request may wait in a queue before processing begins.

Are the voices based on real human recordings?

Yes. The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0), derived from LibriVox public-domain audiobook readings. Every generated file carries an inaudible watermark that marks it as synthetic speech. This ensures transparency about the synthetic nature of the audio while maintaining high quality and natural intonation.

How many characters can I send in one request?

You can send up to 4,096 characters per request. For longer texts, you should split the input into parts of this limit and join the resulting audio files. Spaces and punctuation count toward the character limit. The pricing is $0.01 per 1,000 characters, so accurate character counting helps manage your budget.

Does the API support streaming audio output?

No. The API does not support streaming, WebSocket, or real-time partial audio. You send text and receive the complete audio file in the response. This synchronous model simplifies integration for chatbots and voice assistants that do not require low-latency incremental audio delivery. You can use the official OpenAI SDKs by changing the base URL.

Is there a limit on the number of API keys?

There is one active key per account. You can issue a new key at any time, and the old one stops immediately. This ensures secure access management without maintaining multiple active credentials simultaneously. Each key is subject to rate limits of 60 requests per minute, or 10 per minute before the first top-up.

Can I control the speed of the synthesized speech?

Yes. You can adjust the speed between 0.25 and 4.0 using the speed parameter. This allows you to slow down speech for clarity or speed it up for brevity. The default speed is 1.0. This feature works with all eight voice presets and all supported audio formats, giving you precise control over output timing.

What audio formats are available for download?

You can request audio in mp3, wav, opus, flac, or pcm formats. These options cover various use cases, from web playback to high-quality archival storage. The endpoint is POST https://api.playvm9.com/v1/audio/speech. You specify the desired format in the response_format field of the JSON body along with the text and voice parameters.

Do I need to pay for SSL certificates for the API?

No. The API endpoint https://api.playvm9.com/v1 uses standard HTTPS connections. Standard HTTPS connections are supported, ensuring secure transmission of your text input and retrieval of audio files without additional infrastructure costs or configuration steps for your development team.

Is my input text used to train a new model?

No. Input text is used only to produce the audio for your request and is not used for training. Generated audio is returned in the response and not stored. Request metadata, such as time, characters, and status, is kept for billing purposes only. Your data remains private and is not repurposed for model improvement.

How long is my API key valid for?

API keys do not have an expiration date. They remain valid until you issue a new key, at which point the old one stops immediately. There are no monthly plans or subscriptions tied to key validity. Your balance also does not expire, allowing you to use your credits whenever needed without time pressure.

First request

curl

curl https://api.playvm9.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

Why Voice Assistants Need Reliable TTS

Sound wave illustration for: Why Voice Assistants Need Reliable TTS

Voice interfaces demand consistency to build user trust. A reliable TTS API ensures that your voice assistant maintains the same vocal identity across every interaction, preventing the jarring effect of switching between different voice models or quality levels. This stability is critical for products where audio is the primary communication channel.

Consistent voice identity across interactions

Consistent voice identity across interactions

Maintaining a single voice profile across all user touchpoints reinforces brand recognition. Whether a user interacts via a smart speaker or a mobile chatbot, the auditory experience remains uniform, reducing cognitive load and creating a cohesive product identity.

Fast synthesis for responsive chatbots

Chatbots require immediate audio feedback to feel conversational. Our API delivers synthesized speech quickly, allowing responses to play before the user loses interest. This speed is essential for maintaining the rhythm of natural dialogue without awkward pauses.

Predictable latency for user experience

This performance allows chatbots to deliver audio responses, ensuring the interaction feels immediate. Users expect near-instantaneous reactions to their queries. By minimizing generation time, your application avoids the frustration of waiting for audio to buffer, keeping the conversation flowing smoothly.

Core API Features for Audio Products

Sound wave illustration for: Core API Features for Audio Products

VoiceStack’s API delivers consistent audio output through eight distinct voice presets, allowing product teams to assign specific personalities to their chatbots or voice assistants. This variety ensures that interactions remain engaging.

