Text-to-Speech API for Museum Audioguide Narration

By ChickenSubway editorial team · Updated 2026-11-18

This page guides indie developers building museum audioguides and tourism experiences through voice selection, technical setup, and licensing. You will learn how to match specific voice personalities—like Aria’s warmth or Atlas’s neutrality—to historical content, ensuring clear narration in noisy gallery environments. We cover practical steps for integrating VoxMuse’s TTS API, from choosing formats to managing speed and character limits.

We address the specific needs of tourism apps, including multi-language support across 23 languages and rights management for publicly sourced voices. Discover how to maintain listener retention, optimize audio quality with synthetic watermarks, and scale your project without complex subscriptions. Whether you are crafting a children’s tour or a serious historical documentary, this guide provides the concrete details you need to implement reliable, character-driven audio narration.

Text-to-Speech API for Museum Audioguide Narration

Frequently asked questions

Which voice is best for history?

Atlas offers a neutral, news-style delivery suitable for historical facts. Orion provides a deep, documentary tone ideal for narrative history. Both voices are clear and professional. You can test each voice with your specific script to determine which best matches the historical context and audience expectations for your audioguide.

Can I use Luna for kids?

Luna has a soft, storytelling voice that works well for children. It is not specifically designed for kids but is gentle and engaging. Ensure the language is set correctly, as Luna supports multiple languages. The voice is clear and easy to understand, making it a practical choice for young audiences.

Is Felix too energetic?

Felix is described as energetic and promo-focused. If you find the default energy too high, you can reduce the speed to 1.0 or lower. This allows you to adjust the pace to suit a museum setting. Test Felix with your specific text to see if the energy level fits your needs.

Do voices sound robotic?

Voices are based on licensed recordings from LibriTTS-R and LibriVox. Each generated file includes an inaudible watermark marking it as synthetic. While not perfect, the voices are designed for clear, realistic narration suitable for audioguides without sounding overly mechanical or artificial.

Can I mix multiple voices?

The API processes one voice per request. To mix voices, generate separate audio files for each voice and join them. You can split long texts into parts, generate each with the desired voice, and combine the resulting audio files in your post-production workflow for a multi-voice audioguide experience.

Which voice speaks Spanish best?

All eight voices support Spanish. Aria, Nova, Luna, and Sage are female; Felix, Milo, Atlas, and Orion are male. Any voice can speak any supported language. Choose the voice that best matches the tone of your museum exhibit, as all support Spanish with equal capability and quality.

Is Orion good for documentaries?

Orion is specifically described as deep and documentary-style. This makes it ideal for serious, historical, or educational content. Its tone is authoritative and engaging, suitable for narrating detailed exhibits. Use Orion when you want a voice that conveys depth and seriousness appropriate for documentary-style museum audio.

Does Nova work for promos?

Nova is described as bright and upbeat, making her suitable for promotional content. Felix is also energetic and promo-focused. Use Nova for lighter, positive announcements or marketing audio within your museum. Adjust the speed if needed to match the desired energy level for your specific promotional material.

Can I adjust speed per voice?

Yes, you can adjust the speed for any voice. The speed parameter accepts values from 0.25 to 4.0. This allows you to slow down complex historical explanations or speed up simple signs. Adjust the speed independently for each request to optimize clarity and pacing for your specific audioguide content.

Are voices gender-specific?

Voices are categorized as female (Aria, Nova, Luna, Sage) or male (Felix, Milo, Atlas, Orion). However, any voice can speak any supported language. The gender distinction is primarily tonal. Choose based on the desired tone rather than strict gender requirements, as all voices are capable of clear narration.

Which voice is calmest?

Sage is described as calm and instructional, making it the calmest option. Luna is soft and storytelling-focused, which is also gentle. Use Sage for clear, relaxed instructions or historical narratives. These voices provide a soothing tone suitable for museum environments where a calm atmosphere is preferred by visitors.

Do voices support accents?

