Frequently asked questions
Do I need to specify the language code for Arabic text?
Yes. The API guesses the language from the script, but Latin text defaults to English. For Arabic script, it is detected automatically. However, setting the language parameter explicitly ensures accurate pronunciation and prevents misinterpretation, especially for mixed scripts.
Can I use a male voice for Arabic announcements?
Yes. All eight voices support Arabic. You can use male voices like Felix, Milo, Atlas, or Orion for announcements. Each voice offers a distinct tone, from energetic to neutral, suitable for different transport contexts without additional cost.
Does the API handle Arabic diacritics automatically?
The API processes the input text as provided. It does not have a custom pronunciation lexicon or SSML support to adjust diacritics manually. Standard Arabic text is synthesized correctly, but specific diacritic nuances may not be guaranteed without testing.
What is the best format for storing Arabic audio files?
Choose MP3 for balanced quality and size, or WAV for lossless audio. FLAC is also available. The choice depends on your storage constraints and playback requirements. All formats are supported via the response_format parameter in your API request.
Can I mix English and Arabic in one request?
Yes. You can include both languages in a single input text up to 4096 characters. The API detects the language from the script. Ensure the text is coherent to maintain natural flow across language transitions in the generated audio.
Is the Arabic voice generation slow compared to English?
Generation speed is roughly real-time, about 25 characters per second, regardless of language. Processing time depends on text length and queue load. Short phrases take a few seconds. There is no specific slowdown for Arabic compared to other supported languages.
Do I need to split long Arabic texts manually?
Yes, if the text exceeds 4096 characters. You must split the text into parts, send separate requests, and join the resulting audio files. The API does not automatically split or join audio for longer inputs within a single call.
Can I use the same API key for multiple languages?
Yes. One API key works for all 23 supported languages, including Arabic. You do not need separate keys for different languages. The key authenticates all requests, and the language is determined by the input text or language parameter.
Is there a limit to how many Arabic announcements I can generate?
Rate limits apply: 60 requests per minute per key. There is no specific limit on the number of Arabic announcements, only the request rate. Longer texts count as multiple requests if split. Failed requests do not consume your limit.
Does the API support voice cloning for Arabic speakers?
No. Custom voice cloning is not offered. You can only use the eight preset voices (Aria, Nova, Luna, Sage, Felix, Milo, Atlas, Orion). These voices are based on licensed recordings and can speak Arabic natively.
Can I adjust the speed of Arabic announcements?
Yes. You can set the speed parameter between 0.25 and 4.0. This applies to all supported languages, including Arabic. Adjusting speed helps match the announcement pace to your transport schedule or audience preferences.
Are there any specific requirements for Arabic script encoding?
Standard UTF-8 encoding is sufficient for Arabic text. The API accepts any valid Unicode characters within the 4096 character limit. No special encoding formats are required beyond standard text input in your JSON request body.
How do I verify that the generated audio is in Arabic?
The API returns the audio file directly. You can listen to the output to confirm the language. The language is determined by your input text or the optional language parameter. There is no metadata field indicating the language in the audio file itself.
Can I use this API for live bus announcements?
Yes, but with caveats. The API is not streaming; it returns audio files after processing. For live use, pre-generate announcements or handle latency. Requests are processed sequentially, so high volume may cause delays during peak times.
Is the API suitable for formal transit announcements?
Yes. Voices like Atlas (neutral, news) or Sage (calm, instructional) are well-suited for formal transit announcements. The clear, synthetic speech is professional and consistent, suitable for public address systems in transport environments.
Do I need an internet connection to play the generated audio?
No. Once you receive the audio file via the API response, you can download and store it locally. Playback does not require an internet connection. The API is only needed for the initial generation and download of the file.
Can I download the audio files in WAV format?
Yes. Set the response_format parameter to 'wav' in your request. This provides lossless audio quality. Other formats like MP3, OPUS, FLAC, and PCM are also available depending on your storage and quality needs.
Is the API free for small transport fleets?
No. There is no free tier, but a free trial is available for new Google sign-ins. You pay $0.01 per 1,000 characters. A free trial of $0.50 is available for new Google sign-ins. All other accounts require a top-up via USDT on TRON to generate audio.
How do I troubleshoot errors in Arabic text processing?
Check the API response for error codes. Ensure text is under 4096 characters and properly encoded. Verify your API key is active. Failed requests are free. Contact support via email if issues persist, as there is no live chat.
Can I use the API on my own server?
Yes. The API is a cloud service accessible via HTTP POST. You can integrate it into any server environment that can make HTTP requests. No specific server hardware is required, just internet connectivity and the ability to handle JSON responses.
