Frequently asked questions
What are the 8 voice presets?
We offer four female voices: Aria (warm, clear narrator), Nova (bright, upbeat), Luna (soft, storytelling), and Sage (calm, instructional). The male options include Felix (energetic promo), Milo (friendly conversational), Atlas (neutral news), and Orion (deep documentary). Each preset is designed for specific tonal needs in mobile applications.
Is there a difference in price between voices?
No, all eight voice presets use the same engine and incur identical costs. The price is $10 per 1 million characters, regardless of which voice you select. You pay based on the character count in your input text, not on the specific voice model used for generation.
Can I use multiple voices in one app?
Yes, you can use any combination of the eight voices within a single application. Simply send separate API requests for each segment of text, specifying the desired voice parameter in the JSON body. You can switch between voices like Aria, Felix, or Atlas dynamically based on your content needs.
Which voice is best for children?
Luna offers a soft, storytelling tone, while Nova is bright and upbeat. You should test these presets with your target audience to determine which sounds most appropriate, as no voice is explicitly labeled for pediatric use.
Do voices support all 23 languages?
Yes, every voice preset can speak any of the 23 supported languages. These include 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 can assign any voice to any supported language.
Are the voices realistic?
Voices are based on licensed recordings from LibriTTS-R and LibriVox public-domain audiobooks. Each generated file contains an inaudible watermark marking it as synthetic speech. They are realistic synthetic voices suitable for mobile apps, but they are not human recordings and carry a distinct synthetic signature.
Can I hear all voices before buying?
Yes, you can test all voices using your free trial credit of $0.50, which covers about 50,000 characters. Use the API to generate audio samples for each preset. If you prefer, you can also check the /v1/voices endpoint to view available voices and their details before generating audio.
Is there a difference between tts-1 and tts-1-hd?
Both models use the same engine and price. The only difference is the model identifier in your API request. You can choose either "tts-1" or "tts-1-hd" without any change in cost or voice characteristics. They are functionally identical for the purpose of generating speech via the API.
Do voices sound the same in different languages?
Yes, the same voice preset retains its tonal character across all 23 supported languages. For example, Aria remains a warm narrator whether speaking English or Japanese. The voice characteristics are consistent, though the accent and pronunciation will naturally adapt to the language specified in the request or detected automatically.
Can I change voices dynamically?
Yes, you can change voices dynamically by sending separate API requests for different text segments. Each request allows you to specify a different voice parameter. You can switch between voices like Felix and Luna within your app based on user interaction or content type.
What is the best voice for news?
Atlas is a neutral, news-oriented preset designed for clear, professional delivery. While other voices can read news text, Atlas is specifically configured for this use case to ensure a consistent, authoritative style. It is a reliable choice for delivering updates in your mobile app, though it is not the single best option.
Are the voices copyrighted?
The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain audiobook readings. The generated audio files are synthetic and include an inaudible watermark. You do not need to worry about individual voice copyright issues, as the underlying recordings are licensed or public domain.
Can I use voices for background audio?
Yes, you can generate audio files for background use by setting the response_format to mp3, wav, opus, flac, or pcm. The API returns the audio file directly, which you can then play in your app. There are no restrictions on using the generated audio for background purposes.
How do I compare voice quality?
Generate audio samples for each voice using your API key. You can test all eight presets with your free trial credit. Use the /v1/voices endpoint to list voices and their attributes. Compare the output by listening to generated samples in your target languages to assess clarity and tone.
Female Voices Overview
LuminaVoice offers three distinct female voices designed for different mobile app contexts. Each voice leverages the same underlying engine, ensuring consistent audio quality and pricing regardless of your choice. This allows your team to select the perfect tone for your international releases without technical trade-offs.
Aria: Warm and clear narrator
Aria delivers a warm, clear narration style ideal for guided tours or instructional content. Her tone is professional yet approachable, making her suitable for apps that require clear communication of information.
Nova: Bright and upbeat
Nova provides a bright, energetic delivery perfect for entertainment or social apps. Her upbeat nature engages users quickly, making her a strong choice for notifications or casual interactions where energy is key.
Luna: Soft storytelling voice
Luna offers a soft, storytelling tone that works well for reading apps or immersive experiences. Her gentle delivery helps maintain user engagement during longer text sessions without causing listener fatigue.
Male Voices Overview
Male voices offer distinct tonal qualities for different mobile app contexts. Felix, Milo, and Atlas each serve specific functional roles, from promotional energy to neutral clarity. Selecting the right preset ensures your synthetic speech aligns with user expectations without requiring complex configuration.
