How to Create Audioguide Audio with TTS API?

By MuseumEcho editorial team · Updated 2026-11-19

This guide explains how to generate professional audioguide audio for museums, tours, and accessible travel content using the MuseumEcho TTS API. We cover practical steps from selecting voices to handling multi-language support, ensuring your startup delivers consistent, high-quality narration without expensive studio overhead.

Readers will learn to prepare text under 4096 characters, split long descriptions, and manage audio generation efficiently. You will discover how to integrate the API with standard HTTP requests, handle rate limits, and test your audio guides using trial credit. This page is designed for developers building scalable audio experiences who need clear, step-by-step instructions.

How to Create Audioguide Audio with TTS API?

Frequently asked questions

What is TTS?

Text-to-speech (TTS) is a cloud API that converts written text into audio files. You send a POST request with your script and receive an audio response. This allows startups to generate audioguide narration programmatically without hiring voice actors for every update.

How does TTS work?

Send a JSON payload to the /v1/audio/speech endpoint with your text, chosen voice, and language. The API processes the request and returns an audio file in formats like MP3 or WAV. Failed requests are free, ensuring you only pay for successful generations.

Is TTS better than human recording?

TTS offers consistency and instant updates without scheduling studio time. While human recordings may offer more nuance, TTS allows you to change narration instantly. Voices are based on licensed recordings, providing a reliable baseline for museum tours without complex production workflows.

Can TTS sound natural?

Our eight presets, like Aria or Felix, are based on licensed recordings from LibriVox. They offer clear, distinct styles suitable for audioguides. Every file includes an inaudible watermark marking it as synthetic speech, ensuring transparency about the audio source for your users.

How many languages are supported?

We support 23 languages, including English, Spanish, German, French, Japanese, and Chinese. Any voice can speak any supported language. If you do not set the language, it guesses from the script, defaulting to English for Latin text.

Can I change the voice later?

Yes, you can switch voices by updating the 'voice' parameter in your next API request. You can also rotate keys or adjust settings like speed. There are no long-term contracts, allowing you to experiment with different narrators for different museum exhibits.

Is the audio copyrighted?

The voices are based on LibriTTS-R (CC BY 4.0) and LibriVox public-domain readings. The generated audio files are returned in the response and not stored by us. You receive the file directly, allowing you to integrate it into your tour app immediately.

Does TTS support emojis?

Emojis are treated as characters within the text input. Since the API processes text up to 4096 characters, emojis count toward this limit. They will be rendered as part of the speech synthesis, though specific pronunciation may vary by voice.

Can I adjust the speed?

Yes, you can set the 'speed' parameter between 0.25 and 4.0. This allows you to slow down narration for accessibility or speed up quick updates. The default speed provides a natural listening pace suitable for most museum audioguide contexts.

What is the max length?

Each request supports up to 4096 characters. For longer texts, split your content into parts and join the resulting audio files. This ensures consistent quality and prevents timeout issues during generation, keeping your audioguide audio crisp and clear.

Do I need internet?

Yes, since this is a cloud API, you need an internet connection to send requests and receive audio files. The API processes requests sequentially, so generation time depends on queue load. You must have connectivity to access the generated audio.

Is it secure?

Input text is used only to produce audio and is not used for training. We keep request metadata for billing but do not store the generated audio. Authentication uses a Bearer token, and you can issue a new key instantly if needed.

Can I use it offline?

No, this is a cloud-based API. You must send text to our servers to receive the audio file. There is no offline mode or local SDK. Your users will need internet access to stream or download the generated audio files.

How to train TTS?

You do not need to train the model. The voices are pre-set presets based on licensed recordings. You simply select the voice that best fits your museum tour style. There is no custom voice cloning or training process required for usage.

Why Use TTS for Audioguides?

Sound wave illustration for: Why Use TTS for Audioguides?

Text-to-speech API lets startups produce audioguide content without hiring voice actors for every language. You send text and receive audio files directly, keeping production costs predictable at $0.01 per 1,000 characters. This model eliminates upfront studio fees and reduces overhead for small teams.

