Frequently asked questions
Which voice is best for IVR?
Atlas offers a neutral, news-style tone ideal for clear menu instructions. For a more conversational feel, Milo is friendly and natural. Sage provides a calm, instructional voice suitable for guided workflows. The choice depends on your brand’s desired tone, but neutral voices like Atlas typically reduce listener fatigue in high-volume call routing.
What is the difference between Aria and Nova?
Aria is a warm, clear narrator voice suited for storytelling or detailed explanations. Nova is bright and upbeat, better for energetic prompts or welcome messages. Both support all 23 languages. Choose Aria for clarity and Nova for engagement. Both are female voices with distinct tonal qualities to match different IVR contexts.
Can voices speak any language?
Each voice can speak all 23 supported languages, including English, German, Spanish, and Japanese. If you do not specify a language, the system guesses it from the script, defaulting to English for Latin text. For non-English Latin-script text, explicitly set the language parameter to ensure accurate pronunciation and delivery.
How much do voices cost?
All voices share the same pricing: $0.01 per 1,000 characters. There are no extra fees for specific voices or languages. Failed requests are free. You pay only for successful synthesis. This flat rate applies regardless of whether you use tts-1 or tts-1-hd models or any of the eight available voice presets.
Are there male and female voices?
Yes, there are four female voices: Aria, Nova, Luna, and Sage. There are also four male voices: Felix, Milo, Atlas, and Orion. All voices support all 23 languages. You can mix male and female voices in your IVR flow to provide variety or assign specific roles to different voice types.
Is there a free voice trial?
Accounts created via Google sign-in receive $0.50 in free trial credit, valid for 7 days. Email/password accounts start without trial credit. This credit covers approximately 50,000 characters. You can test voices and latency without immediate payment. Top-ups via USDT (TRC-20) are required for continued usage after the trial or balance depletion.
Can I customize voice speed?
Yes, you can adjust the speed between 0.25 and 4.0. This allows you to slow down complex instructions or speed up simple prompts. The default speed is 1.0. Adjusting speed does not affect the character count or pricing. Faster speeds may reduce listener comprehension, so test with your specific IVR content.
What is the quality of the audio?
Audio is generated using licensed recordings from LibriTTS-R and LibriVox. Files contain an inaudible watermark marking them as synthetic. Supported formats include MP3, WAV, Opus, FLAC, and PCM. Opus is efficient for IVR bandwidth. Quality is consistent across all voices, providing natural-sounding speech suitable for professional call center environments.
Do voices support punctuation?
Yes, voices interpret standard punctuation for natural pauses and intonation. Spaces and punctuation count as characters in your usage. You can include commas, periods, and other standard marks in your input text. The API processes these naturally without requiring special SSML tags, keeping integration simple for standard text inputs.
Is the API easy to integrate?
The API uses a standard JSON POST request shape similar to OpenAI’s speech endpoint. You can use the official OpenAI SDK by changing the base URL. Authentication uses a Bearer token. Simple HTTP requests work too. This standardization makes integration straightforward for developers familiar with common RESTful TTS services.
What is the latency like?
Typical speed is roughly real-time, about 25 characters per second for long texts. Short phrases may take a few seconds. Requests are processed sequentially, so high load may cause queueing. For IVR, plan for slight delays. Use caching for static menu items to minimize API calls and reduce latency impact.
Are voices watermarked?
Yes, every generated audio file carries an inaudible watermark identifying it as synthetic speech. This is automatic and does not affect audio quality or playback. The watermark helps distinguish AI-generated speech from human recordings. It is transparent to the end-user and does not require any additional processing or configuration on your part.
Can I use voices for hold music?
Yes, you can generate spoken content for hold messages. However, voices are not traditional hold music; they generate speech. For background music, you would need separate audio files. You can use voices to announce wait times or updates. Pricing is per character, so long hold messages will incur character costs.
What is the character limit?
Each request supports up to 4,096 characters. For longer texts, split them into parts and join the audio files. This limit applies to the input text field. Spaces and punctuation count toward the limit. Plan your IVR scripts to fit within this constraint or implement client-side splitting logic for longer announcements.
How do I choose a voice?
Select based on tone: Atlas for neutral news, Milo for friendly conversation, Aria for clear narration, or Felix for energetic promos. All voices support all 23 languages. Test each voice with your specific IVR scripts to ensure clarity. Consider listener fatigue; neutral voices like Atlas are often best for repetitive menu loops.
