Multilingual TTS API for Global Transport Fleets

By Paisita editorial team · Updated 2026-11-19

This page details the Paisita Voices API, a reliable text-to-speech solution engineered for large transport fleets requiring consistent, multilingual public announcements. We provide a straightforward integration path for teams managing mixed-language regions, eliminating the need for enterprise contracts or mobile app dependencies while ensuring unified audio updates across your entire network.

You will find specific technical requirements, supported language lists, and voice profiles optimized for transit environments. The page outlines our character-based pricing, data privacy standards, and deployment reliability, giving you all the concrete facts needed to evaluate whether this API fits your operational workflow. Use this information to plan your integration or proceed directly to key generation.

Multilingual TTS API for Global Transport Fleets

Frequently asked questions

Is the Arabic voice generation suitable for formal transit announcements?

Yes. Atlas provides a neutral, news-style voice ideal for formal announcements. Luna delivers a soft, storytelling voice, while Sage provides a calm, instructional tone. All voices support Arabic.

Can I use the same voice for both English and Arabic announcements?

Yes. Each of the eight preset voices can speak any of the 23 supported languages, including Arabic and English. You do not need separate voices for different languages. Simply specify the ISO language code in the request to ensure correct pronunciation, or leave it unset for automatic detection based on the text script.

Do I need an internet connection to play the generated audio files?

The API requires an internet connection to generate the audio files via the POST endpoint. However, once generated, the audio files are returned in the response. You can download and store these files locally or on your servers, allowing playback in your transport vehicles or stations without an active internet connection.

How many characters can I send in a single API request?

You can send up to 4096 characters per API request. For longer announcements, split the text into parts of this limit and join the resulting audio files. This ensures consistent generation quality. Each request counts toward your character-based billing, so precise splitting helps manage costs and ensures complete audio output for all content.

Are the voices based on real human recordings?

Yes. All eight voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain audiobook readings. Each generated file also carries an inaudible watermark to mark it as synthetic speech,

Is there a limit to how many announcements I can generate per minute?

Yes. The rate limit is 60 requests per minute per API key. Before the first top-up, this limit is 10 requests per minute. Requests are processed sequentially, so under heavy load, a request may wait in a queue. This applies to all accounts, ensuring fair usage across the network without requiring enterprise contracts.

Does the API support streaming audio for live updates?

No. The API does not support streaming, WebSocket, or real-time partial audio. You must send the full text and receive the complete audio file in the response. This fits standard transport announcement workflows where files are generated, stored, and played back, rather than requiring live, low-latency audio streaming for immediate broadcast.

Can I clone a specific announcer's voice for my fleet?

No. We do not offer custom voice cloning, voice uploading, or voice design. You must choose from the eight preset voices: Aria, Nova, Luna, Sage, Felix, Milo, Atlas, or Orion. These licensed recordings are available for all supported languages. If you need a specific announcer, you must select the closest preset voice.

What happens if an API request fails? Am I charged?

No. Failed requests are free. If synthesis fails, the charge is returned in full. You only pay for successful generations based on the character count. This protects your budget during network issues or invalid inputs. Ensure your text is under 4096 characters and uses a valid voice to minimize potential errors during generation.

Do you offer an SLA for 99.9% uptime during peak hours?

No. We do not offer an SLA, uptime guarantee, or enterprise contracts. The service is designed for reliable character-based generation without strict uptime promises. Requests are processed sequentially, so performance may vary under load. This model suits transport teams that can handle occasional queue delays without requiring guaranteed 99.9% availability for critical broadcasts.

Is there a WordPress plugin available for my transport website?

No. We do not offer a WordPress plugin, browser extension, or mobile SDK. You can use the official OpenAI SDKs by changing the base URL, or use plain HTTP requests. This gives you full control over integration into your existing transport management systems without relying on specific platform plugins or third-party wrappers for your announcements.

Can I download the audio files in WAV format for broadcast?

Yes. The API supports WAV, MP3, OPUS, FLAC, and PCM formats. You can specify the desired format in the request body. WAV is suitable for broadcast quality, while MP3 offers smaller file sizes. All formats are generated from the same engine, ensuring consistent audio quality regardless of the format you choose for your transport announcements.

Do you support phoneme-level pronunciation adjustments?

No. The API does not support SSML, phoneme dictionaries, or custom pronunciation lexicons. Pronunciation is handled automatically by the voice engine based on the input text and language code. If specific pronunciation is critical, you may need to adjust the input text spelling. We do not offer fine-grained phoneme control for individual words or sounds.

How do I renew my API key if it gets compromised?

You can issue a new key at any time from your account dashboard. When a new key is generated, the old one stops working immediately. This ensures secure access for your transport team. You can manage keys via the /get-api-key/ page, where you can also view your balance, usage, and trial credit status for auditing.

First request

curl

curl https://api.paisita.info/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 transport teams choose API over plugins

Sound wave illustration for: Why transport teams choose API over plugins

Transport fleets require consistent audio updates across mixed-language regions without the overhead of enterprise contracts. A centralized API allows teams to manage voice announcements from a single endpoint, ensuring uniform deployment across all vehicles and stations.

