Frequently asked questions
Which voice is best for news?
Atlas is our neutral, news-oriented voice, but any preset works for news. Choose based on your preferred tone: Aria offers warm clarity, while Milo provides a friendly, conversational style. All voices support all 23 languages, so you can select the voice that fits your specific editorial style.
Can I use multiple voices?
Yes, you can use any of the eight voices in your requests. Since each request generates a separate audio file, you can split longer articles into parts and assign different voices to different sections. You just need to concatenate the resulting audio files to create a multi-voice broadcast.
How do I choose a voice?
Select a voice based on the desired tone: Aria for clear narration, Nova for upbeat content, Luna for storytelling, Sage for instruction, Felix for energetic promos, Milo for conversation, Atlas for neutral news, or Orion for deep documentary styles. All voices support all 23 languages, so choose based on mood rather than language constraints.
Is Atlas good for hard news?
Atlas is specifically described as neutral and suitable for news delivery. It provides a standard, clear tone appropriate for hard news without emotional bias. You can adjust the speed between 0.25 and 4.0 to match your broadcast pace while maintaining clarity for your audience.
Does Orion sound too deep?
Orion is described as deep and suited for documentary styles. If you find the depth too heavy for your specific news format, you can adjust the speed parameter between 0.25 and 4.0 to alter the perceived pitch and pacing. Alternatively, consider using Atlas or Sage for a more standard news tone.
Can Milo read interviews?
Milo is described as friendly and conversational, making it suitable for interview-style content. You can generate audio for each speaker using the same voice or switch voices to differentiate speakers. Since each request produces a separate file, you can edit and join the clips to simulate a dialogue.
Which voice is calmest?
Sage is described as calm and instructional, while Luna is soft and storytelling-oriented. Both are suitable for calm delivery. Sage is ideal for clear, measured news updates, whereas Luna works well for softer, narrative-driven segments. You can also reduce the speed to enhance the calm effect.
How does speed affect quality?
Speed ranges from 0.25 to 4.0 and affects playback rate, not synthesis quality. The engine uses the same underlying model regardless of speed. Adjusting speed allows you to fit more content into a broadcast or slow down complex segments for clarity. The character limit remains 4096 characters per request.
What format is best for web?
MP3 offers broad player support and efficient compression. OPUS provides even better compression for static file delivery, though our API returns complete files rather than supporting live streaming. WAV and FLAC ensure lossless quality with larger sizes, while PCM is raw audio. For your news platform, choose MP3 or OPUS for the best balance of compatibility and file size.
Do voices sound different in languages?
No, the same voice preset is used across all 23 supported languages. The voice characteristics remain consistent regardless of the language spoken. You can set the language explicitly or let the system guess it from the script, but the voice timbre and style do not change between languages.
Can I change voices mid-article?
Yes, by splitting the article into multiple requests. Each request generates a separate audio file. You can assign a different voice to each segment in your code. Then, join the audio files on your server or client side to create a single continuous audio file with varied voices.
Is there a voice for kids news?
There is no specific kids voice, but Nova is bright and upbeat, which may suit younger audiences. Luna is soft and storytelling-oriented, which could also work for children's content. You can adjust the speed to be slower and clearer. All voices support all 23 languages for diverse audiences.
How do I test voices?
Send a POST request to the speech endpoint with your text, desired voice, and language. The API returns the audio file directly. You can also check the /v1/voices endpoint to see all available voices, languages, and formats. Use the free trial credit to test without immediate cost.
Neutral News Voice
Choosing the right voice is critical for digital news platforms where clarity and trust are paramount. Curisum Audio offers eight distinct presets, each designed to suit specific editorial tones. For hard news, neutrality and articulation are often prioritized over expressive flair to ensure the information lands clearly without distracting the listener with unnecessary emotional cues.
Atlas voice description
Atlas is designed as a neutral, news-oriented voice. It delivers text with a steady, clear cadence that avoids excessive emotional inflection. This makes it ideal for reading straight news reports, press releases, or factual updates where the content itself should carry the weight, not the narrator’s personality.
Suitability for hard news
Atlas works best for segments where accuracy and speed are key. Its consistent tone helps listeners focus on the facts rather than the delivery style. It pairs well with ISO language codes to ensure proper pronunciation in the 23 supported languages, from English to Arabic and beyond.
Listener retention rates
While Atlas provides a clear, distraction-free listening experience, specific data on listener retention rates or cognitive load improvements are not tracked or published by Curisum. The voice is optimized for clarity, allowing your audience to consume news efficiently without invented metrics to guarantee engagement levels.