Voices do not have built-in accent options. They speak in a standard accent for the selected language. You can influence tone by choosing a specific voice, but you cannot add a regional accent like Scottish or Texan. The voices are designed for clear, standard pronunciation in each supported language.

Is Atlas neutral enough?

Atlas is described as neutral and news-style. This makes it highly suitable for factual, unbiased narration. It avoids emotional inflection, providing a clear and direct delivery. Use Atlas for historical facts, exhibit descriptions, or any content where a neutral, professional tone is required for clarity and accuracy.

Can I clone a specific voice?

No, custom voice cloning is not offered. You cannot upload your own voice or clone a specific person. The API provides eight fixed presets based on licensed recordings. You can adjust speed and tone, but you cannot create a unique voice clone for your brand or specific narrators.

Narrative Voices for History

Sound wave illustration for: Narrative Voices for History

Selecting the right voice is critical for museum audioguides where narrative tone shapes the visitor's experience. Our API offers specific presets designed to match historical content without requiring complex engineering.

Use Aria for warm, clear narration of historical events. Choose Luna for soft, storytelling modes that enhance emotional connection. Opt for Sage when delivering calm, instructional directions to visitors.

Aria: Warm and clear

Aria provides a warm, clear narrator voice ideal for standard historical tours. Her tone remains consistent and easy to follow, making her suitable for general exhibit descriptions.

The voice works well for long-form audio guides where clarity is paramount. It reduces listener fatigue during extended visits.

  • Ideal for general exhibit descriptions
  • Reduces listener fatigue on long tours

Luna: Soft storytelling

Luna’s soft, storytelling quality adds emotional depth to historical narratives. She is best suited for personal stories or dramatic historical accounts.

Her gentle tone helps engage visitors in immersive experiences without sounding overly formal or stiff.

  • Best for personal historical narratives
  • Adds emotional depth to exhibits

Sage: Calm instructional

Sage offers a calm, instructional voice perfect for wayfinding and educational content. His tone is neutral yet engaging, guiding visitors through complex layouts.

Use Sage for technical explanations or step-by-step tour instructions where clarity and pace are essential.

  • Ideal for wayfinding and directions
  • Clear instructional tone

Energetic Voices for Promos

Sound wave illustration for: Energetic Voices for Promos

Promotional content within museum exhibits demands energy to capture visitor attention quickly. Our API provides distinct vocal profiles tailored for high-impact narration, ensuring your audio guides remain engaging without sounding monotonous.

These voices are optimized for short, punchy messages that drive engagement, whether highlighting a new exhibit or encouraging a visit to the gift shop.

Nova: Bright and upbeat ['Nova delivers a bright, energetic tone ideal for modern art galleries or interactive displays. Her clear delivery keeps listeners engaged during fast-paced tours.', 'Use Nova for dynamic transitions between exhibits or to highlight special temporary exhibitions that require a sense of urgency and excitement.'] ['Ideal for modern and contemporary exhibits', 'Bright tone for quick transitions', 'Engages younger demographics effectively']

Felix: Energetic promo ['Felix is designed specifically for promotional content, bringing an energetic drive to your audio guide. His voice cuts through background noise, making him perfect for announcements.', 'Deploy Felix for limited-time offers, event reminders, or calls to action where a strong, persuasive voice is necessary to influence visitor behavior.'] ['High energy for promotional impact', 'Clear projection in noisy environments', 'Persuasive tone for calls to action']

Milo: Friendly conversational ['Milo offers a friendly, conversational style that feels like a knowledgeable local guide. This approach makes historical facts feel accessible and less formal.', 'Use Milo for narrative-driven tours where storytelling is key, helping visitors connect emotionally with the history being presented without feeling lectured.'] ['Approachable tone for storytelling', 'Conversational style for narrative tours', 'Builds rapport with visitors']

Authoritative Voices for News

Sound wave illustration for: Authoritative Voices for News

For institutions delivering factual historical data or breaking cultural news, voice neutrality is critical. Atlas provides a neutral, news-style delivery that presents information without emotional bias, ensuring the content remains the focus rather than the narrator’s personality.

