Frequently asked questions
Which voice is best for children's tours?
Nova is a bright, upbeat female voice well-suited for children’s tours. Luna offers a soft, storytelling style for narrative content. Both voices support 23 languages, allowing you to create engaging, multilingual audio guides that appeal to young audiences without needing separate voice actors for each language version.
Can I use multiple voices in one tour?
Yes, you can switch voices between requests. Since each request generates a separate audio file, you can split your tour into segments and assign different voices (e.g., Felix for promo, Luna for stories) to each part. You then join the audio files to create a cohesive tour experience.
Do voices sound different in other languages?
No, the same eight voice presets are used across all 23 supported languages. Each voice maintains its specific tonal character—like Aria’s warm narration or Atlas’s neutral news style—regardless of the language spoken. You simply specify the language code in the request to ensure correct pronunciation and accent.
What is the difference between tts-1 and tts-1-hd?
Both models use the same engine and cost $10 per 1 million characters. The only difference is the output format: tts-1-hd generates higher-fidelity audio, while tts-1 offers standard quality. Choose based on your storage or bandwidth needs. Character limits and processing logic remain identical for both options.
Are the voices male or female only?
The eight presets include both female and male voices. Female options are Aria, Nova, Luna, and Sage. Male options are Felix, Milo, Atlas, and Orion. This mix allows you to select voices that best fit your museum’s tone, whether you prefer a calm instructional voice or an energetic promo style.
Can I change the speed of the voice?
Yes, you can adjust the speed between 0.25 and 4.0. This allows you to slow down complex historical explanations for clarity or speed up short announcements. The API accepts a speed parameter in the JSON body, giving you precise control over the pacing of your audioguide content.
Do voices work well with accents?
Voices handle accents naturally within the 23 supported languages. If you omit the language parameter, Latin script defaults to English. For other Latin-script languages, set the language parameter to ensure the voice applies the correct pronunciation rules for that language. Note that the API selects from eight preset voices; it does not offer specific regional accent variants (e.g., Australian vs. British English) beyond the language-level selection.
Is there a voice for news updates?
Atlas is a neutral, news-oriented male voice ideal for updates. He delivers clear, straightforward speech suitable for time-sensitive information or historical facts. You can use the tts-1 model with Atlas to generate concise audio segments that inform visitors without adding emotional bias to the historical narrative.
Which voice is best for museums?
Aria offers a warm, clear narrator style, while Sage provides a calm, instructional tone. Both are excellent for museum contexts. Aria suits general tours, while Sage is ideal for detailed exhibits. You can also use Luna for storytelling elements to engage visitors with historical narratives in a soft, approachable way.
Can I use the voices in a podcast?
Yes, you can generate audio for podcasts using any of the eight voices. Each file includes an inaudible watermark marking it as synthetic speech. Since input text is not used for training, your content remains private. You can produce episodes in multiple languages using the same voice presets for consistency.
Are the voices consistent across updates?
Yes, the voice presets are stable. Since the voices are based on licensed recordings from LibriTTS-R and LibriVox, the tonal characteristics remain consistent. You can regenerate audio for updated tour content using the same voice ID, ensuring your audioguide series maintains a uniform audio identity across all episodes and languages.
Do I need to credit the voices?
The voices are based on LibriTTS-R (CC BY 4.0) and LibriVox readings. The source license requires attribution when you redistribute the audio. While the API does not enforce this automatically, visible credit remains your legal obligation under the source license. The generated files include an inaudible watermark, but this does not replace the need for explicit attribution required by CC BY 4.0.
Can I clone my own voice?
No, custom voice cloning is not offered. You cannot upload your own voice or design a new voice. You must choose from the eight available presets: Aria, Nova, Luna, Sage, Felix, Milo, Atlas, or Orion. These licensed voices are ready to use immediately via the API without additional setup.
What is the best format for mobile?
