Editorial Team, Voice Licensing, and Data Policy Overview

By TransitVoice editorial team · Updated 2026-11-19

SpeakPath is an independent storefront for a text-to-speech API dedicated to the transport industry. We specialize in delivering standardized, human-sounding audio announcements for stations and vehicles, helping product teams meet accessibility and safety standards without complex infrastructure. This page explains who we are, where our voices come from, and how we handle your data.

Unlike larger tech platforms, we are not affiliated with OpenAI, ElevenLabs, or any other brand. We focus on specific technical limits, such as our 60 requests per minute cap and 4096-character input constraint, to ensure consistent quality. Read on to understand our voice licensing, data policy, and support model so you can deploy with confidence.

Editorial Team, Voice Licensing, and Data Policy Overview

Frequently asked questions

Who runs this site?

This site is an independent storefront for the SpeakPath text-to-speech API. We help product teams in the transport industry deploy reliable, human-sounding announcements. Our API delivers standardized audio files that meet accessibility and safety standards without complex infrastructure.

Are you affiliated with OpenAI?

No. We are not affiliated with OpenAI, ElevenLabs, Google, Amazon, Microsoft, or any other brand mentioned in comparisons. Our API uses the same request shape as the OpenAI speech endpoint, but we are a separate entity.

Where do the voices come from?

Our eight voice presets are based on licensed recordings from LibriTTS-R (CC BY 4.0) and public-domain audiobook readings from LibriVox. Each generated file carries an inaudible watermark that marks it as synthetic speech.

Is the license CC BY 4.0?

The source recordings are licensed under CC BY 4.0. However, we do not offer custom voice cloning or uploading your own voice. We provide eight specific presets for use in your applications.

Do you store my audio files?

Generated audio is returned in the response and not stored. Input text is used only to produce the audio and is not used for training. We keep request metadata for billing purposes.

Do you use my text for training?

No. Input text is used only to produce the audio and is not used for training. Generated audio is returned in the response and not stored. Request metadata is kept for billing.

How do I contact support?

Support is by email only. We do not offer a phone number or live chat. You can reach us via the contact page or email link provided on our site.

Is there a phone number?

No. We do not have a phone number. Support is provided via email only. There is no live chat option available for customer inquiries.

Do you offer an SLA?

No. We do not offer an SLA, uptime guarantee, or enterprise contract. There is no on-premise install option or data-center choice available for our service.

Do you have certifications?

No. We do not have certifications such as SOC 2, HIPAA, or GDPR certificates. We are an independent storefront without these specific compliance documents.

Is GDPR certified?

No. We do not have a GDPR certificate. While we follow data practices like not storing audio, we do not hold formal certification for this regulation.

Can I request data deletion?

Request metadata (time, characters, status) is kept for billing. Input text is used only to produce the audio and is not stored. You can contact us via email to discuss your data retention preferences.

Do you sell user data?

We do not sell user data. Input text is used only to produce the audio and is not used for training. Generated audio is returned in the response and not stored.

What is your return policy?

Failed requests are free: if synthesis fails, the charge is returned in full. There is no monthly plan or subscription.

Who we are

Sound wave illustration for: Who we are

SpeakPath is an independent storefront for a text-to-speech API for the transport sector. We focus on delivering reliable, human-sounding announcements for stations and vehicles, ensuring product teams can meet accessibility and safety standards without managing complex infrastructure.

We are not affiliated with OpenAI, ElevenLabs, or any other major brand. Our API follows a standardized request shape compatible with OpenAI SDKs, but we operate as a distinct entity with our own infrastructure, pricing, and voice models designed specifically for clear, clear-cut public address systems.

Independent editorial team

Independent editorial team

We built SpeakPath to solve the specific challenge of generating clear, consistent audio for transport announcements without the overhead of enterprise-grade platforms.

Focused on transport industry

Focused on transport industry

Our voices are tuned for clarity in noisy environments like train platforms and bus interiors. We prioritize natural pacing and clear enunciation over dramatic expression, ensuring passengers understand critical safety information instantly.

No affiliation with OpenAI or others

No affiliation with OpenAI or others

While our API endpoint structure mirrors OpenAI’s for easy integration, we are completely separate. We do not use their models, servers, or branding. SpeakPath runs on its own infrastructure with its own voice engine and pricing model.

Voice licensing

Sound wave illustration for: Voice licensing

SpeakPath’s voice presets are built on licensed recordings from the LibriTTS-R dataset. These voices include Aria, Nova, Luna, and Sage on the female side, and Felix, Milo, Atlas, and Orion on the male side. Each voice provides a distinct tone suitable for transit announcements, from warm narrations to clear, instructional guidance.

By leveraging established datasets, we ensure consistent quality across all 23 supported languages, allowing your product team to deploy standardized audio files that meet accessibility standards efficiently.

