Text-to-Speech API for Transport Announcements

By TransitVoice editorial team · Updated 2026-11-19

This guide helps transport product teams select the right voice and configuration for station and in-vehicle announcements. We break down the eight available presets by tone, explaining how Atlas, Sage, Aria, and others fit specific use cases like news updates or customer service.

You will also find technical details on language support, latency planning, and integration with the OpenAI SDK. Learn how to handle 4096-character limits, manage licensing via LibriTTS-R, and identify synthetic audio using the inaudible watermark.

Text-to-Speech API for Transport Announcements

Frequently asked questions

What is the best voice for emergency alerts?

Q: Which voice is best for emergency alerts? A: Atlas (neutral, news) or Orion (deep, documentary) provide clear, authoritative tones suitable for alerts. You can adjust the speed between 0.25 and 4.0 to ensure urgency. Each voice is a preset based on licensed recordings, ensuring consistent quality for critical safety messages without requiring custom design.

Is Atlas better than Milo for news?

Atlas is described as neutral and news-oriented, while Milo is friendly and conversational. Atlas suits standard news delivery, whereas Milo works better for conversational updates. Both are available presets. You can test both with the same text to see which fits your specific transport communication style best.

Can I use Luna for station announcements?

Yes, Luna is a soft, storytelling voice preset. It works well for station announcements where a calming tone is preferred. You can generate audio via the API using the Luna voice ID. The output is a standard audio file that meets accessibility standards without complex infrastructure.

Does Orion work for subway systems?

Orion is a deep, documentary-style voice. Its clear, steady tone is well-suited for subway system announcements. You can send text up to 4096 characters per request. The API returns an audio file that you can deploy directly to your station playback systems.

Which voice is clearest in noisy areas?

Clarity depends on the voice and speed settings. Atlas (neutral) or Aria (warm, clear narrator) are good choices. You can increase the speed up to 4.0 to make speech punchier. Test the audio in your actual environment to ensure intelligibility over background noise.

Can I use Felix for promo messages?

Yes, Felix is an energetic, promo-oriented voice. It is ideal for promotional announcements or service updates. You can generate these audio files quickly using the API. The cost is $0.01 per 1,000 characters, making it affordable for frequent promotional content.

Is Sage too calm for urgent updates?

Sage is a calm, instructional voice. While calm, you can increase the speed parameter up to 4.0 to make it sound more urgent. This allows you to retain Sage’s clarity while adjusting the tempo to match the urgency of your transport updates.

Do voices sound different in Spanish?

The same 8 voice presets are used for all supported languages, including Spanish. The voice characteristics remain consistent, but the pronunciation adapts to the language. You must specify the language code in the request to ensure accurate delivery of the Spanish text.

Can I use Aria for customer service?

Yes, Aria is a warm, clear narrator voice. It is well-suited for customer service announcements or information updates. You can generate audio files via the API and deploy them to your systems. The tone is professional yet approachable for passenger interactions.

Is Nova too bright for announcements?

Nova is described as bright and upbeat. This may be too energetic for standard announcements but works well for positive updates or station events. You can adjust the speed or choose a calmer voice like Luna or Sage if Nova feels too lively.

Do I need to attribute the voices?

The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox. While the API delivers the audio, attribution requirements depend on your specific license compliance. The generated audio contains an inaudible watermark marking it as synthetic speech.

Can I use the same voice for all languages?

Yes, any of the 8 voice presets can speak any of the 23 supported languages. The voice characteristics remain consistent across languages. You simply specify the target language in the API request to ensure correct pronunciation and intonation for that language.

Is the watermark audible?

No, the watermark is inaudible. It is embedded in the generated audio file to mark it as synthetic speech. This allows for identification without affecting the listening experience for passengers or customers in your transport network.

How do I choose between tts-1 and tts-1-hd?

Both models use the same engine and price. The choice depends on your quality preference. Both are available for the same rate of $0.01 per 1,000 characters. Test both with your specific text to decide.