MP3 is widely supported on mobile devices for both tts-1 and tts-1-hd models. You can also request WAV, Opus, FLAC, or PCM. For mobile audioguides, MP3 offers a practical balance of quality and file size, ensuring reliable playback across various heritage and tourism apps without requiring specialized codecs.
Aria: Warm Narrator
Aria offers a warm, clear narration style ideal for museum history exhibits. Her voice guides visitors through historical narratives with clarity, ensuring that complex timelines and cultural contexts remain accessible without becoming distracting.
Ideal for history exhibits
Aria’s tonal quality suits historical content where clarity matters. Her narration style helps preserve the integrity of historical facts while keeping the listening experience engaging for diverse visitor groups.
- Suitable for static exhibits
- Works well with background ambient sound
Clear and engaging
Her delivery balances warmth with precision, making her well-suited for longer audio scripts. This clarity helps maintain visitor attention throughout extended tour segments without requiring frequent pauses.
For detailed historical accounts, Aria’s consistent pacing ensures that information is absorbed effectively, reducing the cognitive load on listeners navigating new environments.
Well-suited for long scripts
Aria’s clear narration style holds up well over extended durations. This makes her a reliable choice for comprehensive tour guides where consistency is key to a cohesive visitor experience.
Nova: Bright Upbeat
Nova brings a bright, upbeat energy to your museum content. As a female voice preset, she delivers clear and engaging narration that keeps visitors attentive during their tour.
Perfect for promotional tours
Nova is ideal for promotional tours where you need to capture attention quickly. Her energetic tone works well for introductory segments or highlights of special exhibitions.
- Engages visitors with a lively presence
- Ideal for introductory tour segments
- Works well for special exhibition highlights
Energetic and inviting
Her inviting voice makes complex historical details feel accessible. Visitors appreciate the clear, upbeat delivery that enhances the overall experience without being overwhelming.
- Accessible delivery for complex details
- Clear and upbeat tone
- Enhances visitor experience
Great for short snippets
Nova excels in short snippets, where every second counts. Her energetic style ensures that brief audio clips remain engaging and informative, perfect for quick guides.
- Optimized for brief audio clips
- Maintains engagement in short formats
- Perfect for quick guide segments

Luna: Soft Storyteller
Luna brings a soft, storytelling quality to museum audio, making her ideal for emotional exhibits that require a gentle touch. Her voice provides a calming presence, allowing visitors to absorb historical narratives without feeling rushed or overwhelmed by high energy.
This preset excels in creating an immersive atmosphere for sensitive displays, where tone matters more than volume. It supports the specific constraints of heritage experiences by delivering clear, synthetic speech that respects the emotional weight of the content.
Soft and immersive
Luna’s voice is designed for soft, storytelling delivery, offering a gentle narration style that complements historical narratives. This makes her suitable for exhibits where a lighter, more engaging tone helps maintain visitor interest without being intrusive.
- Gentle narration style for historical narratives
Ideal for emotional exhibits
For exhibits that evoke strong feelings, Luna’s soft tone provides a calming presence. She helps guide visitors through sensitive content with a steady, reassuring voice that enhances the emotional impact of the display.
Her voice works well for personal stories or poignant historical moments, ensuring the audio feels connected to the exhibit’s emotional core rather than just informational.
- Supports sensitive content with steady tone
- Enhances emotional impact of displays
Calming presence
Luna offers a soothing delivery that can help reduce the cognitive load on visitors, allowing them to focus on the visual elements of the exhibit. Her voice is neutral enough to remain professional yet warm enough to feel inviting.
This calming quality is particularly useful for long scripts or repetitive audio loops, where a consistent, gentle voice prevents listener fatigue.
- Reduces cognitive load for visitors
- Prevents listener fatigue in loops
Sage: Calm Instructor
Sage offers a calm, instructional voice designed for structured information delivery. Its neutral tone ensures clarity without drawing attention to the narrator, making it ideal for museum exhibits where the content takes center stage. The voice remains consistent and professional, supporting long-form narration without listener fatigue.