Orion offers a deep, documentary-style tone ideal for long-form historical narratives. Its authoritative presence adds gravity to serious subjects, making it suitable for memorial sites or detailed era-specific guides where a respectful, commanding voice enhances the listener's immersion in the historical context.

Atlas: Neutral news style

Atlas delivers clear, unembellished narration perfect for time-sensitive updates, exhibit changes, or factual historical recitations. Its consistent pacing ensures listeners receive information without distraction, making it ideal for museum announcements or rapid-fire educational content where clarity outweighs emotional resonance.

  • Neutral tone avoids bias in historical facts
  • Consistent pacing for rapid information delivery
  • Ideal for exhibit updates and announcements

Orion: Deep documentary

Orion’s deep timbre creates a sense of gravitas, suitable for memorial sites or deep-dive historical segments. This voice adds a layer of seriousness and respect, enhancing the listener’s emotional connection to significant historical events without overwhelming the narrative with excessive drama.

Use Orion for long-form content where listener fatigue is a concern. Its steady, resonant delivery maintains engagement over extended periods, ensuring the historical narrative remains compelling from start to finish without requiring frequent voice changes.

  • Deep timbre adds gravitas
  • Ideal for memorial sites
  • Reduces listener fatigue in long tours

Professional tone options

Selecting the right professional tone depends on the specific historical context. Atlas suits neutral, factual reporting, while Orion enhances narrative depth. Both voices maintain high clarity, ensuring that complex historical data is communicated effectively to diverse audiences.

Pair these voices with appropriate speed settings (e.g., 0.75–1.0) to match the complexity of the content. Neutral tones work well with standard speeds, while deeper voices may benefit from slight slowing to ensure every historical detail is clearly heard by visitors in noisy museum environments.

  • Match tone to historical context
  • Adjust speed for content complexity
  • Optimize for noisy environments

Matching Voice to Content

Sound wave illustration for: Matching Voice to Content

Choosing the right voice ensures your audioguide matches the tone of the exhibit. A mismatched voice can distract visitors from the historical content or make serious facts feel trivial. Aligning vocal characteristics with narrative style improves comprehension and engagement for diverse museum audiences.

We provide distinct presets that range from warm narration to neutral reporting. This allows indie developers to curate a consistent auditory experience without needing custom voice design. The goal is seamless integration of speech into the physical space.

Children's tour voices

Use Luna for soft, storytelling sequences that feel gentle and inviting. Nova works well for bright, upbeat segments that keep younger listeners engaged. These presets avoid the monotony of standard robotic narration, making history feel accessible and fun for families.

  • Luna: soft, storytelling tone for gentle narratives
  • Nova: bright, upbeat energy for interactive play
  • Clear enunciation aids early language development

Serious historical facts

Atlas provides a neutral, news-style delivery ideal for factual exhibits. Orion’s deep, documentary voice adds gravity to significant historical events. Sage offers a calm, instructional tone for detailed explanations. These voices ensure that sensitive or complex topics are treated with appropriate seriousness and clarity.

  • Atlas: neutral style for factual data
  • Orion: deep tone for documentary depth
  • Sage: calm instruction for complex details

Engaging interactive guides

Felix delivers an energetic, promo-style voice perfect for interactive kiosks or quick prompts. Milo offers a friendly, conversational tone for user assistance. These voices respond well to short, repetitive interactions, reducing listener fatigue during longer museum visits. Adjust speed to match the user’s walking pace.

  • Felix: energetic style for quick prompts
  • Milo: friendly tone for user assistance
  • Adjust speed (0.25–4.0) for pacing

Language Compatibility

Sound wave illustration for: Language Compatibility

VoxMuse provides a unified voice engine where any of the eight presets can deliver natural narration in any of the 23 supported languages. This flexibility allows museum developers to maintain consistent character voices across diverse visitor demographics without managing separate voice models for each region or language group.

All voices speak all 23 languages

Aria, Nova, Luna, Sage, Felix, Milo, Atlas, and Orion each 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. You select the voice based on tone rather than language constraints, simplifying content management for international exhibits.