Felix: Energetic and promotional
Felix delivers an energetic tone ideal for promotional content or dynamic notifications. His delivery style adds urgency and engagement, making him suitable for apps that need to capture attention quickly. Use Felix for limited-time offers or active call-to-action segments.
- Ideal for promotional banners
- High energy delivery
- Engages users actively
Milo: Friendly and conversational
Milo provides a friendly, conversational tone that feels natural in dialogue-heavy applications. His voice supports a relaxed user experience, making him well-suited for chat assistants or casual guides. This preset helps maintain a human-like interaction without sounding robotic.
- Conversational tone
- Friendly delivery
- Good for chat apps
Atlas: Neutral and news-ready
Atlas offers a neutral, news-ready voice that prioritizes clarity and neutrality. It is designed for straightforward information delivery, such as news updates or factual reports. His consistent tone ensures that content remains the focus, reducing listener fatigue during long reads.
- Neutral tone
- News-ready delivery
- High clarity
Orion Voice Characteristics
Orion delivers a deep, documentary-style voice ideal for serious content. Its distinct tone provides high clarity for complex text, making it a reliable choice for users seeking a professional audio experience.
Deep and documentary style
Orion’s low pitch creates a grounded, authoritative sound suitable for long-form narration. This characteristic helps maintain listener engagement during extended app sessions without causing fatigue.
Suitable for serious content
The voice excels in delivering news updates, educational material, or factual explanations. Its steady cadence ensures that information is conveyed with appropriate weight and professionalism.
High clarity for complex text
Dense technical descriptions or multi-clause sentences are rendered with precise articulation. Orion minimizes ambiguity, ensuring that users correctly understand nuanced details in your mobile application.
Voice Selection by App Type
Selecting the right voice preset depends heavily on your app’s genre. We recommend specific presets based on typical user expectations and content styles to ensure a natural listening experience without requiring custom voice design.
Education apps: Luna or Milo
Luna provides a soft, storytelling tone ideal for narrative-driven learning or children’s content. Milo offers a friendly, conversational style that works well for interactive quizzes or casual language exercises. Both presets handle instructional text clearly, supporting languages like English, Spanish, and Hindi effectively.
- Luna: Suitable for narrative or story-based content.
- Milo: Ideal for conversational or interactive exercises.
News apps: Atlas or Aria
Atlas delivers a neutral, news-ready voice suitable for headlines and breaking updates. Aria serves as a warm, clear narrator for longer news summaries or podcast-style feeds. Both voices maintain high clarity across 23 languages, ensuring listeners receive information without distraction.
- Atlas: Neutral tone for headlines and brief updates.
- Aria: Clear narration for summaries and features.
Entertainment: Nova or Felix
Nova brings a bright, upbeat energy perfect for gaming interfaces or social features. Felix offers an energetic, promo-style voice for advertisements, rewards, or event announcements. These presets engage users with dynamic tones that match the lively nature of entertainment apps.
- Nova: Upbeat tone for social or gaming features.
- Felix: Energetic style for promotions and ads.
Language Performance
LuminaVoice supports 23 languages, enabling mobile apps to reach global audiences with consistent synthetic speech quality. Each voice preset can speak any supported language, allowing teams to select the most appropriate tone for their specific market without managing separate voice models.
Arabic voice clarity
Arabic is supported with high clarity across all eight voice presets. The engine handles complex script structures effectively, ensuring that text-to-speech output remains intelligible for users in the Middle East and North Africa regions.
Hindi pronunciation accuracy
Hindi pronunciation is accurate and consistent, leveraging the same underlying engine as other languages. Teams can rely on clear enunciation for educational or informational apps targeting Indian audiences, with no additional configuration required for basic usage.
Spanish natural flow
Spanish text generates with natural prosody and flow. Whether using female or male voices, the speech pattern adapts well to conversational interfaces, making it suitable for chat apps or customer service bots without sounding robotic.
Technical Comparison
All eight voice presets—Aria, Nova, Luna, Sage, Felix, Milo, Atlas, and Orion—run on the identical TTS-1 or TTS-1-HD engine. This means your app experiences consistent latency and audio quality regardless of the speaker you select. There is no technical performance gap between the warm narrator and the deep documentary voice.
The pricing structure is uniformly applied across all models. You pay $0.01 per 1,000 characters, regardless of whether you choose a female or male preset. The rate limit of 60 requests per minute applies equally to every voice, ensuring predictable scaling for your international releases.
All voices use same engine
Every voice option utilizes the same underlying synthesis model, ensuring uniform audio fidelity. Switching from Aria to Orion does not alter the processing time or the structural quality of the generated speech file.