Neutral and professional

Sound wave illustration for: Neutral and professional

For transport environments where clarity and neutrality are paramount, specific voice presets ensure announcements are heard and understood without distraction. Atlas provides a neutral, news-style tone ideal for real-time updates, while Sage offers a calm, instructional quality perfect for guiding passengers through procedures.

Atlas: Ideal for news-style updates

Atlas delivers a neutral, authoritative delivery that mirrors standard transit information systems. Its tone is professional and direct, making it suitable for live departure boards, platform alerts, and general service announcements where emotional inflection is unnecessary.

  • Clear enunciation for busy station environments
  • Neutral tone avoids misinterpretation of urgency
  • Ideal for high-frequency, short-duration alerts

Sage: Best for instructional content

Sage is characterized by a calm, instructional quality that guides listeners through steps without sounding robotic. It works well for safety instructions, boarding procedures, and accessibility guidance where steady pacing helps passengers process information.

  • Steady pacing for complex instructions
  • Calm tone reduces passenger anxiety
  • Consistent delivery for repeated messages

Use cases for clear, direct messaging

These voices excel in scenarios requiring standardized communication across diverse passenger groups. By removing subjective tone, Atlas and Sage ensure that safety and operational messages are perceived uniformly, regardless of the listener's background or the ambient noise level.

Warm and engaging

Sound wave illustration for: Warm and engaging

Transport communication often requires a balance between authority and approachability. While Atlas provides the necessary neutrality for safety-critical updates, many passenger interactions benefit from a warmer tone that reduces stress and feels more human.

Aria: Clear narrator for general info

Aria offers a warm, clear narration style ideal for general station announcements. Her tone is professional yet inviting, making her suitable for route information, platform changes, and general guidance where clarity is paramount but a robotic feel should be avoided.

Nova: Upbeat for customer service

Nova brings a bright, upbeat energy perfect for customer-facing messages. She excels in welcoming passengers, providing friendly reminders about amenities, or announcing delays with a positive spin. Her voice helps maintain a pleasant atmosphere even during disruptions.

When warmth improves user experience

Warm voices can significantly reduce passenger anxiety during wait times or route changes. Using Aria or Nova for non-urgent updates creates a more engaging experience compared to purely neutral announcements.

Energetic and dynamic

Sound wave illustration for: Energetic and dynamic

Energetic voices bring vitality to transport announcements, keeping passengers engaged and informed. Felix and Milo add a dynamic layer to standard updates, making routine information feel more personal and less robotic.

These presets are ideal when you want to capture attention quickly without sounding overly formal. They work well for real-time updates that benefit from a touch of personality.

Felix: Perfect for promotions

Felix brings an energetic, promo-style tone that works well for special offers or service highlights. His voice cuts through background noise, making him ideal for advertising new routes or discounts.

Use Felix when you need to convey urgency or excitement. His clear, upbeat delivery ensures that promotional messages are heard and remembered by passengers waiting at stations.

  • Ideal for seasonal discounts
  • High energy for new route launches
  • Clear delivery in noisy environments

Milo: Conversational and friendly

Milo offers a friendly, conversational tone that feels like a helpful companion. He is perfect for general wayfinding, gate changes, or friendly reminders that enhance the passenger experience.

His approachable style reduces the stress of travel by making announcements feel less like commands and more like guidance. This is particularly effective for customer service updates.

  • Great for wayfinding
  • Friendly reminders
  • Customer service updates

Boosting engagement with energy

Using energetic voices can significantly improve passenger engagement. When announcements sound human and lively, passengers are more likely to pay attention to critical information.

Test these voices in your actual environment. The right energy level can make a difference in how well your announcements are received, especially during peak hours.

Soft and narrative

Sound wave illustration for: Soft and narrative

Luna and Orion bring a narrative quality to transport announcements, transforming routine updates into engaging experiences. Luna’s soft, storytelling tone works well for general information, while Orion’s deep, documentary style suits detailed route descriptions or safety briefings.