8 preset voices for distinct personalities

The platform provides eight curated voices, split evenly between female and male tones. Options range from Aria’s warm narration to Felix’s energetic promo style, enabling developers to match the tone of their application precisely. Each voice is derived from licensed recordings, ensuring high-quality synthetic speech for various use cases.

  • Aria: warm, clear narrator
  • Nova: bright, upbeat
  • Luna: soft, storytelling
  • Sage: calm, instructional
  • Felix: energetic, promo
  • Milo: friendly, conversational

23 languages with automatic detection

Support spans 23 languages, including English, Spanish, German, and Japanese. The API automatically detects the language from the input script, defaulting to English for Latin text if unspecified. For non-English Latin-script content, setting the language parameter ensures accurate pronunciation and prevents misidentification.

Standard formats: MP3, WAV, OPUS, FLAC, PCM

Audio output is available in five standard formats: MP3, WAV, OPUS, FLAC, and PCM. This flexibility allows seamless integration into diverse voice interfaces, from lightweight web apps using MP3 to high-fidelity systems requiring uncompressed WAV or FLAC. Each generated file includes an inaudible watermark to identify it as synthetic speech.

Voice Library for Chatbot Personality

Sound wave illustration for: Voice Library for Chatbot Personality

VoiceStack offers eight preset voices designed to give your chatbot a distinct, consistent personality. By selecting from warm narrators, energetic presenters, or calm instructors, product teams can tailor the auditory experience to match their brand’s tone.

Each voice is built on licensed recordings, ensuring a professional baseline. The library supports all 23 available languages, allowing any voice to speak any supported language, which simplifies localization for global voice assistant applications.

Female voices: Aria, Nova, Luna, Sage

Female voices provide versatile options for different interaction styles. Aria offers a warm, clear narration suitable for guides. Nova brings a bright, upbeat energy ideal for quick updates. Luna delivers a soft, storytelling tone for longer content, while Sage provides a calm, instructional voice for help desks.

These presets allow developers to create nuanced conversational flows. The distinct timbres help users distinguish between different bot functions or personas within a single application.

  • Aria: Warm, clear narrator
  • Nova: Bright, upbeat
  • Luna: Soft, storytelling
  • Sage: Calm, instructional

Male voices: Felix, Milo, Atlas, Orion

Male voices cover a range of professional and casual tones. Felix is energetic and promotional, while Milo offers a friendly, conversational style for everyday interactions. Atlas provides a neutral, news-like delivery, and Orion offers a deep, documentary tone for authoritative content.

Using these voices alongside female options allows for diverse character creation. This variety helps maintain user engagement by avoiding auditory monotony in long-running voice assistants.

  • Felix: Energetic, promo
  • Milo: Friendly, conversational
  • Atlas: Neutral, news
  • Orion: Deep, documentary

Licensing from LibriTTS-R and LibriVox

All voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain audiobook readings. This foundation ensures high-quality, natural-sounding synthetic speech without the need for custom voice cloning.

Each generated file carries an inaudible watermark marking it as synthetic speech. This provides transparency regarding the audio source, which is important for user trust in voice interfaces.

Technical Implementation Details

Sound wave illustration for: Technical Implementation Details

The API endpoint POST https://api.playvm9.com/v1/audio/speech accepts a JSON body, mirroring the structure of the OpenAI speech endpoint. This design allows product teams to integrate VoiceStack’s TTS engine into existing codebases.

Requests require standard fields such as model, input, voice, and response_format. The system processes these inputs to generate audio files in formats like MP3, WAV, or OPUS. By adhering to a familiar request shape, teams can maintain consistent integration patterns across different voice interface components.

RESTful endpoint compatible with OpenAI SDKs

The endpoint POST https://api.playvm9.com/v1/audio/speech mirrors the OpenAI speech endpoint structure. This compatibility allows teams to use official OpenAI SDKs by simply changing the base_url to https://api.playvm9.com/v1. The same request shape and authentication headers ensure that existing integrations work with minimal code adjustments, reducing development time for chatbot and voice assistant products.