Documentary Style Voice
Orion brings a deep, documentary-style presence to your audio content. Its grounded tone adds authority to complex narratives, making it ideal for long-form journalism where listener trust is paramount.
Orion voice description
Orion is designed for depth and gravitas. It delivers a consistent, low-frequency tone that commands attention without sounding artificial. This preset excels when you need to convey seriousness and reliability in your news segments.
Depth and gravitas
The voice maintains a steady pace that aids comprehension during dense reporting. Listeners often retain more information when the audio carries a sense of established authority, which Orion provides naturally.
- Deep, resonant tone
- Consistent pacing
- High clarity
Feature story applications
Use Orion for investigative pieces or historical context segments. Its documentary style suits longer texts where maintaining listener engagement over several minutes is critical. Pair it with slow speeds for maximum impact.
- Investigative journalism
- Historical context
- Long-form features

Conversational Tone
Milo offers a friendly, conversational tone ideal for digital news platforms aiming for approachability. As a male preset, it delivers clear articulation without the formality of hard news, making it suitable for casual segments or interview-style readings where a relaxed atmosphere is desired.
Milo voice description
Milo is designed for a natural, conversational flow. It avoids the stiffness of traditional broadcast styles, providing a warm and engaging listening experience that feels personal rather than institutional. This voice preset is one of eight available options, each tuned for specific tonal needs.
- Friendly and conversational male voice
- Suitable for casual news segments
- Clear articulation for readability
Interview style reading
When reading transcripts of interviews, Milo’s tone helps maintain listener engagement during Q&A sessions. Its natural pacing aids comprehension, though it does not include built-in pause controls or SSML support. Developers should include explicit punctuation in the input text to indicate desired breaks in speech.
Casual news segments
For lighter news topics or daily updates, Milo provides a relaxed delivery that reduces listener fatigue. It pairs well with shorter texts under 4096 characters. Since requests are processed sequentially, longer casual segments may experience slight delays under high load, so plan accordingly for real-time expectations.
Instructional Voice
Sage is the calm, instructional voice designed for clarity and precision. Its steady cadence makes it ideal for explaining complex news topics or delivering educational segments where listener comprehension is priority over emotional flair.
Sage voice description
The voice offers a neutral, professional tone that reduces listener fatigue during long reads. It avoids dramatic inflections, ensuring the focus remains on the information being delivered.
Educational content
Use Sage for breaking down technical details or summarizing policy changes. Its consistent pacing helps audiences absorb dense information without distraction, making it perfect for explainer segments in digital news.
Clear articulation
Each word is enunciated clearly, reducing ambiguity in names or technical terms. This clarity ensures your audience retains the core message, even when listening on mobile devices with background noise.
Warm Narrator
Curisum Audio’s Aria voice brings a warm, clear narrative style ideal for human-interest stories and soft news features. Her tone is approachable, helping listeners connect with subject matter on a personal level without requiring complex infrastructure. This makes her a reliable choice for indie developers aiming to deliver engaging audio content that feels natural and inviting.
As a warm narrator, Aria excels in long-form content where emotional resonance matters. Her clear articulation ensures that every word is understood, making her suitable for features that require a gentle, storytelling approach. This voice type helps maintain listener engagement while delivering nuanced content with ease.
Aria voice description
Aria is a female voice preset known for her warm, clear narration style. She is designed to sound approachable and natural, making her ideal for soft news and human-interest stories. Her tone is distinct from the neutral delivery of Atlas, adding a layer of warmth that helps listeners connect with the story’s subject matter on a personal level. This makes her a reliable choice for indie developers aiming to deliver engaging audio content that feels natural and inviting.
- Warm, clear narration
- Approachable tone
- Ideal for soft news
Soft news features
Aria is particularly well-suited for soft news features, where the focus is on human stories rather than hard facts. Her warm tone helps listeners connect with the subject matter on a personal level, making her a reliable choice for content that requires a gentle, storytelling approach. This voice type is perfect for features that require emotional resonance without the need for explicit emotion tags.
Unlike the neutral delivery of Atlas, Aria adds a layer of approachability that helps listeners connect with the story’s subject matter on a personal level. This makes her a reliable choice for indie developers aiming to deliver engaging audio content that feels natural and inviting.