Luna: Soft storytelling voice

Luna offers a warm, clear narrator voice ideal for passenger information screens or app notifications. Her tone reduces the mechanical feel of automated messages, making waiting passengers feel more at ease during delays or service changes.

  • Best for: Station delays, general passenger info, app notifications

Orion: Deep, documentary style

Orion provides a serious, authoritative depth suitable for detailed explanations or safety instructions. His voice commands attention without sounding aggressive, making him perfect for emergency alerts or complex route overviews that require clear, calm delivery.

  • Best for: Safety briefings, complex route details, emergency alerts

Using tone for context

Match the voice to the urgency. Use Luna for routine updates to maintain a pleasant atmosphere. Switch to Orion for critical information where clarity and gravitas are needed. Both voices support all 23 languages, ensuring consistent tone across international routes.

Remember that these voices are synthetic and carry an inaudible watermark. This helps identify generated content if required by local transparency laws.

Language considerations

Sound wave illustration for: Language considerations

SpeakPath’s eight voices are not tied to specific regions. Each voice can pronounce all 23 supported languages, from Arabic to Turkish, with consistent tonal quality regardless of the target language.

All voices support 23 languages

The API supports 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 of these languages, ensuring uniform brand voice across your global network.

No language-specific voice limits

There are no restrictions assigning specific voices to specific languages. You can use Aria for English announcements or Luna for French storytelling without changing your voice roster. This flexibility simplifies management for transport operators with multi-lingual stations.

Testing voices in target languages

Because the system guesses language from script by default, set the language parameter for non-English Latin-script text to ensure accuracy. Test short phrases in each target language to verify clarity, especially for critical safety announcements or directional instructions in noisy environments.

Technical selection criteria

Sound wave illustration for: Technical selection criteria

Selecting the right technical parameters ensures announcements are understood in busy stations and moving vehicles. Our API provides standardized audio files that meet accessibility standards without requiring complex infrastructure.

By tuning speed and format, you guarantee clarity across different hardware environments. The following criteria help you optimize for both audio fidelity and playback reliability in transport contexts.

Clarity in noisy environments

For station platforms or busy terminals, voice choice matters more than volume. Atlas delivers neutral, news-style clarity ideal for direct messaging, while Sage offers calm instructional tones that cut through background noise effectively.

Use Sage for step-by-step boarding instructions or safety guidelines. Its steady pace ensures listeners can process information even in high-decibel areas where rapid speech might be missed.

  • Atlas for neutral, news-style updates
  • Sage for calm, instructional content
  • Clear articulation aids comprehension in high-noise zones

Speed range compatibility

The API supports speed settings from 0.25x to 4.0x, allowing you to adjust pacing for specific environments. Use slower speeds (0.75x–0.9x) for complex route changes or elderly passengers. Faster speeds (1.2x–1.5x) work well for brief, repetitive alerts.

Test your chosen speed with local hardware. While real-time generation handles long texts efficiently, short phrases may take a few seconds to synthesize, so plan buffer times accordingly.

  • Speed range: 0.25x to 4.0x
  • Slower speeds for complex instructions
  • Faster speeds for brief, repetitive alerts

Format support (MP3/WAV)

Choose MP3 for efficient storage and bandwidth usage, or WAV for uncompressed, high-quality playback. Both formats are widely supported by transport PA systems. MP3 reduces storage needs while maintaining clarity for most voice announcements.

WAV files ensure maximum fidelity for critical safety alerts where audio compression artifacts must be avoided. Select the format that best matches your existing hardware capabilities and storage constraints.

  • MP3: efficient storage and bandwidth
  • WAV: uncompressed, high-fidelity playback
  • Both formats widely supported by PA systems

Licensing and rights

Sound wave illustration for: Licensing and rights

SpeakPath’s voices are synthesized from licensed recordings in the LibriTTS-R dataset. This foundation ensures that our audio outputs are built upon high-quality, professionally recorded speech samples.