  • Standard RESTful POST method
  • Compatible with official OpenAI SDKs
  • Identical request shape to OpenAI

JSON request body with standard fields

Requests are sent as JSON bodies containing model ("tts-1" or "tts-1-hd"), input (text up to 4096 characters), voice, and response_format. Optional fields include speed (0.25–4.0) and language (ISO code). This standardized structure simplifies payload construction for developers building audio-driven applications, ensuring predictable behavior across different voice personalities and output formats.

  • model: tts-1 or tts-1-hd
  • input: text string, max 4096 chars
  • response_format: mp3, wav, opus, flac, pcm

Base URL configuration for existing integrations

Configure your SDK or HTTP client by setting the base URL to https://api.playvm9.com/v1. Authentication is handled via the Authorization: Bearer tts-... header using your unique API key. This simple configuration change enables immediate access to the TTS engine, allowing teams to swap or add voice capabilities without modifying the underlying request logic or data parsing.

  • Base URL: https://api.playvm9.com/v1
  • Auth: Bearer token in header
  • No custom SDK required

Language Support and Detection

Sound wave illustration for: Language Support and Detection

VoiceStack supports 23 languages, including English, Spanish, German, French, Japanese, Korean, Arabic, Hindi, and others. Each voice can speak any supported language, ensuring consistent audio quality across diverse user bases without regional model dependencies.

Support for major global languages

Our API covers 23 languages, from Arabic and Chinese to Swahili and Turkish. Every voice preset, whether Aria or Atlas, handles any supported language natively. This allows product teams to build unified voice interfaces for global markets without managing separate voice models for each region.

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

Automatic language guessing from script

When the language field is omitted, the system infers the language from the input script. Latin script text defaults to English. For non-English text using Latin characters, such as German or Dutch, the API may default incorrectly if the language parameter is not explicitly set.

Explicit language parameter for precision

For reliable results with non-English Latin-script text, always set the language parameter to the appropriate ISO code. This ensures the synthesis engine applies the correct phonetic rules. Languages like Vietnamese, Thai, and Ukrainian are not supported, so requests for those scripts will fail or default to English.

Language Support and Detection

Pricing Model for Scalable Products

Sound wave illustration for: Pricing Model for Scalable Products

VoiceStack charges $0.01 per 1,000 characters, scaling linearly with usage. This flat rate simplifies budgeting for chatbot integrations. Generated audio is returned in the response and not stored, keeping storage costs separate from processing fees.

Flat rate per million characters

Billing is calculated at $10 per 1 million characters, including spaces and punctuation. This predictable model allows product teams to estimate costs accurately based on expected conversation lengths without hidden tiers.

No monthly fees or seat limits

Accounts operate on a pay-as-you-go basis with no monthly subscriptions or per-seat charges. You only pay for the synthesis you consume, making it ideal for startups and fluctuating workloads.

Failed requests incur zero cost

If synthesis fails, the charge is returned in full. This ensures you are not penalized for transient API errors or invalid inputs, protecting your operational budget during testing and production.

Data Privacy and Processing

Sound wave illustration for: Data Privacy and Processing

VoiceStack processes your input text solely to generate the requested audio file. The text is not used for training our models, ensuring your proprietary dialogue or content remains distinct from the voice engine’s development data. Generated audio is returned in the response and not stored on our servers, minimizing data retention risks.

We retain request metadata—such as timestamp, character count, and status—exclusively for accurate billing and usage tracking. This data does not include the audio content or the raw input text, ensuring your linguistic data remains private while providing transparent records of API consumption.

Input text used only for synthesis

Your text is processed strictly to produce the audio output. It is not ingested into our training datasets, preserving the uniqueness of your chatbot’s dialogue or voice assistant scripts.

No training on generated audio data

The audio files returned in the response are not stored long-term. We do not use your generated speech to improve our voice models, ensuring your specific use cases remain separate from our general engine development.