Do you provide sample Arabic text for testing?
No. The API does not provide built-in sample text. You must supply your own Arabic text in the input field. You can use any standard Arabic phrase to test voice quality and language detection before full deployment.
Is the API accessible from mobile devices?
Yes. You can send requests from any device with internet access and HTTP capabilities. The API is platform-agnostic. You can integrate it into mobile apps or use it directly from mobile browsers or scripts.
Can I use the API for radio broadcasting?
Yes. The generated audio files can be used for radio broadcasts. Ensure you comply with any licensing requirements. The audio quality is suitable for broadcast, depending on the format chosen (e.g., WAV or FLAC).
Do I need to pay extra for Arabic language support?
No. Arabic is one of the 23 supported languages at the same price. There are no additional fees for specific languages. The cost is based solely on the character count, regardless of the language used.
How do I renew my API key?
You can issue a new key at any time from your account dashboard. The old key stops working immediately when the new one is issued. This is useful for security or rotation without downtime, as long as you update your integration.
Can I use the API for emergency announcements?
Yes. The API is reliable for emergency announcements, but consider latency. Since requests are processed sequentially, high volume may delay responses. Pre-generate critical messages if possible. The synthetic voice is clear and consistent for urgent communications.
Is the API secure for passenger data?
Input text is used only to produce audio and is not stored or used for training. Request metadata is kept for billing. The API does not collect personal passenger data beyond the text you provide. Ensure your own transmission is secure.
Can I integrate the API with my existing transport software?
Yes. The API uses standard JSON over HTTP, similar to OpenAI’s format. You can integrate it with any software that can send HTTP POST requests. Use the OpenAI SDK by changing the base URL for easier integration.
Do you offer technical support for API integration?
Support is available via email only. There is no phone support or live chat. You can contact us for help with integration issues. Response times are not guaranteed. Check the documentation for details.
Can I use the API for multilingual announcements?
Yes. You can generate announcements in any of the 23 supported languages, including Arabic and English. You can mix languages in one request or generate separate files for different languages. The API handles language detection automatically or via parameter, but Latin text defaults to English.
Understanding Arabic script in TTS
The system detects Arabic script to infer the language automatically, but explicit configuration ensures consistency across transport announcements. Relying solely on script detection can lead to errors if mixed-language texts are processed without clear boundaries.
Setting the language parameter explicitly overrides default guessing, guaranteeing that Arabic text is synthesized correctly regardless of surrounding Latin characters. This precision is critical for fleet-wide audio updates in multilingual regions.
Latin vs Arabic script detection
Arabic characters are identified by their Unicode range. When the input contains only Arabic script, the system defaults to Arabic pronunciation rules without requiring additional configuration.
- Auto-detection works for pure Arabic text blocks.
Manual language code setting
Specify language: "ar" in the JSON body to enforce Arabic synthesis. This prevents misinterpretation when Arabic text appears alongside English or other Latin-script announcements in the same batch.
- Use
arfor clear, unambiguous Arabic output.
Default language guessing
If no language is set, Latin script defaults to English. Non-English Latin text may be mispronounced, so always define the language for mixed-content transport scripts to ensure clarity for diverse passenger groups.
- Latin text defaults to English unless specified.
Selecting the right voice for Arabic
Paisita Voices offers eight distinct presets suitable for Arabic transport announcements. Each voice is based on licensed recordings from LibriVox, ensuring high-quality synthetic speech. All eight voices can natively speak Arabic, allowing you to choose a tone that matches your fleet's identity.
Selecting the correct voice depends on your announcement style. Whether you need a calm instructional tone for safety updates or an energetic voice for promotional messages, the API supports all 23 languages with any of the available presets. This flexibility ensures consistency across mixed-language regions.
Female voices: Aria, Nova, Luna, Sage
For clear, professional announcements, consider Aria for warm narration or Sage for calm, instructional updates. Luna offers a soft storytelling tone suitable for detailed route information, while Nova provides a bright, upbeat delivery for general alerts. These voices ensure that Arabic text is rendered with natural prosody.
Using a female voice can help distinguish between different types of announcements, such as separating safety warnings from regular schedule updates. The consistent vocal quality helps passengers identify the source of the information quickly.
- Aria: warm, clear narrator
- Nova: bright, upbeat
- Luna: soft, storytelling
- Sage: calm, instructional
Male voices: Felix, Milo, Atlas, Orion
Felix delivers an energetic tone ideal for promotional messages or urgent alerts. Milo offers a friendly, conversational style suitable for driver-passenger communication. Atlas provides a neutral, news-style delivery, making it a reliable choice for standard route announcements. Orion’s deep voice is suited for documentary-style or formal updates.