LibriTTS-R source data

The LibriTTS-R dataset contains hours of clean, read-aloud speech. By using this data, we ensure that the acoustic characteristics of our voices meet professional standards for clarity and consistency, which is critical for public address systems.

  • High-quality, noise-free source recordings
  • Standardized speech patterns for consistent output

CC BY 4.0 license details

All voice data is provided under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. This allows us to legally synthesize and distribute audio files generated by our API. You do not need to worry about additional licensing fees for the voice data itself when using SpeakPath.

  • CC BY 4.0 licensed data
  • No additional voice licensing fees

Public domain audiobook origins

Much of the source material originates from LibriVox, featuring public-domain audiobook readings. These recordings are from works where copyright has expired or been waived, ensuring that the underlying content is free from restrictive copyright claims. This background supports reliable, long-term use of your announcements.

  • Rooted in LibriVox public-domain readings
  • Minimizes copyright restrictions for commercial use
Licensing and rights

Watermarking and detection

Sound wave illustration for: Watermarking and detection

Every audio file generated by SpeakPath contains an inaudible watermark that marks the content as synthetic speech. This technical feature ensures that your transport announcements are distinctly identifiable as AI-generated, providing a reliable method for internal verification and quality control without affecting the listener's experience.

Inaudible synthetic watermark

The watermark is embedded directly into the audio waveform, making it imperceptible to human ears while remaining detectable by specialized algorithms. This allows you to verify the authenticity of your announcement files at any stage of your production workflow, ensuring that the final output matches the generated source.

Marking content as AI-generated

By embedding this marker, SpeakPath provides a clear signal that the voice is not a human recording. This is particularly useful for large-scale deployments where you need to distinguish between voice-clipped library assets and dynamically generated announcements, helping your team maintain accurate records of content origins.

Compliance with transparency laws

As regulations regarding synthetic media evolve, having a detectable watermark helps demonstrate transparency. While we do not offer legal advice, this built-in marking supports your efforts to meet emerging standards that require clear identification of AI-generated audio in public broadcasting and customer communication channels.

Speed and latency

Sound wave illustration for: Speed and latency

SpeakPath generates audio at roughly real-time speed for long texts, processing about 25 characters per second. This predictable rate helps transport teams estimate deployment timelines accurately without complex infrastructure overhead.

Real-time generation on long texts

Long announcements, such as station platform updates, are processed efficiently. The system handles up to 4096 characters per request; you must split longer texts and join the audio files. This ensures consistent pacing and synchronous audio output for continuous broadcasts.

Queueing under load

Requests are processed sequentially. Under high load, a request may wait in a queue before synthesis begins. This linear processing guarantees audio quality but means simultaneous requests might experience slight delays compared to parallel systems.

Planning for latency

Short phrases typically take a few seconds to generate. Plan your API integration to account for this initial latency, especially for real-time triggered events. For critical safety alerts, consider pre-generating common messages to ensure instant playback availability.

Common use cases

Sound wave illustration for: Common use cases

Transport announcements require clarity and consistency. SpeakPath provides standardized audio files that meet accessibility and safety standards without complex infrastructure. Product teams can deploy reliable, human-sounding updates for stations and vehicles with minimal engineering effort.

Station platform announcements

Use Atlas or Sage for clear, direct messaging. Send up to 4096 characters per request to generate concise updates about delays or gate changes. The API returns standard formats like MP3 or WAV, ready for integration with existing PA systems.

  • Clear instructions for boarding and alighting
  • Real-time delay notifications
  • Platform change alerts

In-vehicle route updates

Nova or Milo add a friendly, conversational tone for interior announcements. Generate audio for specific stops or route segments. Adjust speed between 0.25 and 4.0 to match vehicle acoustics and passenger density.