- Ideal for soft news
- Engages listeners
- Natural storytelling
Emotional resonance
While Aria lacks explicit emotion tags, her natural inflection provides a subtle emotional depth suitable for soft news. This makes her a reliable choice for content that requires emotional resonance without the need for complex infrastructure. Her tone is distinct from the neutral delivery of Atlas, adding a layer of warmth that helps listeners connect with the story’s subject matter on a personal level. This makes her a reliable choice for indie developers aiming to deliver engaging audio content that feels natural and inviting.
Aria’s warm, clear narration style makes her ideal for human-interest stories and soft news features. Her tone is approachable, helping listeners connect with the subject matter on a personal level without requiring complex infrastructure. This makes her a reliable choice for indie developers aiming to deliver engaging audio content that feels natural and inviting.
- Subtle emotional depth
- Natural inflection
- No explicit emotion tags
Upbeat Announcer
Nova brings a bright, upbeat energy to your digital news platform. Designed for segments that require a positive tone, this female voice cuts through the noise with clarity and enthusiasm, making it ideal for keeping listeners engaged during daily updates.
Nova voice description
Nova is characterized by a bright and upbeat delivery style. As a female preset, she offers a distinct alternative to deeper male voices, providing a fresh auditory experience for recurring news segments. Her tone is consistently positive, which helps maintain listener interest without sounding monotonous.
- Female voice preset
- Bright and upbeat character
- Positive tonal quality
Breaking news energy
When delivering breaking news or quick updates, Nova’s energetic style ensures the information feels urgent yet approachable. The upbeat nature prevents the delivery from becoming too serious or dull, which is crucial for maintaining audience attention during time-sensitive reports.
Positive tone delivery
Nova excels in delivering content with a consistently positive tone. Whether reading weather updates, community highlights, or light news features, her bright delivery adds a layer of warmth and approachability that resonates well with general audiences.
Storytelling Voice
Luna brings a soft, storytelling quality to digital news, making her ideal for long-form content that requires a gentle, engaging tone. Her voice style helps maintain listener interest over extended periods, turning standard articles into immersive audio experiences.
Luna voice description
Luna is designed for soft news features and emotional resonance. Her clear, warm delivery ensures that complex narratives are accessible and engaging, helping listeners stay connected to the story without feeling overwhelmed by dense information.
Long-form content
Use Luna for deep dives, investigative reports, or historical overviews. Her steady pacing supports longer texts, keeping the audience engaged from start to finish. She handles nuanced storytelling better than more energetic presets, ensuring clarity even in detailed segments.
Engaging narrative
Luna excels at maintaining a narrative flow, making her perfect for feature stories that require emotional depth. Her voice adds a layer of intimacy to news pieces, helping listeners feel more connected to the subjects being discussed.
Energetic Promo
The Felix voice brings an energetic, promotional quality to your audio content, making it ideal for short, high-impact segments. Its distinct tone captures attention quickly, perfect for delivering concise updates or announcements that require immediate listener engagement.
This preset is particularly effective for time-sensitive news bits where energy and clarity are paramount. It helps maintain audience interest during brief transitions or promotional interludes within your digital news platform.
Felix voice description
Felix is designed to be energetic and promotional. His voice cuts through with clarity and drive, ensuring that your message is delivered with enthusiasm and precision. This makes him an excellent choice for segments that need to stand out.
Short segments
Use Felix for brief updates, weather alerts, or sponsor messages. His style is optimized for short bursts of text, ensuring that every word counts. This approach keeps your content dynamic and prevents listener fatigue during rapid-fire updates.
High energy delivery
The high energy of Felix’s voice helps convey urgency and importance. It’s perfect for breaking news snippets or special announcements where a lively tone can enhance the sense of immediacy and keep your audience engaged.

Language Selection
Curisum Audio supports 23 languages, including English, Spanish, and Japanese. You can specify the language via the language parameter to ensure correct pronunciation, especially for Latin-script texts that might otherwise default to English.
Our system supports cross-language capabilities, allowing any voice to speak any supported language. This flexibility lets you maintain a consistent audio brand across diverse regional editions of your news platform.
Matching voice to language
Select the language code (e.g., fr for French) to guide the synthesis engine. If omitted, Latin script defaults to English. For non-English Latin text, always set the language to avoid mispronunciation.
Cross-language capabilities
Any voice preset can pronounce any of the 23 supported languages. You can switch voices and languages between requests without needing separate models, maintaining audio consistency across different language editions of your news site.