Male voices often project well in noisy transport environments. The neutral tone of Atlas ensures that the message is perceived as authoritative and clear, which is critical for safety-critical information in large transport teams.
- Felix: energetic, promo
- Milo: friendly, conversational
- Atlas: neutral, news
- Orion: deep, documentary
Neutral tone selection
For large transport teams requiring consistent announcements, Atlas is often the preferred choice due to its neutral, news-style delivery. It avoids emotional bias, ensuring that information is delivered objectively. This neutrality is particularly important when broadcasting to diverse audiences in mixed-language regions.
When the specific emotional tone is less important than clarity and consistency, a neutral voice reduces the risk of misinterpretation. All voices carry an inaudible watermark, confirming the synthetic origin of the audio for compliance purposes.
Preparing your announcement text
Before sending text to the API, verify that your announcement content fits within the 4096-character limit per request. Spaces and punctuation count toward this total, so plan your script length accordingly. For transport teams broadcasting mixed-language routes, ensure the language parameter is explicitly set to 'ar' to prevent incorrect English default detection.
Character limit considerations
Each API call processes up to 4096 characters. Exceeding this limit causes the request to fail. Calculate your total script length, including spaces, before generating audio.
- Limit: 4096 characters per request
- Includes spaces and punctuation
- Failures are free
Handling special characters
Arabic text may include punctuation or diacritics. The API handles standard Unicode characters without issue. SSML is not supported. Ensure your text is plain UTF-8 encoded.
- UTF-8 encoding required
- No SSML support
- Standard Unicode punctuation works
Splitting long texts
For announcements exceeding 4096 characters, split the text into smaller segments. Generate audio for each segment individually, then join the resulting audio files. This ensures no data loss and maintains playback continuity.
- Split at logical sentence breaks
- Generate each segment separately
- Join audio files client-side
Making the API request
Send a POST request to https://api.paisita.info/v1/audio/speech with your API key in the Authorization header. The JSON body defines the synthesis parameters, ensuring the system processes your transport announcements correctly.
JSON payload structure
Include the model (tts-1 or tts-1-hd), input text, and voice preset. Ensure the text does not exceed 4096 characters. The system accepts standard JSON formatting for reliable parsing.
- model: tts-1 or tts-1-hd
- input: your announcement text
- voice: e.g., atlas or nova
Setting the language parameter
Explicitly set the language field to "ar" for Arabic. Without this, the API may default to English for Latin-script inputs, though Arabic script detection usually works correctly. Precise tagging ensures accurate pronunciation.
Choosing the response format
Select response_format from mp3, wav, opus, flac, or pcm. MP3 is recommended for general playback due to its balance of quality and file size. WAV offers lossless audio.
Handling the response
The API returns the generated audio as binary data within the HTTP response body. You must handle this raw stream directly. The response headers indicate the content type, ensuring your client library correctly identifies the audio format you requested.
Receiving binary audio data
Upon successful synthesis, the response contains the raw audio bytes. Your application should read this stream immediately to avoid memory issues. The data matches the response_format specified in your request, whether MP3, WAV, or other supported formats.
- Stream the response body directly to disk or memory.
- Verify the Content-Type header matches your requested format.
- Close the connection after reading the full payload.
Saving MP3 or WAV files
Write the received binary data to a file with the correct extension. For transport announcements, MP3 is efficient for storage, while WAV provides uncompressed quality for immediate playback systems. Ensure your file naming convention includes the announcement ID for easy tracking.
- Use .mp3 for compressed, widely compatible playback.
- Use .wav for high-fidelity internal systems.
- Append metadata tags if your player supports them.
Error handling basics
If synthesis fails, the API returns an error JSON object with a status code and message. Failed requests are free, so you can retry without incurring charges. Log the error details to diagnose issues like invalid text or rate limits.
- Check HTTP status codes (200 vs 4xx/5xx).
- Parse the error JSON for specific failure reasons.
- Retry on transient errors; failed attempts are refunded.
Testing in production
Before deploying Arabic announcements to transport fleets, validate the API integration using the $0.50 free trial credit. This allows testing up to 50,000 characters within a seven-day window without financial commitment. Verify that the endpoint correctly interprets Arabic script and generates audio files compatible with your playback hardware.
Using the free trial credit
Sign in via Google to activate the trial balance. Use this credit to test voice presets like Aria or Felix with Arabic text. Note that the trial expires in seven days, so schedule your validation tests accordingly.