Updates are instant: change the text in your database, regenerate the audio, and deploy. No re-recording sessions are needed when exhibit details shift or new artifacts are added. The API handles the synthesis, so your team focuses on curating accurate tour content rather than managing audio logistics.

Cost-effective production

Pricing is simple: $10 per million characters. Failed requests are free, so you only pay for successful generations. With a free trial of $0.50, you can test your first few tours before committing to top-ups via USDT on the TRON network.

Easy updates to content

Exhibitions change frequently. If a plaque description is updated, you simply send the new text to the API and replace the audio file. This ensures your audioguide always reflects the current state of the museum without waiting for studio availability.

Consistent voice quality

Every file uses one of eight licensed voices, ensuring uniform tone across your entire library. Whether narrating a history tour or providing quick facts, the voice remains recognizable and professional. Audio quality is consistent, with synthetic watermarks marking files as generated.

Choosing the Right Voice

Sound wave illustration for: Choosing the Right Voice

MuseumEcho provides eight distinct voice presets, each optimized for specific audioguide contexts. Selecting the correct voice ensures your audio content matches the intended atmosphere, whether you are delivering historical facts or immersive storytelling.

Aria for narration

Aria offers a warm, clear narrator tone suitable for general tours. Her delivery is consistent and professional, making her ideal for standard museum exhibits where clarity is paramount. She handles descriptive text efficiently, maintaining a natural pace that keeps visitors engaged without sounding robotic.

  • Warm and clear tone
  • Ideal for standard exhibits
  • Consistent delivery

Atlas for facts

Atlas provides a neutral, news-style voice perfect for delivering historical data, dates, and technical specifications. His straightforward delivery ensures that complex information is conveyed with precision. Use Atlas when your audioguide focuses on factual accuracy and requires a neutral, authoritative presence.

  • Neutral, news-style delivery
  • High precision for data
  • Authoritative tone

Orion for stories

Orion brings a deep, documentary quality to your content, enhancing narrative-driven tours. His voice adds gravity and immersion to historical stories or legendary accounts. Use Orion when you want to evoke emotion and draw listeners deeper into the historical context of your site.

  • Deep, documentary quality
  • Enhances narrative immersion
  • Evokes historical gravity

Preparing Your Text

Sound wave illustration for: Preparing Your Text

Before sending text to the API, prepare your content to fit technical constraints. Each request accepts a maximum of 4,096 characters, including spaces and punctuation. This limit ensures stable generation and predictable billing.

Keep under 4096 chars

Check your text length before every API call. If your description exceeds the limit, the request will fail. Since costs are calculated per character, precise input helps you estimate expenses accurately without overpaying for unused capacity.

Split long descriptions

For exhibits with extensive histories, divide the text into logical segments of up to 4,096 characters. Generate audio for each segment individually, then join the resulting audio files on your server. This approach prevents timeout errors and keeps processing times manageable.

Format for readability

Use clear sentences and avoid complex punctuation. The TTS engine reads text linearly, so straightforward phrasing improves clarity. Remove unnecessary symbols or markdown that the speech engine cannot interpret, ensuring a smooth listening experience for museum visitors.

Generating Audio

Sound wave illustration for: Generating Audio

Generate audioguide audio by sending a POST request to the API endpoint with your text, voice selection, and format preferences. The system processes the input and returns the synthesized audio file directly in the response body, allowing you to integrate audio into your app or website without managing separate storage servers for each generated clip.

Send POST request

Use the OpenAI SDK or plain HTTP to send a JSON payload to POST https://api.siap234.cc/v1/audio/speech. Include your Authorization: Bearer key, the model (tts-1 or tts-1-hd), the input text (max 4096 characters), and the desired voice preset. Specify the response_format (mp3, wav, opus, flac, or pcm) and optional speed (0.25–4.0) to match your audioguide's pacing.

