Frequently asked questions
Which voice is the most realistic?
Realism is subjective. All eight voices use licensed recordings from LibriTTS-R and LibriVox. Each file carries an inaudible watermark marking it as synthetic. Choose based on your stream’s tone: Aria for warm narration, Felix for energetic promos, or Orion for deep documentary styles. There is no single "most realistic" voice.
Can I mix voices in a single stream?
Yes. Since each request generates a separate audio file, you can mix voices by concatenating outputs. Split texts into parts under 4096 characters, generate each with a different voice, and join the files. This works for any voice combination like Aria, Nova, or Atlas within your stream workflow.
Do all voices support all 23 languages?
Yes. Any of the eight voices can speak any of the 23 supported languages. If you do not set the language, the system guesses it from the script. For non-English Latin text, explicitly set the language parameter to ensure accurate pronunciation and avoid defaulting to English.
What is the 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. Use "tts-1" or "tts-1-hd" interchangeably. Both support all eight voices, all formats, and the same speed range. There is no quality or cost difference between the two model strings.
Are the voices suitable for children's content?
Yes. Voices like Luna (soft, storytelling) and Nova (bright, upbeat) are well-suited for children's content. All voices are synthetic and carry an inaudible watermark. They are based on licensed recordings, making them safe for commercial use without additional licensing fees beyond the API usage cost.
Can I use these voices in paid advertisements?
Yes. The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain readings. You can use them in paid advertisements, streams, and content. The generated audio is returned in the response and not stored, ensuring you own the output file for your commercial projects.
Do the voices have accents?
The voices are designed to be clear and neutral within their supported languages. While specific regional accents are not a primary feature, the voices handle the 23 supported languages accurately. Set the language parameter explicitly to ensure the correct pronunciation for your target audience without unintended accent variations.
How do I choose the right speed?
Adjust speed between 0.25 and 4.0 to match your stream’s pace. Faster speeds suit energetic promos with Felix, while slower speeds work for Luna’s storytelling. Typical generation takes roughly 25 characters per second. Test different speeds to ensure clarity and emotional impact for your specific content type.
Is there a limit to voice changes?
No. You can change voices per request. Each API call allows a different voice selection. There is no limit to how often you switch between Aria, Nova, Luna, Sage, Felix, Milo, Atlas, or Orion. Just ensure each text input stays under 4096 characters per request.
Do I need to pay extra for HD voices?
No. Both "tts-1" and "tts-1-hd" models cost $10 per 1 million characters. There is no price difference between the standard and HD models. You pay based on character count, regardless of which model identifier you use in your API request. The engine and quality are identical.
Can I download the voice models?
No. You receive generated audio files (mp3, wav, opus, flac, pcm), not the voice models themselves. The voices are based on licensed recordings but are not available as standalone downloadable models. You can only download the synthesized audio output from each API request.
Are the voices customizable?
No. You can adjust speed (0.25–4.0) and select from eight presets. There is no SSML, emotion tags, phoneme dictionaries, or custom pronunciation lexicons. You cannot upload your own voice or clone voices. Each voice has a fixed tone, such as Felix’s energetic promo style or Orion’s deep documentary voice.
How long do the generated files last?
Generated audio files are returned in the response and not stored by the API. You must save them immediately after generation. Input text is used only to produce audio and is not retained for training. Only request metadata (time, characters, status) is kept for billing purposes.
Can I use the voices offline?
No. The voices are generated via a cloud API. You send text and receive an audio file. There is no offline SDK or local model available. You need an internet connection to make requests to the endpoint and receive the synthesized audio files for your streams.
Do the voices sound natural in noise?
The voices are clear and designed for clarity. However, background noise in your stream can affect perception. Use high-quality formats like WAV or FLAC for best results. The voices themselves do not have built-in noise cancellation. Ensure your audio output is clean for the best listener experience.
Is there a voice for horror streams?
Orion (deep, documentary) or Sage (calm, instructional) can suit horror tones. Orion’s deep voice works well for atmospheric narration. Sage’s calm style can create tension. While not explicitly labeled "horror," these voices provide the depth and seriousness suitable for horror content when paired with appropriate speed and format.
Can I combine multiple voices?
Yes. Generate separate audio files for each voice segment and join them. Split texts into parts under 4096 characters. Use different voices for each part, then concatenate the resulting audio files. This allows dynamic voice switching within a single stream or video without real-time streaming limitations.