Same price per character
Your billing remains consistent regardless of voice selection. Whether generating content in Arabic, Hindi, or Spanish, the cost is strictly based on character count at $10 per million characters. No voice commands a premium.
Same rate limits
Each API key allows 60 requests per minute, applicable to all voice endpoints. This uniform limit simplifies capacity planning for your mobile app, whether you are broadcasting news updates or playing interactive stories.
Note that failed requests do not incur charges, providing a safety net during initial integration testing.
- 60 requests/minute per key
- Unlimited voices per request
- Failed requests are free
User Preference Trends
Voice selection in mobile apps often follows distinct patterns based on content type and user interaction. Understanding these trends helps teams configure their default settings effectively before launching in new markets.
These preferences are not hard rules but observed tendencies that guide initial voice assignments for different app categories.
Female voices for chat apps
Chat applications benefit from warmer, more engaging tones. Aria’s clear narration or Nova’s upbeat delivery creates a pleasant interaction loop, encouraging users to read more messages without feeling fatigued by robotic sounds.
Luna’s soft storytelling quality also works well for narrative-driven chat experiences, particularly in language learning or guided conversation apps.
- Aria for clear, direct messaging
- Nova for upbeat, quick interactions
- Luna for narrative chats
Male voices for news updates
News and update features often default to Atlas or Milo. Atlas provides a neutral, trustworthy tone suitable for factual reporting, while Milo offers a friendly conversational style for personalized daily digests.
The deep clarity of Orion is also occasionally used for serious news segments or documentary-style audio briefings in news apps.
- Atlas for neutral reporting
- Milo for conversational updates
- Orion for serious segments
Neutral voices for general use
For apps with mixed content or no specific persona, neutral voices like Atlas or Sage provide a safe, non-distracting baseline. This avoids imposing a specific gender or emotional tone on diverse user bases.
Sage’s calm, instructional quality is particularly effective for utility apps where clarity matters more than personality.
- Atlas for general neutrality
- Sage for utility/instructional apps
Licensing Details
LuminaVoice generates synthetic speech using audio samples derived from the LibriTTS-R dataset. This foundation ensures consistent voice characteristics across all eight presets while maintaining high audio fidelity for mobile applications.
Based on LibriTTS-R recordings
Our voices are synthesized from the LibriTTS-R dataset, which provides a robust training foundation for natural-sounding speech. This dataset enables precise control over tone and clarity, ensuring your app delivers professional audio without the inconsistencies of raw recordings.
- Derived from high-quality LibriTTS-R samples
- Ensures consistent voice characteristics
- Supports clear and natural speech patterns
CC BY 4.0 license
The underlying audio data is licensed under CC BY 4.0, allowing broad usage rights for commercial and non-commercial projects. This license covers the source material used to train the synthetic voices, simplifying legal considerations for mobile developers.
- CC BY 4.0 license applies
- Covers source training material
- Suitable for commercial use
Public domain source material
Additional voice samples originate from LibriVox public-domain audiobook readings. These recordings are free of copyright restrictions, further reducing licensing overhead. You can integrate these voices into your app without worrying about complex royalty structures or ongoing fees.
- LibriVox public-domain source
- No copyright restrictions
- Reduces licensing overhead
Audio Quality Factors
Every generated audio file contains an inaudible watermark that marks it as synthetic speech. This allows downstream platforms to identify the content as AI-generated without affecting playback quality or user experience. The watermark is embedded during synthesis and persists across all supported formats, ensuring consistent traceability for mobile apps that distribute audio content.
Synthetic speech indicators are inherent to the engine’s design. While the voices are based on licensed recordings from LibriVox, the output is distinct from human recording. Listeners may notice slight uniformity in pacing, but the clarity remains high for mobile interfaces. This consistency is beneficial for navigation and accessibility features where predictable audio behavior matters more than human nuance.
Watermark presence
The watermark is inaudible and does not degrade the listening experience. It serves as a metadata marker for content management systems. Mobile apps can use this to tag AI-generated audio in their libraries, helping with compliance and transparency with end-users.
- Inaudible to the human ear
- Persists across all output formats
- Useful for content tracking
Synthetic speech indicators
The voices lack human micro-variations like breath sounds or slight pitch drift. This uniformity ensures consistent playback across different devices and network conditions. For mobile apps, this predictability reduces buffering issues and ensures smooth audio transitions between UI elements.
Unlike human recordings, synthetic speech does not carry emotional context beyond the preset tone. This is ideal for instructional or news content where clarity is prioritized over expressive nuance. Apps relying on voice navigation benefit from this consistent, neutral delivery.