Voices from LibriTTS-R

Voices from LibriTTS-R

All eight voice presets are derived from the LibriTTS-R dataset. This foundation ensures that every generated file maintains a natural, human-like quality essential for passenger announcements. Whether you need a calm voice for station updates or an energetic tone for promotions, the dataset provides the necessary range.

  • 8 distinct voice presets available
  • Covers warm, clear, bright, and deep tones
  • Consistent quality across 23 languages

CC BY 4.0 license

The underlying voice data is licensed under CC BY 4.0. This allows us to distribute the audio files generated by our API while respecting the original creators' rights. You can use the generated audio for your transport announcements without worrying about complex licensing fees for the voice model itself.

Source: LibriVox public domain

The recordings originate from LibriVox, a collection of public-domain audiobook readings. These high-quality, licensed recordings form the basis of our synthetic speech engine. By using these established sources, we provide a stable and legally sound foundation for your text-to-speech needs in the transport industry.

  • Rooted in public-domain audiobook readings
  • High-quality original recordings
  • Transparent data origins

Data policy

Sound wave illustration for: Data policy

SpeakPath is an independent storefront for a text-to-speech API focused on the transport industry. We are not affiliated with OpenAI, Google, or any other major platform.

Input text not used for training

When you send text to our API, it is processed solely to generate the audio file you receive. We do not use your announcement scripts, station names, or route data to train our speech models. Your content remains yours, ensuring that proprietary transit information stays private and is not repurposed for our general voice datasets.

Audio not stored after response

The generated audio files are returned directly in the HTTP response and are not permanently stored on our servers. Once the audio is delivered to your system, we do not keep a copy of the file. This minimizes storage costs and ensures that your audio assets remain under your direct control, reducing data retention risks.

Metadata kept for billing

We retain request metadata, including the timestamp, character count, and status code, strictly for billing and usage tracking purposes. This data helps us accurately charge you based on the $0.01 per 1,000 characters rate and troubleshoot any issues. We do not store the actual audio content or the input text alongside this metadata.

Technical limits

Sound wave illustration for: Technical limits

SpeakPath is a practical tool for transport announcements, but it runs on shared cloud infrastructure without enterprise-grade guarantees. You should plan your integration knowing that we do not offer an SLA or uptime guarantee. If the service experiences delays, you may need to implement retry logic in your own application layer.

No SLA or uptime guarantee

We do not promise specific availability percentages or financial credits for downtime. This keeps our costs low and the API simple for product teams who just need reliable audio files without complex infrastructure contracts.

60 requests per minute limit

Each API key is limited to 60 requests per minute. Before your first top-up, this limit is lower at 10 requests per minute. If you exceed this limit, requests will queue or return an error. Plan your batch generation accordingly.

4096 character input limit

Each API request can process up to 4,096 characters of text. For longer announcements, you must split the text into chunks and join the resulting audio files. This ensures consistent generation speed and prevents timeouts during synthesis.

Technical limits

Support contact

Sound wave illustration for: Support contact

We keep our support simple and focused on what matters for your transport announcements. All communication happens through email, ensuring a clear written record of your queries and our responses.

Email support only

Send your questions to our support address. We prefer email because it allows us to attach relevant documentation, code snippets, or API references directly to our replies. This ensures you have a complete record of the solution for future reference.

No phone or live chat

We do not offer phone lines or live chat widgets. This eliminates the noise of constant interruptions and lets our engineers focus on resolving your technical issues thoroughly.

Response times vary

Response speed depends on the complexity of your request and our current volume. Simple questions about voices or pricing are usually answered quickly, while issues involving API integration or specific transport announcement workflows may take longer. We appreciate your patience as we ensure accuracy.

Honest limitations

Sound wave illustration for: Honest limitations

We offer eight consistent, licensed voices but do not allow you to upload your own recordings for custom voice cloning. Each generated file carries an inaudible synthetic watermark to mark it as AI-generated speech.

No custom voice cloning

Our voices are based on licensed LibriTTS-R recordings. While we provide eight distinct presets suitable for transport announcements, we do not currently support uploading a unique voice sample to create a custom clone.

No SSML support

The API accepts plain text input. We do not support SSML tags for controlling pronunciation, pauses, or emphasis. You can control the overall speaking rate using the speed parameter (0.25–4.0), but fine-grained audio structure must be handled by splitting your text into separate requests.

No real-time streaming

We deliver standardized audio files (mp3, wav, opus, flac, pcm) after full synthesis. There is no WebSocket or streaming endpoint for partial audio. You must wait for the complete file to be generated and returned, which typically takes roughly 25 characters per second.

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. Tether (USDT)
  3. openai-node on GitHub
  4. OpenAI API reference: create speech
  5. IETF RFC 7845: Ogg Opus
  6. TRON developer hub: TRC-20
  7. LibriVox public-domain audiobooks
  8. openai-python on GitHub
  9. LibriTTS-R corpus (OpenSLR 141)
  10. LibriTTS-R paper (arXiv 2305.18802)
  11. MDN: Web Speech API
  12. Creative Commons Attribution 4.0

More on this site