How do I get the best quality audio?
Use WAV or FLAC formats for lossless quality. MP3 and Opus are compressed. Set speed between 0.25 and 4.0 for clarity. Ensure text is under 4096 characters. Failed requests are free, so test samples before generating long texts. Use the official OpenAI SDK or direct HTTP requests for reliability.
What is the best format for streaming?
MP3 is widely supported and small. WAV and FLAC offer lossless quality. Opus is efficient for web streaming. PCM is raw data. Choose based on your streaming platform’s requirements. MP3 is generally sufficient for most streamers, while FLAC is better for archival or high-fidelity needs.
Do I need to attribute the voices?
The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox. While the API output is synthetic, attribution to the source recordings may be required depending on your use case. Check the specific license terms for LibriTTS-R and LibriVox to ensure compliance with your content distribution.
Methodology: How We Evaluate Voices
We evaluate the eight synthetic voices using three core criteria relevant to streamers: clarity under load, energy consistency, and long-session durability. Each voice is tested against standard gaming commentary scripts to ensure intelligibility and engagement without vocal fatigue.
Clarity and Articulation Standards
Clarity is measured by character-to-second generation speed and articulation. We verify that consonants remain distinct even at high speeds, ensuring viewer comprehension during fast-paced gameplay moments.
Energy and Engagement Metrics
Energy is assessed by the preset's tonal character and pacing. We test each voice to ensure it maintains appropriate intensity for hype moments without sounding monotonous or overly artificial.
Durability for Long Sessions
Durability is evaluated by listening to generated audio over extended periods. We check for listener fatigue, ensuring the voice remains pleasant and clear during hour-long streams without becoming grating.
Voice 1: Aria (Warm Narrator)
Aria is defined by her warm, clear narrator quality, making her an ideal choice for streamers who want a comforting presence without sounding artificial. She excels at maintaining listener engagement during long streams, offering a consistent vocal baseline that doesn’t fatigue the audience over hours of gameplay.
Ideal Use Cases
Aria handles standard narration and strategic explanations with high clarity. Her steady pacing works well for background commentary during gameplay, ensuring the voice remains present but not intrusive. She is particularly effective for content requiring a steady, reassuring tone, such as walkthroughs or relaxed gaming sessions.
- Standard gameplay commentary
- Strategy explanations
- Background narration
- Relaxed gaming sessions
Tone Analysis
The vocal profile of Aria is smooth and approachable. She avoids the robotic edge common in TTS by using natural inflections that mimic a human narrator. This warmth helps bridge the gap between synthetic audio and human connection, making her suitable for audiences who prefer a gentle auditory experience.
- Smooth vocal profile
- Natural inflections
- Gentle auditory experience
- Approachable tone
Best For Gaming Commentary
For streamers who need a reliable voice to fill silence or explain strategies, Aria offers consistency. Her clear articulation ensures that instructions are understood without requiring repeated listening. This reliability makes her a solid choice for maintaining flow during complex gameplay moments.
Voice 2: Nova (Bright Upbeat)
Nova delivers a bright, upbeat tone ideal for energetic gaming streams. Her clear articulation and lively pacing make her suitable for dynamic commentary where listener engagement is key.
Ideal Use Cases
Nova excels in fast-paced segments like live gameplay commentary or quick intro segments. Her energy helps maintain viewer attention during high-action moments.
- Live gameplay commentary
- Quick intro segments
- Event announcements
Tone Analysis
She combines warmth with clarity, avoiding the monotony of standard narrators. Her voice remains crisp even at higher speeds, ensuring no loss of intelligibility during rapid delivery.
Best For High Energy Streams
Use Nova for highlight reels or competitive match updates where momentum matters. Her style supports the excitement of gaming without sounding artificial or overly dramatic.
- Highlight reels
- Competitive match updates
- Viewer engagement segments
Voice 3: Luna (Soft Storyteller)
Luna delivers a soft, storytelling voice that excels in immersive content. Her warm tone reduces listener fatigue during extended streaming sessions, making her ideal for narrative-heavy segments where clarity and emotional nuance matter more than high energy.
As one of eight available presets, Luna supports all 23 languages without requiring specific voice changes. She is generated via the tts-1 model, ensuring consistent quality and predictable costs for your voiceover pipeline.