  • Base URL: https://api.siap234.cc/v1
  • Method: POST
  • Auth: Bearer token
  • Max input: 4096 characters

Receive audio file

Upon successful processing, the API returns the audio file directly in the HTTP response body. The format matches your response_format parameter, ensuring compatibility with standard audio players in mobile apps or web interfaces. You can save this file locally or play it directly in your application.

Handle response

Check the HTTP status code to confirm success. Failed requests incur no charge, so you can safely retry if needed. For long texts exceeding 4096 characters, split the input into chunks, generate each audio file separately, and concatenate them in your application logic before playback.

Handling Long Texts

Sound wave illustration for: Handling Long Texts

Long exhibit descriptions often exceed the 4096-character limit per request. To handle this, split your text into logical segments, such as individual exhibit entries or short narrative blocks, ensuring each part stays within the character boundary.

Split into parts

Break your script into chunks of up to 4096 characters. Process each segment individually via separate API calls. This ensures no data loss and keeps request sizes manageable for the synthesis engine.

  • Maximum 4096 characters per request
  • Split at natural sentence or paragraph boundaries
  • Keep context consistent between segments

Join audio files

Once you receive the audio files for each segment, concatenate them into a single track. Use a standard audio library to merge the files seamlessly. This creates a continuous listening experience for the user without noticeable gaps between exhibits.

  • Merge outputs in chronological order
  • Use consistent formats (e.g., mp3 or wav)
  • Ensure smooth transitions between segments

Manage timing

Generation takes roughly real-time duration, plus a few seconds for short phrases. Under load, requests may queue, so plan for slight delays. For long audioguides, generate segments asynchronously to avoid blocking the user interface during content loading.

  • Approx. 25 characters generated per second
  • Requests process sequentially under load
  • Asynchronous generation improves UX

Multi-Language Support

Sound wave illustration for: Multi-Language Support

Audioguide travelers come from all over the world, so your content must match their native language. The API supports 23 languages, but you must explicitly set the language field for non-English Latin text to ensure correct pronunciation. Without this code, the system guesses the language, which often defaults to English and causes errors.

This explicit control prevents mispronunciations in mixed-script texts and ensures consistent quality across your museum or tour app. It allows you to offer localized experiences without hiring separate voice actors for each region.

Set language code

Always include the ISO 639-1 language code in your request body. This overrides auto-detection and guarantees the voice speaks the intended language correctly. For example, set language: "es" for Spanish or language: "ja" for Japanese.

This is critical for Latin-script languages other than English, such as German or Dutch, where the default guess might fail. It ensures the synthetic voice applies the correct phonetic rules for that specific language.

  • Use ISO 639-1 codes (e.g., 'fr', 'de', 'ko').
  • Set it for all non-English Latin text to avoid default English pronunciation.
  • The same 8 voices can speak any of the 23 supported languages.

Auto-detection limits

If you omit the language field, the API guesses the language based on the script. Latin text defaults to English. If you send French text without setting the language, it will likely be read in English.

This limitation means you must be explicit for any language not English. Relying on auto-detection is risky for mixed-language content or less common Latin scripts, leading to awkward or incorrect audio output.

  • Latin text defaults to English if unspecified.
  • Detection is script-based, not context-aware.
  • Non-English Latin text requires explicit language codes.

23 languages available

We support 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 of the 8 voice presets can speak any of these languages.

This broad coverage allows you to create unified audioguide apps for international tourist destinations without managing multiple voice providers. Note that Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, and Bengali are not supported.

  • Supports 23 major global languages.
  • All 8 voices (Aria, Nova, Luna, Sage, Felix, Milo, Atlas, Orion) support all 23 languages.
  • Not supported: Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, Bengali.

Testing in Production

Sound wave illustration for: Testing in Production

Testing ensures your audioguide workflow works before releasing content to visitors. Use your trial credit to generate sample audio files and verify that the synthesis matches your expectations for tone and clarity.