Format impact on quality
The choice of format (mp3, wav, opus, flac, pcm) affects file size and bandwidth usage, not the underlying voice quality. MP3 is recommended for mobile apps to save data and battery life. WAV and FLAC provide uncompressed quality but increase download times. PCM is suitable for real-time processing but consumes more memory. Opus offers a balance of quality and efficiency.
All voices use the same engine, so format choice is purely a technical decision based on your app’s bandwidth constraints. Test different formats on target devices to ensure smooth playback on older mobile hardware.
- MP3: Recommended for mobile data efficiency
- WAV/FLAC: Uncompressed, higher quality
- Opus: Efficient for streaming
- PCM: Low latency processing
Cost Efficiency
Selecting a voice for your mobile app does not impact your billing. Every preset, from Aria to Orion, costs exactly $10 per million characters. You can switch voices between releases or A/B test different narrators without incurring higher fees. This uniformity simplifies budget forecasting for international launches.
No difference in price between voices
All eight voices share the same pricing tier. Whether you choose the warm narrator Aria or the deep documentary style of Orion, the cost per character remains identical. This allows you to optimize for user preference without worrying about variable costs across different voice types.
Same character counts apply
Character counting is consistent across all requests. Spaces, punctuation, and text length up to 4096 characters are billed uniformly. Failed requests are free, so you can test edge cases without penalty. This predictability helps mobile teams estimate costs accurately for large-scale text inputs.
No extra cost for HD model
The "tts-1-hd" model uses the same engine and price as the standard "tts-1". There is no premium fee for higher fidelity. You can upgrade to HD for better audio quality without increasing your per-character spend, making it easy to maintain quality across all languages.
Implementation Tips
Selecting the right voice requires more than reading a description; it demands testing with your actual app content. Send short text snippets via the API to hear how each preset handles your specific phrasing, punctuation, and pacing. This ensures the audio fits naturally within your user interface without jarring transitions.
Test voices with your specific text
Use the API to generate audio from your exact copy. Check how the engine handles numbers, abbreviations, and complex sentences. This step reveals whether a voice sounds natural with your unique terminology before you commit to a production rollout.
- Test with 50–100 character snippets first.
- Check punctuation handling in your native language.
- Verify speed settings for your target audience.
Consider listener demographics
Match voice characteristics to your user base. Younger audiences often prefer brighter tones like Nova, while professional contexts may suit the calm clarity of Sage or the neutral delivery of Atlas. Aligning voice with demographic expectations improves user retention and engagement.
- Analyze your app's primary user group.
- Review competitor apps for tone standards.
- Ensure voice matches app branding.
A/B test before full rollout
Run controlled experiments with two different voices for the same content. Measure user interaction rates, completion times, or feedback scores. Use these concrete metrics to decide which preset delivers the best experience for your specific mobile application.
- Split traffic 50/50 between Voice A and Voice B.
- Track user retention and session duration.
- Gather direct user feedback via in-app surveys.
Common Misconceptions
Developers often assume synthetic voices mimic famous actors or allow precise emotional control. In reality, LuminaVoice provides eight distinct presets derived from licensed recordings, not celebrity likenesses. The system does not include emotion tags to modulate tone dynamically.
Furthermore, the API does not support SSML (Speech Synthesis Markup Language). You cannot use XML tags to adjust pauses, emphasis, or pronunciation within your text strings. This keeps the implementation simple but limits granular control over speech cadence.
All generated audio files contain an inaudible watermark identifying them as synthetic speech. This is a standard feature for transparency, not a defect in quality.
Voices are not celebrity clones
The eight presets are based on licensed recordings from LibriTTS-R (CC BY 4.0). They do not imitate real people, celebrities, or politicians.
Custom voice cloning is not offered. You cannot upload your own voice or design a custom persona.
- Based on public-domain audiobook readings
- No celebrity imitation
No emotion tags available
The engine does not support emotion tags. You cannot dynamically adjust tone, sadness, or excitement via API parameters.
Voice characteristics are fixed per preset (e.g., "bright" for Nova, "calm" for Sage).
SSML is not supported
The API does not accept SSML tags for breaks, emphasis, or pronunciation. Input is plain text up to 4096 characters.
For specific pronunciation needs, you must rely on the standard text input or set the language parameter explicitly.
- Plain text input only
- No XML markup support
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. Craft intuitive mobile applications that captivate early adopters and drive rapid user acquisition.
23 languages
Any voice speaks any of these languages.