- Activate trial by signing in with Google
- Test up to 50,000 characters
- Trial expires in 7 days
- No credit card required
Monitoring character usage
Track character consumption in the dashboard after each request. Since pricing is $0.01 per 1,000 characters, accurate monitoring prevents unexpected charges once the trial ends. Failed requests do not consume credit, ensuring you only pay for successful audio generation.
- Check balance in the dashboard
- Pricing: $0.01 per 1,000 characters
- Failed requests are free
- Spaces and punctuation count as characters
Verifying audio quality
Listen to generated MP3 or WAV files for clarity in noisy transport environments. Ensure the chosen voice, such as Atlas for news-style announcements, maintains a neutral tone suitable for public address systems. Verify that the inaudible watermark is present without affecting audio fidelity.
- Test audio in noisy environments
- Select neutral voices like Atlas
- Verify inaudible watermark presence
- Check MP3/WAV compatibility
Common issues with Arabic text
Arabic text often triggers language guessing errors because the system defaults to English for Latin-script inputs. If your announcement text uses Arabic script, the model may still attempt English pronunciation unless you explicitly define the language code. This results in unnatural, phonetically incorrect audio for transport announcements.
Voice selection impacts clarity significantly. While all eight voices support Arabic, some deliver better articulation for complex transport terminology. For example, Atlas provides a neutral tone suitable for news-style updates, while Luna offers a softer delivery for station announcements. Test each voice to ensure it meets your team's standards.
Missing language tag errors
The API guesses the language from the script if the language parameter is omitted. Arabic script is correctly identified, but Latin-script Arabic text may default to English. Always set language: "ar" to ensure accurate pronunciation and avoid misinterpretation of transport-specific terms.
Incorrect language detection leads to audible errors in announcements. For example, Arabic words written in Latin script might be read as English, causing confusion for passengers. Explicitly defining the language parameter ensures consistent output across all your announcements.
- Set
language: "ar"for all Arabic text. - Verify script type before sending requests.
- Test with sample phrases to confirm accuracy.
Incorrect voice selection
Not all voices perform equally with Arabic. Male voices like Atlas and Orion often provide clearer, more authoritative tones for transport announcements. Female voices like Aria and Luna may sound too soft or storytelling-oriented for urgent updates. Choose voices that match the urgency of your announcements.
Experiment with different voices to find the best fit for your specific use case. Consider the context of each announcement, such as boarding calls versus delay notifications, to select the most appropriate voice.
- Use Atlas for neutral, clear announcements.
- Try Orion for deep, documentary-style updates.
- Test Luna for softer, informational messages.
Character count limits
Each API request is limited to 4096 characters. Arabic text can consume character counts differently due to script complexity. Ensure your announcements fit within this limit to avoid truncation. Split longer texts into multiple requests and join the audio files for seamless playback.
Monitor your character usage closely, especially when generating multiple announcements. Failed requests do not incur charges, but successful ones do. Plan your text lengths to optimize efficiency and reduce costs.
- Limit each request to 4096 characters.
- Split long texts into smaller parts.
- Join audio files for continuous playback.

Optimizing for transport use
For transport announcements, clarity is paramount. Adjusting the speed parameter between 0.25 and 4.0 allows you to optimize audio for noisy environments or specific listener demographics. Slower rates ensure intelligibility in crowded stations, while faster rates suit brief alerts.
Selecting the correct response format ensures compatibility with your existing PA systems. MP3 offers small file sizes for quick deployment, while WAV provides uncompressed quality for high-fidelity speakers. Choose formats that balance storage efficiency with playback reliability across your fleet.
Speed adjustments for clarity
Fine-tune the speed parameter to match ambient noise levels and announcement length. Lower speeds (e.g., 0.75) enhance comprehension for complex route information, while higher speeds (e.g., 2.0) work for simple alerts. Test various rates to ensure consistent intelligibility across different vehicle acoustics.
Format selection for playback
Choose between MP3, WAV, OPUS, FLAC, or PCM based on your hardware constraints. MP3 is efficient for storage, while WAV provides uncompressed quality. Avoid formats that require additional decoding steps if your playback devices are limited.
- MP3: Best for network efficiency
- WAV: Highest compatibility
- OPUS: Efficient for streaming
Batch processing strategies
Split long texts into parts of up to 4096 characters to manage request limits effectively. Process multiple announcements concurrently where possible, but be aware that requests are processed sequentially under load. Monitor your 60 requests per minute limit to avoid delays during peak hours.