  • Stop-by-stop navigation cues
  • Transfer information
  • Next station previews

Emergency alert messages

Atlas provides a neutral, news-style voice ideal for urgent situations. Split longer emergency texts into parts of 4096 characters and join the audio files. Ensure high clarity even in noisy environments by selecting appropriate response formats.

  • Safety instructions
  • Evacuation guidance
  • Immediate hazard warnings

Integration tips

Sound wave illustration for: Integration tips

Integrating SpeakPath into your transport announcement pipeline is straightforward because the API mimics the standard OpenAI speech endpoint. You can use the official OpenAI SDKs by simply pointing the base URL to https://api.55gg.info/v1 and supplying your Bearer token. This allows you to leverage existing client libraries for Python, Node.js, or Go without writing custom HTTP parsers, ensuring a smooth deployment for your product team.

Using OpenAI SDK

The API uses the same request shape as OpenAI’s speech endpoint. Change your SDK’s base_url to https://api.55gg.info/v1 and pass your tts-... key in the Authorization header. This compatibility means you can use the official OpenAI SDKs directly, reducing development time and maintenance overhead for your engineering team.

  • Use official OpenAI SDKs with modified base_url
  • Standard JSON body with model, input, voice, response_format
  • Supports Python, Node.js, Go, and other SDK languages

Handling 4096 char limits

Each request accepts up to 4096 characters. For longer announcements, such as complex station schedules, split your text into chunks of this size, generate each audio file separately, and join them programmatically. This ensures consistent audio quality and prevents truncation or errors during synthesis.

  • Split text into 4096-character segments
  • Generate audio files for each segment
  • Concatenate audio files in your preferred format (mp3, wav, opus, flac, pcm)

Error handling strategies

Failed requests are free, so you can retry them without worrying about credit consumption. Implement exponential backoff for transient errors and log request metadata for billing audits. Since requests are processed sequentially, monitor your queue status if you expect high volume, and consider staggering requests to avoid latency spikes.

  • Retries are cost-free on failure
  • Log metadata for billing and debugging
  • Stagger requests to manage queue latency

Final recommendations

Sound wave illustration for: Final recommendations

Choosing a single voice ensures your passengers recognize your brand instantly, whether they are listening to platform updates or in-vehicle route information. Consistency reduces cognitive load and builds trust over time.

Stick to one voice for consistency

Using the same voice across all announcements, from emergency alerts to routine schedule changes, creates a familiar auditory identity. This uniformity helps passengers quickly identify official communications amidst background noise.

  • Maintain brand recognition across all touchpoints
  • Reduce passenger confusion during critical updates

Test before full deployment

Generate sample announcements for your specific routes and stations before going live. Verify that the audio quality meets accessibility standards and that the text-to-speech engine handles your local terminology correctly.

  • Validate clarity in noisy environments
  • Check pronunciation of specific station names

Monitor user feedback

Pay attention to passenger comments regarding audio clarity and volume. If users report difficulty understanding certain announcements, consider adjusting the speed parameter or switching to a clearer voice preset like Atlas or Sage.

  • Track common comprehension issues
  • Adjust speed settings based on real-world usage

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. Pay attention to passenger comments regarding audio clarity and volume.

Avatar of the voice aria
ariafemale
Avatar of the voice nova
novafemale
Avatar of the voice luna
lunafemale
Avatar of the voice sage
sagefemale
Avatar of the voice felix
felixmale
Avatar of the voice milo
milomale
Avatar of the voice atlas
atlasmale
Avatar of the voice orion
orionmale

23 languages

Any voice speaks any of these languages.

Sources and further reading

  1. Wikipedia
  2. LibriTTS-R paper (arXiv 2305.18802)
  3. LibriVox public-domain audiobooks
  4. openai-python on GitHub
  5. Node.js documentation
  6. MDN: Web Speech API
  7. Tether (USDT)
  8. Creative Commons Attribution 4.0
  9. curl documentation
  10. W3C WAI: making audio and video accessible
  11. openai-node on GitHub
  12. IETF RFC 7845: Ogg Opus

More on this site