Clear and instructional
Sage delivers information with precise articulation and steady pacing. This clarity helps visitors absorb complex historical details or directional instructions without distraction. The voice avoids emotional peaks, maintaining a reliable rhythm that suits both static displays and moving tour routes.
- Consistent pacing for complex narratives
- Clear articulation of proper nouns and dates
- Low emotional variance reduces listener fatigue
Best for educational content
Use Sage for detailed exhibit descriptions, timeline narrations, or interactive learning stations. Its neutral delivery supports educational material where accuracy and clarity matter more than dramatic flair. Visitors can focus on the historical facts rather than the performance style.
Neutral tone
The voice avoids strong personality traits, ensuring it blends seamlessly with various exhibit themes. It works well for informational desks or general background narration where the goal is to inform rather than entertain. This neutrality allows the audio to adapt to different contexts without sounding out of place.
- Adapts to diverse exhibit themes
- Avoids distracting emotional cues
- Suits both indoor and outdoor installations
Felix: Energetic Promo
Felix brings high energy to your promotional tours, making him ideal for ads and short-form content. His lively tone keeps visitors engaged without sounding forced, ensuring that your museum’s highlights get the attention they deserve.
High energy for ads
Use Felix to create dynamic audio snippets that capture attention quickly. His energetic delivery is perfect for highlighting special exhibits or upcoming events, ensuring your message stands out in a busy museum environment.
Engaging and lively
Felix’s voice is designed to be engaging and inviting, making complex information feel approachable. Whether you are introducing a new gallery or explaining a interactive exhibit, his tone helps maintain visitor interest throughout the tour.
Short form content
Ideal for brief, impactful audio clips, Felix excels in short-form content. His clear and energetic style ensures that even quick updates or announcements are delivered with clarity and enthusiasm, fitting seamlessly into any guided experience.
Milo: Friendly Conversational
Milo delivers a natural, conversational voice ideal for interactive museum tours. His friendly tone helps visitors feel guided rather than lectured, making complex exhibits feel accessible.
Natural dialogue style
Milo sounds like a knowledgeable guide rather than a robotic announcer. This conversational pacing keeps visitors engaged without demanding intense focus, perfect for self-guided audio walks.
- Sounds like a real person speaking
- Reduces listener fatigue during long tours
- Ideal for casual, non-academic audiences
Best for interactive tours
Use Milo for tours where visitors interact with exhibits or answer prompts. His approachable style suits questions, trivia, or 'choose your own adventure' style museum paths.
- Perfect for Q&A style audio content
- Works well with interactive exhibit triggers
- Encourages visitor participation and engagement
Approachable tone
Milo’s warm demeanor makes technical or historical details feel less intimidating. He bridges the gap between academic content and casual tourism, ensuring broad audience appeal.
- Reduces perceived complexity of content
- Appeals to families and casual tourists
- Creates a welcoming atmosphere for all ages
Atlas: Neutral News
Audience members at information desks require precise, unembellished delivery. Atlas provides a neutral, news-style voice that prioritizes clarity over emotional inflection, ensuring that critical operational details are understood without distraction.
Professional and clear
Atlas delivers consistent, straightforward narration ideal for wayfinding, opening hours, and safety instructions. The neutral tone avoids interpretive bias, making it suitable for high-traffic areas where information must be received quickly and accurately.
Ideal for information desks
Use Atlas for static informational content such as directory listings, exhibit rules, or emergency procedures. Its steady cadence supports repetitive listening without causing listener fatigue, ensuring guests retain essential details during their visit.
Neutral, news
Atlas is described as neutral and news-oriented, offering an authoritative yet unbiased delivery. This makes it distinct from warmer voices like Aria or Luna, focusing purely on the accurate transmission of factual data.