Security and compliance
Secure delivery of transport announcements relies on robust API key management. Each account maintains one active key, which can be rotated instantly via the dashboard. When a new key is issued, the previous one stops immediately, ensuring no unauthorized access persists during transitions. This mechanism simplifies security audits for large teams without requiring complex certificate infrastructure.
API key management
The API key authenticates every request. Rotate keys directly from the account page. Old keys cease working at once, preventing legacy device conflicts.
- One active key per account.
- Instant invalidation of old keys.
- No expiration date for active keys.
Data privacy for announcements
Input text is used solely to produce the audio file and is not stored or used for training models. Generated audio is returned in the response and is not retained on our servers. Only request metadata, such as timestamp, character count, and status, is kept for billing purposes. This ensures that specific announcement content remains transient.
- Text not used for training.
- Audio not stored post-delivery.
- Metadata kept only for billing.
Watermark verification
Every generated audio file contains an inaudible watermark identifying it as synthetic speech. This feature allows transport operators to verify that announcements are AI-generated rather than human-recorded, which may be relevant for compliance or content auditing. The watermark is embedded directly into the audio stream and persists across format conversions.
Support and troubleshooting
Support for Paisita Voices is handled exclusively via email. There is no phone number or live chat available for immediate assistance. All inquiries regarding API integration, voice selection, or technical issues are directed to our support team through the contact page.
Email support contact
Send your questions to the address provided on our contact page. Include your API key and a clear description of the issue. Response times vary based on volume, but we prioritize transport-related queries.
Rate limit explanations
Each API key allows 60 requests per minute. Before your first top-up, this limit is reduced to 10 requests per minute. If you exceed the limit, you will receive a 429 error. Requests are processed sequentially, so under load, a request may wait in a queue.
- 60 requests/minute after top-up
- 10 requests/minute before first top-up
- Requests are processed one after another
Billing dispute process
Failed synthesis requests are free, with charges returned in full. If you notice an incorrect charge, contact us via email with your account details. We do not offer SLAs or certified data handling guarantees, but we ensure accurate billing based on character counts.
Future updates and features
The API infrastructure is periodically updated to enhance synthesis quality and efficiency without altering the core interface. These updates ensure consistent performance for high-volume transport announcements.
We monitor voice capabilities and developer tools to align with standard industry practices, ensuring your integration remains compatible with evolving requirements.
Model updates
The underlying synthesis models may receive improvements to audio clarity and processing speed. These changes apply to both tts-1 and tts-1-hd engines.
Voice additions
New voice presets may be introduced to support additional languages or tones. Existing voices will remain available, but new options could expand your announcement library.
SDK improvements
While we do not maintain an official SDK, updates to the OpenAI-compatible endpoint structure will be documented. Developers using standard HTTP clients or the OpenAI SDK will benefit from improved base URL compatibility.
Community and documentation
All technical references, endpoint definitions, and parameter details are centralized in the documentation, accessible directly from the site’s navigation. This resource provides the definitive schema for the /v1/audio/speech endpoint, ensuring your integration aligns precisely with current API capabilities without ambiguity.
For developers seeking quick implementation, example code repositories demonstrate standard requests using the OpenAI SDK with the adjusted base URL. These examples cover common transport scenarios, such as sending multilingual texts and handling binary audio responses, allowing your team to deploy announcements with minimal configuration overhead.
API docs location
The complete API reference is hosted at /docs/ and includes detailed descriptions of all available voices, supported languages, and response formats. It outlines the exact JSON structure required for the POST /v1/audio/speech request, including parameters for speed, model selection, and language codes.
This documentation also clarifies the distinction between tts-1 and tts-1-hd models, ensuring you select the appropriate engine for your specific audio quality and cost requirements. It serves as the primary reference for any integration or troubleshooting activity.
Example code repositories
We provide ready-to-use code snippets in our documentation section, demonstrating how to send text and receive audio files via cURL or popular HTTP libraries. These examples illustrate the correct headers, including the Authorization: Bearer token, and show how to parse the binary response.
The examples specifically address transport use cases, such as splitting long announcement texts into segments under 4096 characters and reassembling them. This ensures that large-scale broadcast systems can handle extensive route updates or safety instructions without exceeding character limits.
User guides
For non-technical stakeholders, user guides explain how to interpret character counts and manage API keys through the dashboard at /get-api-key/. These guides detail how to monitor usage and understand billing, ensuring transparent cost management for your team.
The guides also cover basic error handling, explaining how to interpret status codes when a request fails or when rate limits are reached. This helps operations teams maintain consistent audio output during high-traffic periods without unnecessary downtime.
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.
23 languages
Any voice speaks any of these languages.