  • Single voice model supports 23 distinct languages
  • Voice personality remains consistent across language switches
  • No need for separate voice assets per region

Accent consistency

The API automatically detects language from the input script, defaulting to English for Latin text if unspecified. For non-English Latin-script content, explicitly setting the language parameter ensures accurate accent and pronunciation. This prevents mixed accents when switching between languages in a single audioguide tour.

Natural phrasing

Clear phrasing enhances listener retention, especially in noisy museum environments. The API handles up to 4096 characters per request, allowing you to structure long historical narratives into coherent segments. Splitting text appropriately ensures smooth transitions and maintains the narrator's rhythm throughout the experience.

Technical Considerations

Sound wave illustration for: Technical Considerations

Selecting the right audio format balances file size, compatibility, and quality for museum devices. MP3 offers broad support with smaller files, while WAV provides lossless clarity for high-end speakers. For modern mobile apps, OPUS ensures efficient bandwidth usage without sacrificing intelligibility. FLAC preserves quality for archival purposes, and PCM is ideal for raw processing pipelines.

Adjusting the speed parameter (0.25–4.0) allows you to match the narration pace to the visitor’s walking speed or device capabilities. Slower speeds improve clarity in noisy environments, while faster speeds keep interactive tours engaging. Test different rates with your target audience to ensure comfortable listening without rushing through historical context.

Audio format selection

Choose MP3 for general mobile apps due to its balance of quality and small file size. Use WAV for high-fidelity playback on dedicated museum audio guides where storage is less constrained. OPUS is efficient for variable mobile networks, while FLAC suits archival needs. PCM allows direct integration with custom audio processing tools or raw playback.

  • MP3: Best for mobile apps and general distribution
  • WAV: Lossless quality for dedicated audio guides
  • OPUS: Efficient streaming for variable network conditions
  • FLAC: High-quality archival storage
  • PCM: Raw data for custom processing

Speed adjustment range

The speed parameter ranges from 0.25 to 4.0, allowing precise control over narration pace. Lower speeds enhance clarity in noisy museum halls, while higher speeds keep interactive tours dynamic. Adjust based on the complexity of the content and the expected listening environment. Test with actual users to find the optimal balance between information density and comprehension.

  • Range: 0.25 to 4.0
  • Slower speeds improve clarity in noisy environments
  • Faster speeds maintain engagement in interactive tours
  • Adjust based on content complexity and listening conditions

Character limit impact

Each API request supports up to 4096 characters. For longer tours, split the text into segments and join the audio files. This approach ensures consistent quality and manageable processing times. Monitor character counts to optimize costs, as pricing is based on character usage. Longer texts may take longer to generate, so plan your audio assets accordingly.

  • Limit: 4096 characters per request
  • Split longer texts into segments
  • Join audio files for seamless playback
  • Monitor character counts for cost optimization

Licensing and Rights

Sound wave illustration for: Licensing and Rights

VoxMuse’s synthetic voices are derived from licensed recordings in the LibriTTS-R dataset, ensuring clear legal grounding for commercial audioguide deployments. This foundation allows indie developers to integrate high-quality narration without navigating complex royalty structures or uncertain ownership trails common in custom voice projects.

LibriTTS-R license

The underlying voice models are trained on data from the LibriTTS-R dataset, which provides a transparent licensing framework. This ensures that the synthetic speech generated for your museum guides rests on a known legal basis, reducing risk for tourism applications.

  • Based on verified dataset licensing

CC BY 4.0 usage

Key source material is governed by the Creative Commons Attribution 4.0 International license. This permission allows you to use the resulting audio files in commercial audioguide tours, digital installations, and visitor apps.

  • Commercial use permitted

Public domain source

Additional voice training data originates from LibriVox public-domain audiobook readings. These recordings are free of copyright restrictions, providing a robust base for voice synthesis that is inherently clear for historical narration and educational content in museum settings.