Ideal Use Cases
Luna is best suited for ambient background narration, lore explanations, and atmospheric intros. Her gentle pacing helps maintain viewer engagement without demanding active attention, allowing streamers to focus on gameplay or chat interaction.
- Background narration for gameplay
- Atmospheric intro sequences
- Lore and world-building explanations
- Relaxing stream ambiance
Tone Analysis
Luna’s vocal profile is characterized by a calm, inviting delivery. Unlike the bright energy of Nova or the neutral tone of Atlas, Luna’s voice carries a subtle warmth that feels personal yet professional. This makes her distinct among the eight presets for content requiring emotional depth without dramatic intensity.
- Warm and inviting delivery
- Calm pacing reduces fatigue
- Subtle emotional nuance
- Professional yet personal feel
Best For Lore and Intro Videos
For streamers crafting narrative-driven content, Luna’s voice ensures that lore is delivered with clarity and emotional resonance. Her tone complements visual storytelling, helping viewers connect with the game’s world. Combined with the API’s 4096-character limit, you can generate substantial intro scripts in a single request, streamlining your production workflow.
- Narrative-driven content
- Clear lore delivery
- Complements visual storytelling
- Efficient script generation
Voice 4: Sage (Calm Instructor)
Sage is designed as a calm, instructional voice, making it the optimal choice for structured content. Its clear articulation ensures that complex gaming mechanics or patch notes are delivered without listener fatigue during long tutorial sessions.
Ideal Use Cases
Use Sage for step-by-step walkthroughs, update logs, or educational streams where clarity outweighs emotional flair. Its steady pace helps viewers retain information without feeling rushed or overwhelmed by high-energy delivery.
- Game update announcements
- Mechanic explanations
- Strategy guides
- Patch note readings
Tone Analysis
Sage delivers a neutral, professional tone with consistent pacing. Unlike Aria’s warmth or Felix’s energy, Sage prioritizes intelligibility and calmness, reducing cognitive load for audiences processing detailed instructions.
Best For Tutorials and Guides
For content creators producing how-to videos or live tutorial streams, Sage’s consistent delivery ensures viewers focus on the content rather than the voice’s emotional cadence. Pair with 1.0x speed for natural pacing.
Voice 5: Felix (Energetic Promo)
Felix is designed for high-impact promotional content, delivering an energetic and punchy delivery that grabs attention immediately. His vocal profile cuts through background noise effectively, making him ideal for short-form clips where clarity and energy are paramount.
As a male preset based on licensed recordings, Felix maintains a consistent, professional tone without sounding artificially stiff. He excels at driving momentum in concise audio segments, ensuring the listener stays engaged throughout the clip.
Ideal Use Cases
Use Felix for trailer voiceovers, sponsor reads, or quick channel updates. His energetic cadence works well for announcements that need to convey urgency or excitement without requiring extensive post-production.
He is particularly effective for promotional segments where brevity is key. Since requests are limited to 4096 characters, Felix’s style ensures every word counts, avoiding filler while maintaining a lively pace.
- Promotional trailers and teasers
- Sponsor advertisements
- Urgent channel announcements
- Short-form social media clips
Tone Analysis
Felix exhibits a high-energy, upbeat delivery that feels modern and direct. The voice avoids excessive dramatization, keeping the tone professional yet lively. This balance prevents listener fatigue during repeated plays.
The articulation is crisp, supporting clear enunciation of consonants and numbers. This makes him suitable for detailed promo scripts where specific dates, codes, or prices must be understood instantly.
Best For Highlight Reels
For highlight reels, Felix’s energetic tone complements fast-paced visual edits. His voice can bridge gaps between gameplay clips, providing context or commentary that enhances the viewing experience.
Use the speed parameter (up to 4.0) to match the rhythm of quick cuts. Felix handles increased speeds well, maintaining intelligibility even when pushed beyond normal real-time generation speeds.
Voice 6: Milo (Friendly Conversational)
Milo brings a natural, approachable warmth to automated voiceovers, making him ideal for streamers who want their content to feel like a one-on-one conversation rather than a broadcast. His friendly cadence helps maintain viewer connection during long streams without sounding robotic or overly scripted.
Ideal Use Cases
Milo excels in casual commentary, reaction videos, and interactive segments where a relaxed tone is key. He works well for reading chat messages, introducing sponsors, or providing ambient voiceover during gameplay that doesn’t require high intensity.