Is there a demo available?
You can generate audio directly via the API using your API key. Sign up at /get-api-key/ to receive your key and trial credit. Use the provided curl examples or SDKs to test voices.
What formats are supported?
Supported formats are MP3, WAV, Opus, FLAC, and PCM. Opus is recommended for IVR due to its low bandwidth usage. MP3 and WAV are widely compatible. FLAC offers lossless quality. PCM is raw audio. Choose the format that best fits your telephony system’s requirements and bandwidth constraints.
Can I switch voices easily?
Yes, simply change the voice parameter in your API request. No reconfiguration or code changes are needed beyond updating the voice name. All voices share the same endpoint and request structure. You can A/B test different voices by switching the parameter in your IVR script or application logic.
Is there a volume discount?
No, there are no volume discounts or tiered pricing. The rate is flat at $0.01 per 1,000 characters regardless of usage volume. There are no monthly plans or minimum spends. Pricing is consistent whether you generate 1,000 characters or 1 million. Top-ups are processed via USDT on the TRON network.
What is the best voice for German?
All eight voices support German. Atlas provides a neutral, clear tone suitable for business IVR. Milo offers a friendly, conversational style. Aria is clear and warm. There is no single "best" voice; it depends on your brand voice. Test Atlas for neutrality or Milo for approachability in German-language menus.
Selection Criteria
Choosing a TTS API for IVR requires balancing audio clarity with cost efficiency. Our audit focuses on latency, character efficiency, and language support to help startups deploy natural-sounding voice menus without overpaying.
Clarity and Naturalness
Audio quality directly impacts customer satisfaction in high-volume call routing. We evaluate voice presets like Aria and Felix for their warmth and energy, ensuring they sound natural rather than robotic.
- Realistic intonation reduces listener fatigue
- Consistent audio quality across different lengths
- Inaudible synthetic watermark for transparency
Language Support
Support for 23 languages allows global reach without managing multiple providers. The API guesses the language from the script but allows explicit setting for non-English Latin text to avoid errors.
- 23 supported languages including German and Spanish
- Automatic language detection for Latin scripts
- Explicit language code support for precision
Cost Efficiency
Pricing is straightforward at $0.01 per 1,000 characters, with no monthly plans or hidden fees. Failed requests are free, ensuring you only pay for successful audio generation.
- $0.01 per 1,000 characters
- No minimum spend or monthly plans
- Free failed requests
Top Female Voices
Selecting the right female voice is critical for IVR clarity and user retention. Our audit focuses on how these presets perform under high-volume call routing conditions, balancing naturalness with processing efficiency.
Each voice is evaluated for latency, character efficiency, and language support, ensuring your startup avoids overpaying while maintaining a professional customer experience.
Aria: Warm Narrator
Aria delivers a warm, clear narration style ideal for standard welcome menus. Her tone reduces listener fatigue during repeated calls, making her a reliable default for general inquiries.
- Clear, warm narration
- Low listener fatigue
- Standard welcome menus
Nova: Bright and Upbeat
Nova’s bright and upbeat delivery works well for promotional holds or quick status updates. The energetic tone keeps callers engaged without sounding robotic, supporting high-volume routing efficiently.
- Bright, upbeat tone
- Promotional holds
- High engagement
Luna: Soft Storytelling
Luna offers a soft, storytelling cadence suitable for detailed error explanations or account updates. Her gentle pace aids comprehension in noisy environments, reducing repeat calls for clarification.
- Soft, storytelling pace
- Error explanations
- Noisy environments
Top Male Voices
Male voices provide the structural backbone for IVR systems, balancing authority with approachability. Our selection prioritizes clarity in noisy call center environments, ensuring callers understand navigation options without repeated prompts. Each preset is tuned for specific tonal expectations, from energetic announcements to neutral information delivery.
Felix: Energetic Promo
Felix delivers high-energy, promotional tones ideal for short welcome messages or service announcements. His distinct cadence cuts through background noise, making him suitable for time-sensitive alerts where engagement is critical.
- Best for: Welcome greetings, promotional updates, urgent alerts
Milo: Friendly Conversational
Milo offers a warm, conversational tone that reduces caller friction. His natural inflection makes complex menu structures feel like a dialogue, improving user retention in longer call flows.