Licensing and Rights

Audio Quality Features

Sound wave illustration for: Audio Quality Features

Every generated audio file carries an inaudible synthetic watermark, clearly marking the content as AI-generated speech. This ensures transparency for museum visitors without disrupting the listening experience. The watermark is embedded directly into the audio data upon synthesis.

Our TTS API provides two models: tts-1 and tts-1-hd. Both use the same engine but differ in fidelity, allowing you to balance audio clarity against processing costs based on your audioguide's specific needs.

Synthetic watermark

Each audio file includes an inaudible synthetic watermark that identifies it as synthetic speech. This feature provides clear provenance for museum collections, ensuring transparency for visitors regarding the origin of the narration without adding any audible artifacts or noise to the playback.

  • Inaudible and transparent to the listener
  • Automatically embedded during synthesis
  • No manual tagging required

High-definition option

Choose between tts-1 and tts-1-hd models. The HD option offers higher fidelity for detailed historical descriptions, while the standard model remains cost-effective for general narration. Both models support the same eight voices and 23 languages, allowing consistent voice casting across your entire tour collection regardless of quality tier selected.

  • tts-1 for standard clarity
  • tts-1-hd for enhanced fidelity
  • Same voice selection for both tiers

File size optimization

Select the optimal output format to manage storage and playback efficiency. Choose mp3 for compact file sizes suitable for mobile apps, or wav and flac for archival-quality preservation. The API returns the audio file directly in the response, ensuring no intermediate storage is required for your immediate integration workflow.

pcm provides raw data for specialized processing pipelines, while opus offers efficient compression. This flexibility allows you to tailor the audio delivery method to the specific constraints of your audioguide platform without sacrificing audio integrity.

  • mp3 for small file sizes
  • wav and flac for high quality
  • opus for efficient streaming

User Experience Impact

Sound wave illustration for: User Experience Impact

Consistent narration keeps visitors engaged throughout a museum tour. By maintaining a steady pace and clear articulation, the API ensures that historical narratives are absorbed without confusion. This reliability reduces the cognitive load on listeners, allowing them to focus on the exhibits rather than deciphering unclear audio.

Listener retention

High-quality synthetic speech mimics natural intonation patterns, which helps maintain attention during longer audio guides. When the voice remains consistent and expressive, visitors are more likely to complete the entire tour experience without losing interest.

Clarity in noisy environments

Museum halls often have ambient noise from crowds or other exhibits. Choosing higher-bitrate formats like WAV or FLAC ensures the audio remains crisp and distinct against background sounds. The clear vocal ranges of voices like Aria and Atlas cut through ambient noise effectively.

Voice fatigue reduction

Long tours can cause listener fatigue if the voice is too monotonous or harsh. Selecting softer voices like Luna for storytelling segments or Sage for instructional parts helps reduce ear strain. Adjusting the speed between 0.25 and 4.0 also ensures comfortable listening for all ages.

Implementation Tips

Sound wave illustration for: Implementation Tips

Designing a museum audio tour requires careful pacing to match the visitor's walking speed. Use the speed parameter (0.25–4.0) to adjust narration duration, ensuring it aligns with exhibit dwell times without rushing or dragging. Shorter texts generate quickly, while longer descriptions may queue during peak hours.

Mixing voices adds variety to long tours, but keep transitions smooth. Use consistent voice personalities for related exhibits to maintain narrative cohesion. Test different speeds to find the optimal balance between clarity and engagement.

Mixing voices for variety

Assign distinct voices to different exhibit sections or historical periods. For instance, use Aria for warm historical narratives and Felix for energetic promotional inserts. This differentiation helps listeners navigate the tour structure without visual cues.

  • Use Aria for historical narratives
  • Use Felix for promotional inserts
  • Maintain voice consistency within sections

Pacing adjustments

Adjust the speed parameter to match the average walking pace of your audience. Slower speeds (0.5–1.0) work well for detailed artifact descriptions, while faster speeds (1.5–2.0) suit brief transitions between exhibits.

Testing with users

Conduct real-world tests in your museum environment. Check audio clarity against background noise and ambient sound. Gather feedback on voice fatigue and pacing preferences to refine your final implementation.