- Chat message readbacks
- Sponsor segment introductions
- Casual gameplay commentary
- Ambient background narration
Tone Analysis
Milo’s voice is grounded in a conversational style, avoiding the high energy of Felix or the deep tone of Orion. He maintains clarity while sounding like a peer rather than a presenter, which reduces listener fatigue during extended sessions.
- Natural pacing
- Approachable timbre
- Low emotional variance
- High intelligibility
Best For Chat Interaction
When automating responses to viewer messages or donations, Milo’s tone feels personal and engaging. His conversational quality makes synthetic speech feel more like a live host interacting with the community, enhancing retention during interactive streams.
- Automated donation thanks
- Viewer message responses
- Interactive Q&A segments
- Community highlight recaps

Voice 7: Atlas (Neutral News)
Atlas serves as the reliable backbone for information-heavy streams. Its neutral, news-style delivery ensures that critical updates are delivered without emotional bias or distracting tonal shifts, making it ideal for keeping viewers informed during gameplay.
Ideal Use Cases
Use Atlas for reading patch notes, server maintenance schedules, or sponsor reads where clarity outweighs personality. The voice avoids the fatigue of high-energy tones during long streams, maintaining a consistent baseline for informational content.
- Server maintenance announcements
- Sponsor integration reads
- Real-time score updates
- Rule clarifications
Tone Analysis
Atlas delivers a steady, professional cadence with precise articulation. It lacks the warmth of Aria or the energy of Felix, offering instead a neutral, authoritative tone that prioritizes information transfer.
Best For Updates and Announcements
Deploy Atlas for time-sensitive stream updates or community news. Its neutral profile ensures the message is received clearly, whether you are announcing a tournament schedule or reading a donation chain without emotional interference.
Pair Atlas with the tts-1 model for standard clarity or tts-1-hd if audio quality is paramount for recorded VODs. The voice works effectively across all 23 supported languages, adapting naturally to local dialects when the language code is specified.
Voice 8: Orion (Deep Documentary)
Orion delivers a deep, resonant voice characterized by a documentary-style delivery. Its low frequency provides a sense of gravity and authority, making it distinct from the brighter presets. This vocal profile is engineered for immersion rather than high-energy engagement, offering a steady, consistent rhythm that holds attention during longer segments.
Ideal Use Cases
Orion excels in narrative-driven content where tone sets the scene. It is particularly effective for lore introductions, cinematic cutscene voiceovers, and atmospheric background narration. The voice maintains clarity even at lower speeds, ensuring that dense narrative text remains comprehensible without sounding monotonous.
- Lore and backstory narration
- Cinematic game intros
- Atmospheric environmental commentary
Tone Analysis
The vocal range is consistently deep, providing a neutral yet commanding presence. Unlike Felix's energetic promo style, Orion avoids sharp inflections, favoring a smooth, steady cadence. This makes it ideal for conveying serious or epic tones without requiring additional sound design to establish mood.
Best For Cinematic Moments
Streamers using Orion for highlight reels or dramatic moments benefit from its ability to underscore tension. Its documentary style pairs well with slower-paced editing. The voice remains clear up to the 4096-character limit, ensuring long cinematic scripts retain their intended gravity throughout generation.
Technical Comparison: Speed and Format
Generation speed averages roughly real-time, producing about 25 characters of text per second of audio for long-form content. Short phrases may take a few seconds due to processing overhead, while extended texts are split into 4096-character chunks and stitched together seamlessly.
Under heavy load, requests queue sequentially, so generation times can increase. Failed requests incur no charge, ensuring you only pay for successful synthesis.
Real-Time Generation Speed
For standard playback, the API generates audio at approximately real-time speeds. This means one second of audio requires roughly 25 characters of input text.
Short phrases may experience slight delays as the engine initializes, while longer texts are automatically split and joined. Requests are processed sequentially, meaning high traffic can introduce queuing delays.
- Average speed: ~25 chars per second of audio
- Max input: 4096 characters per request
- Queued processing under load
MP3 vs WAV Quality
MP3 offers efficient file sizes suitable for streaming and web embedding, while WAV provides uncompressed, lossless audio for high-fidelity editing. Both formats maintain the same voice quality generated by the model.