Use trial credit

Sign in to get $0.50 in trial credit, valid for 7 days. Generate approximately 50,000 characters of audio to test different voices and languages without spending real funds.

  • Create a Google account to receive the trial credit immediately
  • Test multiple voices like Aria for narration or Atlas for facts
  • Verify audio formats like MP3 or WAV are compatible with your app

Check audio quality

Listen to the generated files to ensure clarity and correct pacing. The API produces roughly real-time audio, so long texts take time to generate. Check that the speed setting (0.25–4.0) suits your audience.

  • Adjust speed settings to match natural speech patterns
  • Verify language detection works correctly for non-English texts
  • Check that audio files are clear and free of artifacts

Monitor usage

Track your character count and request status in the dashboard. Failed requests are free, so monitor failures to identify text formatting issues. Keep an eye on your rate limit of 60 requests per minute.

  • Watch your balance and character usage in the dashboard
  • Log failed requests to improve text input quality
  • Stay under the 60 requests per minute limit
Testing in Production

Deployment Tips

Sound wave illustration for: Deployment Tips

Efficient deployment requires treating audio as a static asset rather than a live stream. Since generation takes time and costs credit, caching results for unchanged exhibits reduces latency and spend.

Cache generated audio

Store generated MP3 or WAV files in your CDN or storage bucket. Map audio URLs to exhibit IDs in your database so you do not regenerate the same content repeatedly.

  • Use the same input text as the cache key.
  • Invalidate cache only when exhibit descriptions change.

Handle rate limits

Your key allows 60 requests per minute. For large tours with many exhibits, pre-generate audio files during off-peak hours or use a queue to smooth traffic bursts.

  • Monitor usage via the dashboard.
  • Retry with exponential backoff if you hit the limit.

Error logging

Log request timestamps, character counts, and status codes. Since failed requests are free, distinguish between network errors and synthesis failures to optimize your generation pipeline.

Accessibility Features

Sound wave illustration for: Accessibility Features

Our TTS API ensures accessible content for diverse users through precise control over audio output. By leveraging standard voice presets and configurable parameters, you can create audio guides that are clear, readable, and adaptable to various listening environments, from quiet museum halls to noisy bus tours.

Accessibility is built into the request structure, allowing developers to adjust speed and select the most suitable audio format for their application’s constraints without needing external post-processing tools.

Clear pronunciation

All eight voice presets are designed for natural, human-like intonation, ensuring that complex historical facts or directions are delivered with clarity. The engine handles standard punctuation and sentence structures effectively, reducing the need for manual phoneme editing in typical museum scripts.

Since voices are based on licensed recordings from LibriVox and LibriTTS-R, the underlying speech patterns are proven to be clear and easy to understand for general audiences.

  • Natural intonation in all 8 voice presets
  • No need for custom phoneme dictionaries
  • Consistent clarity across 23 supported languages

Adjustable speed

You can adjust the speech rate between 0.25x and 4.0x using the speed parameter. This allows you to slow down audio for elderly visitors or speed up content for quick transit announcements without re-generating the audio file.

A standard speed is approximately 25 characters per second, which is considered natural for most listeners. You can fine-tune this to match the specific pacing requirements of your audioguide app.

  • Speed range: 0.25x to 4.0x
  • Adjustable per request
  • Ideal for varied listening contexts

Multiple formats

The API supports five output formats: mp3, wav, opus, flac, and pcm. This flexibility ensures compatibility with various web players, mobile apps, and embedded audio systems used in tourism devices.

Choose mp3 or opus for efficient storage, or wav and flac for high-fidelity playback in premium museum installations. pcm is available for raw audio processing needs.

  • Supported formats: mp3, wav, opus, flac, pcm
  • Optimized for web and mobile players
  • High-fidelity options available

Common Use Cases

Sound wave illustration for: Common Use Cases

Text-to-speech enables consistent, scalable audio content for tourism and transit environments. By generating audio programmatically, teams can update guide content instantly without re-recording studio sessions, ensuring information stays current.