- Best for: Menu navigation, account inquiries, support routing
Atlas: Neutral News
Atlas provides a consistent, neutral voice perfect for standard information retrieval. With no emotional bias, he ensures clarity for status updates, balance checks, or transaction confirmations across diverse demographics.
- Best for: Status updates, transaction confirmations, standard info
Specialty Voices
For IVR systems requiring precise instruction delivery, Sage offers a calm, instructional tone that reduces cognitive load during complex menu navigation. Orion provides a deep, documentary-style voice ideal for premium or corporate branding, adding authority to wait messages and error announcements.
Sage: Calm Instructional
Sage is optimized for clarity in high-volume call routing. Its steady pacing ensures instructions are understood even in noisy environments, making it suitable for detailed menu options or account verification steps.
- Steady pacing for complex instructions
- Low listener fatigue
- Ideal for verification steps
Orion: Deep Documentary
Orion’s deep timbre conveys trust and stability, making it effective for brand reinforcement during hold times or premium service announcements. It maintains neutrality without sounding robotic, ensuring a professional tone.
- Deep, authoritative tone
- Suitable for premium branding
- Neutral yet engaging
Language Compatibility
IVR systems require precise language handling to ensure callers understand menu options regardless of their native tongue. Our API supports 23 languages, allowing you to route calls to the appropriate voice module based on user preference or detected language.
This flexibility ensures that your call center operates efficiently across diverse demographics without needing separate voice engines for each language, maintaining consistent quality and cost structure.
German Voice Performance
German is fully supported with natural-sounding voices like Aria and Felix. The TTS engine handles German phonetics accurately, ensuring clear pronunciation for complex menu items and instructions.
- Full support for standard German pronunciation
- Compatible with all 8 voice presets
- No additional cost for German-specific rendering
Multilingual Capabilities
The API supports 23 languages including Arabic, Chinese, Danish, Dutch, English, Finnish, French, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish.
Each voice can speak any of these languages, allowing you to mix languages within a single IVR flow if needed, though sticking to one language per call is recommended for clarity.
- 23 supported languages
- Any voice can speak any language
- Consistent audio quality across all languages
Default Language Detection
If the language field is not set, the API guesses the language from the script. Latin script text defaults to English.
For non-English Latin-script text, such as Spanish or French, you must explicitly set the language parameter to ensure accurate pronunciation and intonation.
This avoids mispronunciations that can confuse callers and increase call duration.
- Auto-detection for Latin script defaults to English
- Explicit
languagefield required for non-English Latin text - Ensures correct pronunciation for all supported languages
Pricing Comparison
IVR systems process high call volumes, making predictable billing essential. Our pricing model charges a flat $0.01 per 1,000 characters regardless of the selected voice or audio format. This uniformity simplifies cost forecasting for startups deploying natural-sounding voice menus without complex tiered structures.
Uniform Pricing Model
The API charges $10 per 1 million characters. Spaces and punctuation count toward the limit. Failed requests incur no charge, protecting your budget during network interruptions or invalid inputs. This ensures you only pay for successful audio generation.
- Flat rate: $0.01 per 1,000 characters
- Spaces and punctuation count as characters
- Failed requests are free
Cost per Voice
Voice selection does not affect the price. Whether you use Aria for warm greetings or Felix for energetic prompts, the character cost remains identical. This allows easy A/B testing of voice personalities without impacting your operational expenses.
Volume Discounts
There are no monthly plans, seats, or minimum spend requirements. You top up via USDT (TRC-20) starting at $5, with credits credited automatically within minutes. Balance does not expire, allowing flexible usage for seasonal call center peaks without paying for idle capacity.
Technical Specs
The API supports five audio formats: mp3, wav, opus, flac, and pcm. This flexibility allows IVR systems to optimize for bandwidth or compatibility, ensuring clear voice playback across various telephony networks without additional conversion steps.
Speed control ranges from 0.25x to 4.0x, enabling fine-tuned pacing for complex menu structures. Combined with the standard 4096-character input limit per request, long scripts can be split and joined seamlessly, maintaining natural flow.
Audio Formats
Supported formats include mp3, wav, opus, flac, and pcm. Selecting opus or mp3 reduces bandwidth usage for high-volume call centers, while wav or flac ensures uncompressed quality for local storage or specific PBX requirements.