Centralized voice management

Centralized voice management

Manage voice profiles and language settings via a single API key. Updates to text or voice presets propagate instantly to the entire fleet, eliminating the need to update individual devices or applications.

No mobile app dependencies

Unlike mobile applications, the API works directly with existing transport hardware and software. There are no app stores, updates, or compatibility issues to manage for each vehicle’s onboard system.

Unified fleet announcements

Deliver consistent passenger information across Arabic, English, and other supported languages. The system handles character limits and audio joining automatically, ensuring seamless playback regardless of the message length.

Supported languages for public transit

Sound wave illustration for: Supported languages for public transit

Paisita Voices supports 23 languages, enabling consistent announcements across diverse passenger demographics. Our API handles text in any supported language without requiring separate voice models or regional integrations.

Each voice is capable of speaking all 23 languages, ensuring uniform audio quality regardless of the target region or language mix in your fleet.

Arabic voice synthesis

Arabic is fully supported with natural pronunciation. You can specify the ISO code in the request to ensure accurate rendering, though the system also attempts to guess the language from the script.

Arabic text is processed with the same reliability as other supported languages, maintaining consistent speed and character limits.

  • Full support for Arabic script
  • Optional ISO language code specification
  • Consistent voice output across all presets

English and European languages

We support 15 European languages including English, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Russian, Swedish, Danish, Finnish, Norwegian, Greek, and Turkish. This covers most major transit networks in Europe and North America.

Set the language parameter explicitly for non-English Latin text to avoid default English fallback.

  • English, German, French, Spanish
  • Italian, Portuguese, Dutch, Polish
  • Russian, Swedish, Danish, Finnish
  • Norwegian, Greek, Turkish

Asian and African language support

Asian languages include Chinese, Japanese, Korean, and Malay. African support includes Swahili. These languages are processed with the same character limits and speed constraints as other languages.

All 23 supported languages are available for immediate integration without additional configuration or voice training.

  • Chinese, Japanese, Korean, Malay
  • Swahili
  • No separate voice models needed

Voice profiles for transit environments

Sound wave illustration for: Voice profiles for transit environments

Consistent audio branding reduces passenger confusion in mixed-language regions. Our eight voice presets ensure announcements remain clear and recognizable across diverse transit environments, from busy metro stations to intercity buses.

Atlas: neutral news style

Atlas delivers a neutral, news-style delivery ideal for standard transit updates. Its clear articulation ensures route changes and safety instructions are understood quickly, even in noisy terminal environments.

Milo: friendly conversational

Milo provides a friendly, conversational tone suitable for customer service announcements. This voice profile works well for welcoming messages or helpful travel tips, enhancing the passenger experience without sounding robotic.

Luna: soft storytelling tone

Luna offers a soft, storytelling tone for longer informational segments. Use this voice for detailed station guides or historical transit facts where a calmer, engaging delivery improves comprehension.

Technical requirements for integration

Sound wave illustration for: Technical requirements for integration

Our API exposes a single POST endpoint at https://api.paisita.info/v1/audio/speech, accepting standard JSON bodies. You submit the text, selected voice, and target language, and receive a direct audio file response. This structure eliminates the need for complex middleware or custom transport layer protocols, allowing your existing fleet management software to handle audio generation with minimal code changes.

RESTful JSON endpoint

Requests are processed synchronously, returning the audio file upon completion. For texts exceeding 4,096 characters, split the input into segments and concatenate the resulting audio files. The endpoint supports mp3, wav, opus, flac, and pcm formats, allowing you to optimize bandwidth and storage based on your network infrastructure.

  • Endpoint: POST /v1/audio/speech
  • Max input: 4,096 characters per request
  • Formats: mp3, wav, opus, flac, pcm

OpenAI SDK compatibility

The request schema matches the OpenAI speech endpoint. By updating the base_url to https://api.paisita.info/v1 and using your Paisita API key, you can utilize official OpenAI SDKs directly. This reduces integration time for teams already familiar with standard TTS libraries, ensuring a straightforward deployment without learning a new API structure.

Standard audio formats

Choose between compressed formats like MP3 and OPUS for efficient data transfer, or lossless WAV and FLAC for high-fidelity playback in stations. PCM is available for raw audio processing. This flexibility ensures compatibility with diverse playback hardware across buses, trains, and station platforms.

Audio quality and watermarking

Sound wave illustration for: Audio quality and watermarking

Every audio file generated by the Paisita Voices API carries an inaudible watermark that marks it as synthetic speech. This ensures clear attribution for public announcements without affecting the listening experience for passengers or staff in transport hubs.

Inaudible synthetic watermark

The watermark is embedded directly into the audio stream upon generation. It serves as a verification marker for synthetic content, distinguishing it from human-recorded announcements in mixed-language regions.