Museum tours

Use warm voices like Aria or Luna to narrate exhibit histories. Generate individual audio files for each artifact, keeping texts under 4,096 characters. Cache these files to reduce API costs during high visitor seasons.

  • Assign specific voices to different historical eras.
  • Split long exhibit descriptions into logical segments.
  • Use MP3 or OPUS for efficient storage.

Bus audio guides

Create clear, neutral narration for stops and landmarks. Atlas provides a professional tone suitable for factual stop announcements. Generate audio for each route segment, allowing easy updates when schedules or routes change.

  • Use concise texts to minimize generation time.
  • Maintain consistent voice across all bus routes.
  • Test audio clarity in noisy environments.

Airport announcements

Use clear, standard voices for flight updates and gate changes. Generate audio for each announcement type, ensuring language codes are set explicitly for non-English destinations. Cache common phrases to handle peak traffic efficiently.

  • Set language codes explicitly for multi-lingual airports.
  • Use neutral voices like Atlas for clarity.
  • Monitor character usage for high-volume announcements.

Troubleshooting

Sound wave illustration for: Troubleshooting

Even with careful preparation, you may encounter common issues when generating audioguide audio. This section addresses typical problems like inconsistent audio quality, unexpected language detection, and hitting rate limits, so you can resolve them quickly and keep your tours running smoothly.

Audio quality issues

Check your text formatting and ensure special characters are properly encoded. If audio sounds rushed, reduce the speed parameter to a value between 0.25 and 4.0. Verify that the selected voice matches the content tone, such as using Atlas for factual content.

  • Verify speed settings (0.25–4.0)

Language detection errors

The API guesses language from script, defaulting to English for Latin text. For non-English Latin-script content, explicitly set the language parameter. Ensure you are using one of the 23 supported languages; Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, and Bengali are not supported.

  • Set explicit language code for Latin text

Rate limit exceeded

You are limited to 60 requests per minute. If you exceed this, wait until the window resets. Monitor your usage dashboard to stay within limits. Failed requests do not count against your character quota.

  • Limit: 60 requests/minute

Next Steps

Sound wave illustration for: Next Steps

Once your audioguide prototype is ready, integrate the API into your application. Use the provided curl example or OpenAI SDK with the base URL https://api.siap234.cc/v1. Ensure your authentication header includes your Bearer key.

Before launching, test the audio with real users. Verify clarity, pacing, and language accuracy. Use your $0.50 trial credit to generate initial samples without cost. Monitor usage metrics to ensure stability.

Integrate API

Send POST requests to /v1/audio/speech with your text and selected voice. Receive MP3 or WAV files directly in the response. Store these audio assets locally or in your CDN to reduce repeated API calls for static content.

  • Use the OpenAI SDK with base_url https://api.siap234.cc/v1
  • Include Authorization: Bearer tts-<your-key> header
  • Set response_format to mp3 for smaller file sizes

Test with users

Play generated audio on target devices to check volume and clarity. Verify that language detection works correctly for non-English Latin scripts. Adjust speed settings between 0.25 and 4.0 for optimal listening experience.

  • Test on mobile devices and speakers
  • Check audio quality at different speed settings
  • Verify language accuracy for mixed-script texts

Scale usage

Monitor your character count and rate limits (60 requests/minute). Top up with USDT on TRON network for higher volumes. Cache generated audio to minimize API calls and reduce costs.

  • Track character usage against your balance
  • Top up with minimum $5 USDT
  • Cache audio files to save costs

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. For archival tours or high-fidelity experiences, WAV and FLAC provide lossless audio.

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. curl documentation
  3. IETF RFC 7845: Ogg Opus
  4. ISO 639 language codes
  5. Python documentation
  6. LibriTTS-R paper (arXiv 2305.18802)
  7. TRON developer hub: TRC-20
  8. Tether (USDT)
  9. openai-python on GitHub
  10. openai-node on GitHub
  11. LibriTTS-R corpus (OpenSLR 141)
  12. MDN: Web Speech API

More on this site