Metadata retained for billing accuracy

We keep request metadata, including time, character count, and status, solely to calculate charges accurately. This ensures transparent billing without retaining the actual content you sent or received.

Limitations and Technical Constraints

Sound wave illustration for: Limitations and Technical Constraints

VoiceStack’s architecture prioritizes simplicity and speed, which introduces specific constraints for large-scale deployments. Understanding these limits helps product teams design robust fallback mechanisms.

4096 character limit per request

Each synthesis request accepts a maximum of 4,096 characters. For longer content, your application must split the text into segments and join the resulting audio files. Short phrases may take a few seconds to generate, while longer texts process at roughly real-time speed.

No SSML or phoneme dictionary support

The API does not support SSML markup or custom pronunciation lexicons. Voice control relies entirely on the eight preset voices and the specified language parameter. This simplifies integration but means you cannot fine-tune emphasis or pause timing at the word level.

Standard rate limits of 60 requests per minute

Each API key is limited to 60 requests per minute. New accounts start with a lower threshold of 10 requests per minute until the first top-up is processed. Requests are processed sequentially, so heavy load may introduce queueing delays.

Rate limits are enforced per key, so distributing load across multiple keys or batching requests can help maintain throughput for high-traffic voice assistants.

Getting Started with VoiceStack

Sound wave illustration for: Getting Started with VoiceStack

Creating an account is the first step toward integrating VoiceStack into your voice assistant or chatbot. You can sign up directly on the /get-api-key/ page using either your Google credentials or a standard email and password combination. This process is designed to be quick, allowing your product team to move from sign-up to testing in minutes without unnecessary friction.

Once registered, you receive a unique API key immediately. This key authenticates your requests to the API endpoint, enabling you to start generating audio files right away. You can also issue a new key at any time, which instantly invalidates the previous one for added security control.

Sign up with Google or email credentials

Creating an account is the first step toward integrating VoiceStack into your voice assistant or chatbot. You can sign up directly on the /get-api-key/ page using either your Google credentials or a standard email and password combination. This process is designed to be quick, allowing your product team to move from sign-up to testing in minutes without unnecessary friction.

Once registered, you receive a unique API key immediately. This key authenticates your requests to the API endpoint, enabling you to start generating audio files right away. You can also issue a new key at any time, which instantly invalidates the previous one for added security control.

Receive API key for immediate testing

Creating an account is the first step toward integrating VoiceStack into your voice assistant or chatbot. You can sign up directly on the /get-api-key/ page using either your Google credentials or a standard email and password combination. This process is designed to be quick, allowing your product team to move from sign-up to testing in minutes without unnecessary friction.

Once registered, you receive a unique API key immediately. This key authenticates your requests to the API endpoint, enabling you to start generating audio files right away. You can also issue a new key at any time, which instantly invalidates the previous one for added security control.

Access documentation and curl examples

The /get-api-key/ dashboard displays your key, balance, and usage statistics alongside a ready-to-use cURL example. This allows you to test the POST request to the speech endpoint immediately. For deeper integration details, visit the /docs/ page to explore the standard JSON body fields and SDK configuration.

You can also start with a free trial credit of $0.50 (valid for 7 days) if you sign in with Google. This provides ample characters to test latency and voice quality before committing to top-ups via USDT on the TRON network.

  • Use the provided cURL example to test the endpoint instantly
  • Check your balance and character usage directly on the dashboard
  • Sign up with Google for immediate trial credit access

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. Your voice assistant must understand context, not just commands, to feel truly conversational to the user.

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. Node.js documentation
  3. W3C WAI: making audio and video accessible
  4. LibriTTS-R corpus (OpenSLR 141)
  5. openai-python on GitHub
  6. OpenAI API reference: create speech
  7. LibriVox public-domain audiobooks
  8. IETF RFC 7845: Ogg Opus
  9. MDN: Web Speech API
  10. ISO 639 language codes
  11. Creative Commons Attribution 4.0
  12. FFmpeg documentation

More on this site