Orion: Deep Documentary
Orion delivers a deep, resonant voice ideal for museum audio guides requiring a serious, documentary-style narration. This preset is designed for heritage content where a grounded, authoritative tone enhances the historical narrative without sounding overly dramatic or casual.
Deep and resonant
The vocal range is lower and steady, providing a consistent auditory backdrop for long-form exhibit descriptions. This depth helps maintain listener engagement during extended audio tours without causing vocal fatigue, ensuring the narration remains clear and distinct.
- Lower frequency range for a grounded sound
- Steady pacing suitable for historical facts
- Reduced pitch variation for consistency
Best for documentaries
Orion is well-suited for documentary-style audio guides where context and accuracy are prioritized over emotional flair. It works effectively for explaining complex historical events or archaeological findings where a neutral, serious demeanor supports the content's credibility.
- Ideal for historical context narration
- Supports detailed exhibit explanations
- Maintains focus on content over voice personality
Serious tone
The tone is professional and restrained, avoiding the brightness of Nova or the warmth of Aria. This makes it appropriate for solemn memorials, serious historical periods, or educational content where a lighter voice might seem incongruous with the subject matter.
- Professional and restrained delivery
- Suitable for solemn or serious topics
- Avoids emotional exaggeration
Comparison by Use Case
Museums and hotels share a need for clear, multilingual narration, but their content structures differ. Museums often produce long, sequential scripts for exhibits, while hotel guides typically use short, contextual snippets for rooms or amenities. MuseVoice handles both by allowing precise speed adjustments and format choices to fit specific playback devices.
Museums vs. Hotels
Museum audioguides usually require longer, continuous narration for historical context. Hotel audio guides benefit from brief, point-of-interest updates. MuseVoice’s API processes texts up to 4096 characters per request, allowing teams to split long scripts into manageable parts for consistent audio quality across both long tours and short snippets.
- Museums: Long scripts, sequential flow, historical context.
- Hotels: Short snippets, location-specific, quick updates.
Short vs. Long Tours
Short tours under two minutes generate quickly, while longer tours may require splitting text into 4096-character segments. Each segment is synthesized independently and joined later. This approach ensures consistent audio quality regardless of script length, preventing listener fatigue from variable pacing or technical limits.
- Split long texts into 4096-character parts.
- Join audio files for seamless playback.
- Adjust speed (0.25–4.0) to fit time limits.
Educational vs. Promotional
Educational content benefits from clear, instructional tones suitable for learning. Promotional content often uses energetic or deep voices to engage listeners. MuseVoice offers eight preset voices, allowing teams to select tones that match their specific content type, from neutral news-style delivery to warm storytelling.
- Educational: Clear, instructional, neutral tones.
- Promotional: Energetic, deep, or warm tones.
- Choose from 8 licensed presets.
Voice Selection Tips
Choosing the right voice ensures your audioguide resonates with visitors. Match the voice’s character to your content’s emotional weight and length. For example, use Aria for warm, historical narratives or Luna for immersive, storytelling exhibits. Conversely, select Felix or Milo for lively, promotional snippets and interactive tours. This alignment keeps listeners engaged without overwhelming them with mismatched tones.
Match tone to audience
Align the voice’s personality with your exhibit’s mood. Use Aria for warm, historical narratives or Luna for immersive, storytelling exhibits. Select Felix or Milo for lively, promotional snippets and interactive tours. This alignment keeps listeners engaged without overwhelming them with mismatched tones.
- Use Aria for warm, historical narratives.
- Select Luna for immersive, storytelling exhibits.
- Choose Felix or Milo for lively, promotional snippets.
Consider language nuances
While any voice can speak any of the 23 supported languages, the API does not offer distinct regional accents (e.g., US vs. UK English). You must explicitly set the language parameter for non-English Latin-script text to ensure accurate pronunciation. Without this, the system defaults to English, which may mispronounce specific regional terms or proper nouns.