  • Test in actual museum environments
  • Check clarity against background noise
  • Gather feedback on voice fatigue

Cost vs Quality

Sound wave illustration for: Cost vs Quality

Choosing between standard and high-definition audio models depends on your audioguide’s context and budget. Both models use the same engine and price tier, so the decision rests entirely on your preference for audio fidelity versus file size constraints.

The standard pricing structure simplifies cost prediction for tourism projects. You pay only for what you generate, with no monthly fees or minimum commitments, making it ideal for indie developers managing variable visitor traffic.

Standard vs HD pricing

Both 'tts-1' and 'tts-1-hd' models operate at the same rate: $10 per 1 million characters. This uniform pricing means you do not pay extra for higher quality, allowing you to upgrade audio clarity without increasing your API costs.

Failed requests are free, ensuring you never pay for audio that does not generate successfully. This protects your budget during testing phases or when handling edge cases in historical texts.

  • Same price for both standard and HD models
  • $0.01 per 1,000 characters including spaces
  • No additional fees for higher fidelity

Character count efficiency

Prices are calculated based on character count, including spaces and punctuation. Understanding this metric helps you estimate costs for long museum descriptions accurately. Each request supports up to 4,096 characters, so longer exhibits require splitting text.

For typical museum content, generation speed is roughly real-time. This means a 10-minute audio guide requires processing approximately 1,500 characters per minute of audio, helping you forecast API usage for large collections.

  • Count includes spaces and punctuation
  • Max 4,096 characters per request
  • No minimum spend required

Trial voice testing

New accounts signed in with Google receive $0.50 in trial credit, valid for 7 days. This provides approximately 50,000 characters to test voice quality and integration before committing to larger purchases.

Use this trial to verify that voices like Aria or Luna match your museum’s tone. Since balances do not expire and top-ups are processed quickly via USDT, you can scale up seamlessly after testing.

  • $0.50 trial credit for Google sign-ins
  • Approximately 50,000 characters available
  • Valid for 7 days

Future Scalability

Sound wave illustration for: Future Scalability

As your museum app grows, the API handles volume scaling automatically without requiring complex infrastructure changes. Requests are processed sequentially, so under high load, synthesis may queue briefly, ensuring stable output without service degradation. This predictable behavior supports steady expansion from small tours to city-wide audio guides.

Adding new languages

The API supports 23 languages, allowing you to expand into new markets without integrating new voice engines. Each voice can speak any supported language, ensuring consistent audio quality across diverse linguistic contexts. You can add new language support simply by updating the language parameter in your requests.

  • Support for 23 languages including Arabic, Chinese, and Swahili
  • Automatic language detection if the language field is omitted
  • Consistent voice quality across all supported languages

Volume growth handling

The API manages increasing request volumes by processing them in a queue. While requests are handled one after another, the system ensures no data loss or corruption. For large-scale deployments, splitting long texts into 4096-character segments helps maintain consistent performance and avoids timeout issues.

  • Sequential processing ensures data integrity under load
  • Text splitting into 4096-character parts optimizes throughput
  • No upper limit on total characters generated per account, but requests are limited to 4096 characters each.

API consistency

The API maintains a consistent request shape across all endpoints, making it easy to integrate with standard tools like the OpenAI SDK. The base URL remains stable, and the JSON body structure does not change, reducing the need for frequent code updates as your application scales.

  • Compatible with OpenAI SDKs via simple base URL change
  • Stable JSON request format across all endpoints
  • No breaking changes expected to the core API structure

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. We build immersive audio stories for independent creators who want to bring museums to life through sound.

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. IETF RFC 7845: Ogg Opus
  3. FFmpeg documentation
  4. curl documentation
  5. LibriTTS-R paper (arXiv 2305.18802)
  6. Node.js documentation
  7. Creative Commons Attribution 4.0
  8. OpenAI API reference: create speech
  9. ISO 639 language codes
  10. openai-node on GitHub
  11. Tether (USDT)
  12. LibriTTS-R corpus (OpenSLR 141)

More on this site