Choose WAV if you plan to apply additional audio effects or need maximum fidelity for post-production. MP3 is optimal for direct integration into stream overlays or VODs.
- MP3: Efficient, web-ready
- WAV: Lossless, editing-friendly
- Other formats: OPUS, FLAC, PCM
Speed Adjustment Effects
The speed parameter allows adjustment from 0.25x to 4.0x. Slowing down improves clarity for complex lore, while speeding up suits fast-paced highlight reels.
Extreme speeds may slightly alter natural intonation but preserve the core voice character. Adjustments are applied during generation, so you do not need external tools.
- Range: 0.25x to 4.0x
- Ideal for pacing control
- No external editing needed
Language Support Across Voices
VoxHype’s API supports 23 languages, including English, Spanish, German, Japanese, and Arabic. All eight synthetic voices can speak any of these languages without restriction, allowing streamers to localize content or engage diverse audiences without switching voice models.
The system guesses the language from the text script if you do not specify one. Latin script defaults to English, which is convenient for English-first streams but requires manual adjustment for mixed-language content.
23 Languages Available
The platform supports Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. This range covers most major gaming markets, enabling VTubers to produce localized intros or commentary in their audience's native tongue.
Default Language Behavior
If the language field is omitted, the API infers the language from the input text. For Latin-script text, it defaults to English. This simplifies requests for monolingual streams but may cause mispronunciation if you feed it non-English text without the explicit tag.
Manual Language Setting Benefits
Explicitly setting the language parameter ensures accurate pronunciation, especially for non-English Latin scripts like Spanish or German. It prevents the engine from forcing an English accent on foreign words, preserving the intended tone of your stream’s voiceover.
Pricing Impact on Voice Choice
Voice selection is driven by character count, not duration. At $0.01 per 1,000 characters, longer scripts cost more regardless of the voice used. Every voice supports speeds from 0.25x to 4.0x, allowing creators to compress audio and reduce costs for fast-paced commentary.
Cost Per Minute of Audio
Audio length varies with speed settings. A 4,000-character script at 1.0x speed generates roughly 160 seconds of audio. Increasing speed to 2.0x halves the duration but does not change the character-based charge. Creators should estimate character counts to predict expenses accurately.
- Rate: $10 per 1 million characters
- Characters include spaces and punctuation
- Speed adjusts duration, not price
HD vs Standard Model Pricing
Both "tts-1" and "tts-1-hd" models share the same pricing structure. The HD model offers higher fidelity but costs the same per character. Streamers can choose the standard model to save processing time during high-volume requests without incurring extra fees.
- Both models cost $0.01/1k chars
- HD provides higher audio quality
- No price difference between models
Trial Credit Usage
New Google sign-ups receive $0.50 in trial credit, valid for 7 days. This covers approximately 50,000 characters, enough for testing voice clarity or generating short intro clips. Email-based accounts start without this credit and require an immediate top-up to begin generating audio.
- Trial: $0.50 for Google sign-ups
- Valid for 7 days
- Covers ~50,000 characters
Final Rankings for Streamers
Choosing the right voice depends on your stream’s specific needs. Aria offers a balanced performance across commentary and general use, providing a warm tone that is engaging without being overwhelming.
For high-energy moments, Nova delivers bright, upbeat clarity perfect for hype segments. If clarity is your priority, Aria remains the top choice for consistent articulation during long gaming sessions.
Best Overall Voice
Aria is a versatile option for streamers who need a reliable narrator voice. Its warm, clear tone works well for both casual commentary and structured segments, making it a safe default choice for most content types.
- Warm, clear narrator tone
- Versatile for commentary and intros
- Consistent articulation over long sessions
Best for Energy
Nova is a strong choice for high-energy streams. Its bright and upbeat character brings immediate engagement to highlight reels, quick reactions, and hype moments, ensuring your audience stays alert and entertained.
- Bright, upbeat character
- Ideal for highlight reels
- Enhances viewer engagement in fast-paced content
Best for Clarity
Aria provides clear articulation that ensures every word is understood. This makes it ideal for detailed explanations or when you need your audience to catch every detail without straining to listen.
- Precise articulation
- Ideal for detailed explanations
- Reduces listener fatigue during long streams
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. We check for listener fatigue, ensuring the voice remains pleasant and clear during hour-long streams without becoming grating.
23 languages
Any voice speaks any of these languages.