Consistent voice output

Voices like Atlas (neutral, news) and Milo (friendly, conversational) deliver uniform tone across thousands of requests. This consistency is critical for large transport teams broadcasting updates across diverse fleets without varying audio quality.

No background noise artifacts

The TTS engine produces clean audio files directly. There is no need for post-processing to remove background hiss or compression artifacts, ensuring the audio is ready for immediate playback on station speakers.

Audio quality and watermarking

Pricing model for large teams

Sound wave illustration for: Pricing model for large teams

Paisita Voices uses a straightforward pay-per-use model: $0.01 per 1,000 characters. Failed requests incur no charge, ensuring you only pay for successful audio generation.

Pay per 1,000 characters

Billing is precise: every character, including spaces and punctuation, counts toward your usage. At $10 per million characters, costs scale linearly with volume, making it predictable for large transport fleets with high-frequency announcement cycles.

No monthly minimums

There are no monthly plans, seats, or minimum spend requirements. You can top up via USDT (TRC-20) with a $5 minimum, and the balance never expires. This structure suits teams with fluctuating announcement needs without locking capital.

Trial credit available

New accounts signing in with Google receive $0.50 in trial credit (approx. 50,000 characters) valid for 7 days. Email-based accounts start without credit. This allows immediate testing of voice quality and integration speed before committing to larger top-ups.

Deployment speed and reliability

Sound wave illustration for: Deployment speed and reliability

Paisita Voices generates audio at roughly real-time speed, processing approximately 25 characters per second. This ensures that long transit announcements are delivered without significant latency, keeping your fleet updates synchronized with schedules.

Real-time generation

Typical short phrases take only a few seconds to synthesize. For longer texts, the system splits input into 4096-character parts and joins the resulting audio files seamlessly, maintaining consistent timing across the entire announcement.

Queue handling under load

Requests are processed sequentially. Under heavy load, a request may wait in a queue before synthesis begins. This ensures stable performance for large transport teams deploying simultaneous updates across mixed-language regions without system crashes.

Instant key rotation

If security requires it, issuing a new API key immediately invalidates the old one. This instant rotation ensures that unauthorized access is cut off immediately, maintaining the integrity of your billing and usage data.

Data privacy for passenger announcements

Sound wave illustration for: Data privacy for passenger announcements

Passenger announcement systems often process sensitive route data or passenger names. Our architecture ensures that the input text you send for synthesis is used solely to generate the audio file and is not used for training our models. This prevents your specific operational data from becoming part of our voice model datasets.

Input text not used for training

When you submit text for a bus or train announcement, that string is processed immediately to create the audio response. It is not retained in a pool for future voice model improvements, ensuring your route specifics remain distinct from our general training data.

No long-term storage of audio

Generated audio files are returned in the response and are not stored on our servers after delivery. This means your fleet's audio announcements are transient, reducing the risk of unauthorized access to your specific content library long after the broadcast.

Metadata kept for billing

We retain request metadata, including timestamp, character count, and status, strictly for billing and usage verification. This minimal data allows us to process your account correctly without storing the actual content of your announcements.

Comparison with traditional TTS systems

Sound wave illustration for: Comparison with traditional TTS systems

Traditional transport TTS systems often require complex SSML markup for basic timing and tone adjustments, adding development overhead. Paisita Voices delivers natural, consistent speech through a simple JSON payload, eliminating the need for custom markup interpretation.

Unlike legacy on-premise solutions that demand dedicated hardware and IT staff, our API integrates directly with existing fleet management software. This reduces deployment time and eliminates the need for specialized voice engineering teams.

No SSML overhead

Legacy systems often rely on verbose SSML tags to control pronunciation and pacing, requiring dedicated parsing logic. Our API accepts plain text and standard parameters, removing the complexity of markup management for simple announcement updates.

  • Plain text input instead of XML markup
  • No custom parsing logic required for basic timing
  • Standard parameters for speed and voice selection

Simper integration

Traditional TTS platforms frequently require proprietary SDKs or complex authentication flows. Our endpoint is compatible with standard OpenAI SDKs, allowing transport teams to integrate using familiar libraries and reducing engineering effort.

  • Compatible with official OpenAI SDKs
  • Standard RESTful JSON structure
  • Easy key-based authentication

Lower operational cost

On-premise TTS solutions incur hardware, maintenance, and licensing costs regardless of usage. Our model charges only for characters processed, with no monthly minimums or seat fees, aligning costs directly with fleet activity.

  • Pay per 1,000 characters
  • No monthly minimums
  • No per-seat licensing fees

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. Her voice remains engaging without fatigue, making her suitable for comprehensive journey updates.

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. openai-node on GitHub
  3. Creative Commons Attribution 4.0
  4. IETF RFC 7845: Ogg Opus
  5. Tether (USDT)
  6. LibriTTS-R corpus (OpenSLR 141)
  7. FFmpeg documentation
  8. ISO 639 language codes
  9. TRON developer hub: TRC-20
  10. OpenAI API reference: create speech
  11. curl documentation
  12. MDN: Web Speech API

More on this site