Accent consistency
Voices are based on licensed recordings and maintain consistent accent profiles. For example, atlas provides a neutral tone suitable for news, while aria offers a warm narrator style. Consistency is key for listener trust.
Speed Adjustments
The speed parameter allows you to fine-tune playback between 0.25 and 4.0 times normal rate. For digital news, adjusting this setting helps match the pacing of your content, whether you are delivering breaking headlines or reading long-form features. You can set the language explicitly to ensure the AI interprets the text correctly before applying speed adjustments.
Optimal speed for news
Standard news reading typically benefits from a speed between 1.0 and 1.5. This range maintains clarity while ensuring efficient delivery of information. Since requests are processed sequentially, higher speeds may slightly reduce generation time, but the audio length scales inversely with the speed factor.
Speed range impact
At 0.25x, audio becomes slow and deliberate, suitable for complex data or foreign language segments. At 4.0x, the voice becomes rapid, potentially obscuring consonants. The API processes one request at a time, so extreme speeds do not significantly alter queue wait times compared to standard speeds.
- Minimum: 0.25x
- Maximum: 4.0x
- Standard: 1.0x
Listener comprehension
Comprehension depends on the listener's preference and the complexity of the text. There is no universal threshold where understanding drops; instead, clarity varies by individual. For hard news, keeping speed near 1.0–1.2x usually preserves articulation best. Always test with your specific audience to find the right balance.
Format Choices
Curisum Audio supports five output formats: mp3, wav, opus, flac, and pcm. Each format serves different technical needs for your news platform, from compact distribution to high-fidelity archival.
MP3 vs WAV
MP3 is the standard for web delivery, offering small file sizes that load quickly on mobile networks. WAV provides uncompressed, lossless audio, preserving every detail of the synthesized speech for archival purposes or professional editing.
- MP3: Efficient compression, ideal for general web playback.
- WAV: Lossless quality, best for editing or high-fidelity needs.
File size considerations
Choosing a format directly impacts your bandwidth and storage costs. MP3 and OPUS files are significantly smaller than WAV or FLAC, reducing data transfer expenses for your users. PCM files are raw and large, suitable only for specific processing pipelines.
- MP3 and OPUS minimize bandwidth usage.
- WAV and FLAC offer higher fidelity but larger files.
Streaming compatibility
All supported formats are delivered as standard audio files via the API response. You can integrate any of these formats into your existing audio players without needing specific codecs or plugins, as the API returns the raw audio data.
- MP3, WAV, OPUS, FLAC, and PCM are all direct file responses.
- No proprietary streaming protocol required; standard HTTP delivery.
Implementation Tips
Implementing Curisum Audio into a news platform requires managing text limits and API reliability. Since each request supports a maximum of 4096 characters, you must split longer articles into sequential segments. The API returns individual audio files for each part, which your infrastructure can concatenate server-side or client-side before playback.
Text chunking strategy
Split your input text into chunks of up to 4096 characters. Ensure boundaries respect sentence structures to avoid abrupt audio cuts. Process requests sequentially, as they are handled one after another, and join the resulting audio files to create a continuous stream for the listener.
- Keep chunks under 4096 characters
- Respect sentence boundaries for natural pauses
- Join audio files sequentially after generation
Error handling in playback
Monitor the API response status for each chunk. If synthesis fails, the charge is returned in full, so retrying is safe. Implement logic to pause playback or display a fallback text summary if a specific segment fails to generate, ensuring the user experience remains uninterrupted.
- Check response status codes
- Retry failed requests freely (refunded)
- Fallback to text if generation fails
User interface design
Design the player to handle dynamic audio sources. Since requests are processed sequentially, indicate loading states clearly. Allow users to adjust speed between 0.25 and 4.0 to suit their listening preference, enhancing accessibility for news consumers who prefer faster or slower delivery rates.
- Show clear loading indicators
- Expose speed controls (0.25–4.0)
- Support dynamic audio source updates
Cost Efficiency
Curisum Audio charges $0.01 per 1,000 characters, with spaces and punctuation counting toward the total. To optimize costs, split long articles into chunks of up to 4,096 characters. This ensures efficient processing and avoids unnecessary character waste while maintaining audio quality.
Failed requests are free, so you can test endpoints without financial risk. Use the free $0.50 trial credit (approx. 50,000 characters) to validate voice selection before committing to paid usage. Balance never expires, allowing flexible budgeting for irregular news cycles.