Test with focus groups
Generate short samples of your actual tour scripts using different voices. Play these back in the actual museum environment to check clarity against background noise. Ask visitors which voice feels most natural and authoritative for the specific content. This practical testing helps you select the optimal voice before committing to full-scale production.
- Generate short samples of actual tour scripts.
- Play back in the actual museum environment.
- Ask visitors which voice feels most natural.
Technical Considerations
Choosing the right audio format ensures compatibility across museum kiosks, mobile apps, and web players. MuseVoice supports MP3, WAV, OPUS, FLAC, and PCM, allowing you to balance quality and bandwidth based on your delivery method.
Speed adjustments range from 0.25x to 4.0x, enabling precise synchronization with exhibit pacing. Slower speeds enhance clarity for complex historical narratives, while faster speeds suit brief informational snippets.
Format compatibility ['MuseVoice outputs MP3, WAV, OPUS, FLAC, and PCM files. MP3 offers universal playback support for web and mobile apps, while WAV ensures broad compatibility with legacy museum kiosks. OPUS provides efficient compression for streaming, and FLAC preserves original fidelity for archival purposes.', 'PCM delivers uncompressed raw audio, useful for direct hardware integration where post-processing is handled elsewhere. Choose based on your player’s codec support and network constraints.'] ['MP3: Universal compatibility for web and mobile.', 'WAV: Broad support for legacy kiosks.', 'OPUS: Efficient for streaming environments.', 'FLAC: Preserves original audio fidelity.', 'PCM: Uncompressed raw audio for hardware integration.']
Speed adjustments ['The API accepts speed values between 0.25 and 4.0. Adjusting speed allows you to match the pacing of physical exhibits, ensuring visitors have enough time to read plaques while listening.', 'Slower speeds (e.g., 0.8x) improve comprehension for complex historical details or non-native speakers. Faster speeds (e.g., 1.5x) are suitable for brief announcements or promotional snippets where time is limited.']
File size impact ['Generated audio files are returned directly in the response and not stored, minimizing server storage needs. File size varies by format and speed, with MP3 and OPUS offering smaller footprints for easier distribution.', 'Larger WAV and FLAC files ensure high quality but require more bandwidth for download. Plan your distribution strategy based on visitor device capabilities and network availability at your sites.']

Licensing and Usage
MuseVoice voices are derived from the LibriTTS-R dataset (CC BY 4.0) and public-domain LibriVox recordings. This foundational licensing ensures the synthetic voices are legally sourced for professional deployment in museum and tourism products.
LibriTTS-R license
Our audio engines utilize data licensed under Creative Commons Attribution 4.0 (LibriTTS-R) and public-domain audiobook readings from LibriVox. This provides a clear, permissive legal foundation for generating synthetic speech without requiring complex additional licensing for the voice models themselves.
Commercial use allowed
You retain ownership of the generated audio files. MuseVoice does not claim rights over your specific output, and input text is used solely for synthesis, not for training our models. Generated audio is returned in the response and not stored on our servers, ensuring your content remains private and ready for immediate commercial use in audioguides or tours.
Watermark details
Every audio file generated includes an inaudible digital watermark marking it as synthetic speech. This watermark helps identify the content as AI-generated without affecting the listening experience, providing transparency for your museum visitors while maintaining professional audio quality.
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. Our audioguide transforms museum visits into immersive storytelling experiences for visitors.
23 languages
Any voice speaks any of these languages.
Sources and further reading
- Wikipedia
- W3C WAI: making audio and video accessible
- ISO 639 language codes
- TRON developer hub: TRC-20
- FFmpeg documentation
- IETF RFC 7845: Ogg Opus
- LibriTTS-R corpus (OpenSLR 141)
- Python documentation
- OpenAI API reference: create speech
- Node.js documentation
- MDN: Web Speech API
- LibriVox public-domain audiobooks