Character usage optimization
Since every character counts toward your $0.01 per 1,000 rate, precise input management matters. Use the 4,096-character limit per request to break down long articles efficiently. This approach keeps generation times predictable and costs aligned with actual content length.
- Count spaces and punctuation as characters
- Split texts at natural sentence boundaries
- Avoid redundant formatting tags
Batch processing benefits
Processing multiple segments sequentially allows you to manage queue times effectively. Since requests are processed one after another, batching helps distribute load. This ensures consistent audio delivery without overwhelming the API rate limit of 60 requests per minute.
- Group related article segments
- Monitor queue status during high volume
- Stagger large deployments
Budget planning
With no monthly plans or minimum spend, you pay only for what you use. Top-ups via USDT (TRC-20) are credited automatically within minutes. This model suits indie developers with variable traffic, ensuring you never pay for idle capacity.
Start with the $0.50 trial credit to estimate character usage for your typical article length. Adjust your voice and speed settings during this trial to understand character-to-audio ratios before scaling up.
- No expiration on balance
- Instant USDT top-ups
- Pay-per-character model
Technical Requirements
Curisum Audio’s API is designed for reliable integration into news workflows. Understanding technical constraints ensures smooth deployment for indie developers managing digital news platforms.
Bandwidth considerations
Audio files are returned directly in the response. For long articles, split text into 4096-character chunks to generate shorter clips, ensuring manageable data transfer without overwhelming your network.
Storage management
Generated audio is not stored by Curisum Audio; it is returned in the immediate response. You must handle local storage or streaming to your users. Request metadata is retained for billing only.
API rate limiting
Each API key supports 60 requests per minute. New accounts start with a limit of 10 requests per minute until the first top-up is processed. Plan your chunking strategy to stay within these bounds.

Future Scalability
As your news platform grows, Curisum Audio scales with your demand without requiring complex infrastructure changes. The API handles volume growth naturally through its queue-based processing, ensuring that even during peak news cycles, your audio generation remains reliable and consistent. You do not need to manage servers or anticipate traffic spikes manually; the system adapts to your request load.
Adding new languages is seamless because all eight voices support all 23 available languages. Whether you expand into Spanish, Hindi, or Turkish markets, you use the same API endpoints and voice presets. This uniformity allows you to scale geographically without learning new integration patterns or managing separate voice models for different regions.
Adding new languages
Since every voice can speak all 23 supported languages, expanding your content library does not require technical reconfiguration. You simply pass the appropriate ISO language code in your request. This means you can launch a French edition or a German update instantly, using the exact same Aria or Atlas voice, ensuring brand consistency across all markets without needing new voice assets.
Accent consistency is maintained because the voice models are trained to handle multiple languages natively. There is no need for separate 'English Aria' and 'French Aria' assets. Your infrastructure remains simple: one API key, one set of endpoints, and dynamic language selection based on the article's target audience.
- All 8 voices support all 23 languages
- No separate voice models per language
- Dynamic language code in API requests
Volume growth handling
Your account supports 60 requests per minute per key. As your output grows, you can monitor usage via the dashboard. If you need to exceed this limit, you can issue a new API key. Note that issuing a new key stops the old one immediately, so you switch to the new key rather than running them in parallel with a single account. For higher throughput, you can manage multiple independent accounts, each with its own key and rate limit.
The API processes requests sequentially, so under heavy load, requests may wait in a queue. This ensures fair resource distribution. For long texts, the 4096-character limit is handled by splitting and joining audio files automatically on your side or via simple client logic. This approach keeps the API stateless and easy to scale.
- 60 requests per minute per key
- Issue new key to switch capacity
- Sequential processing with queuing
Infrastructure needs
Since the API is stateless and returns audio files directly, your infrastructure requirements remain minimal. You do not need dedicated speech synthesis servers or GPU instances. Your team only needs to manage the API key, handle the JSON responses, and store the resulting audio files. This reduces your operational overhead significantly compared to self-hosted TTS solutions.
Bandwidth considerations are straightforward: you receive standard audio formats (MP3, WAV, etc.). Storage management depends on your retention policy. Since generated audio is not stored by Curisum Audio, you are responsible for your own storage. This gives you full control over your data and costs, with no surprise fees for long-term storage or egress.
- Stateless API, no dedicated servers
- Standard audio formats (MP3, WAV)
- User-managed storage and bandwidth
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. news outlets often overlook the creative spirit of indie developers in their coverage.
23 languages
Any voice speaks any of these languages.