- mp3, wav, opus, flac, pcm
Speed Control
Adjustable speed from 0.25 to 4.0 allows optimization for listener comprehension. Slower speeds improve clarity for complex instructions, while faster speeds reduce hold times during peak traffic periods.
Character Limits
Each request accepts up to 4096 characters. For longer scripts, split the text into parts, generate audio files, and join them programmatically. This approach ensures consistent latency and avoids truncation.

Use Case Recommendations
Choosing the right voice for IVR depends on context. Welcome menus need warmth to reduce abandonment, while error messages require clarity to prevent repeat calls. Hold messages benefit from calm tones to minimize perceived wait time.
Our API handles these variations with specific voice presets, ensuring consistent quality across all call flows without complex configuration.
Welcome Menus
Use Aria or Nova for welcoming callers. Their warm, clear, or upbeat tones create a positive first impression. This reduces early drop-offs in high-volume call centers.
The natural pacing helps listeners process information quickly, especially when combined with the standard 1x speed setting.
- Aria: Warm, clear narrator for general greetings.
- Nova: Bright, upbeat tone for promotional calls.
Error Messages
Atlas provides a neutral, news-like delivery for error codes. Its clarity ensures digits and instructions are understood even in noisy environments.
Use Milo for conversational errors, such as 'Please try again.' Its friendly tone softens the frustration of a failed transaction.
Keep messages under 4096 characters to avoid splitting, ensuring a seamless audio response.
- Atlas: Neutral, ideal for clear instructions.
- Milo: Friendly, good for conversational prompts.
Hold Messages
Luna or Orion work best for hold music alternatives. Luna’s soft storytelling tone feels less repetitive, while Orion’s deep documentary voice adds authority.
These voices reduce listener fatigue during long wait times. Set speed to 1.0 for natural pacing.
Unlike music, TTS provides relevant information without requiring additional licensing for background tracks.
- Luna: Soft, reduces fatigue during holds.
- Orion: Deep, authoritative for updates.
Auditory Testing
Auditory testing focuses on how synthetic speech performs in real-world IVR environments. We evaluate latency, clarity, and listener fatigue to ensure your voice menus remain effective under pressure.
The API returns audio in formats like MP3 and WAV, allowing you to balance quality and bandwidth. Speed control (0.25–4.0) lets you optimize playback time without losing intelligibility.
Latency Impact
Short phrases take a few seconds to generate, while longer texts process at roughly real-time speed (25 characters per second). Requests are queued sequentially, so high load may add slight delays.
Use the speed parameter to reduce wait times for users. Faster playback can mask minor latency, but keep it above 1.0x for clarity.
- Short phrases: a few seconds
- Long texts: ~25 chars/sec
- Sequential processing under load
Clarity in Noise
Voices like Atlas (neutral, news) and Sage (calm, instructional) project well in noisy call center environments. Their clear enunciation reduces mishearing rates.
Test audio in your specific acoustic environment. Use MP3 or WAV for broad compatibility.
- Atlas: neutral, clear enunciation
- Sage: calm, instructional tone
- MP3/WAV: broad compatibility
Listener Fatigue
Repetitive menus can cause fatigue. Mix voices like Luna (soft, storytelling) and Nova (bright, upbeat) to keep listeners engaged.
Avoid speeds above 2.0x for complex information. Use the language parameter to ensure correct pronunciation, reducing listener effort.
Regularly audit your IVR script. Remove redundant phrases and keep messages concise to maintain a pleasant user experience.
- Mix voices to reduce monotony
- Keep speed below 2.0x
- Audit scripts for conciseness
Integration Ease
The API follows the standard OpenAI request shape, allowing you to reuse existing integration code by simply updating the base URL to https://api.sotmal.info/v1. This compatibility minimizes development overhead for startups already familiar with common TTS endpoints.
API Compatibility
Send a POST request to /v1/audio/speech with JSON body parameters including model, input, voice, and response_format. The endpoint returns an audio file directly, simplifying the response handling logic in your call routing scripts.
- Standard JSON body structure
SDK Support
We do not provide a proprietary SDK; instead, we support the official OpenAI SDKs. By changing the base_url configuration, you can use the same libraries you already trust for text-to-speech operations without learning a new interface.
Documentation Quality
Clear specifications for all 8 voices and 23 supported languages are available in our docs. We provide concrete curl examples and field definitions, ensuring you can test endpoints immediately without guessing parameter formats or required headers.
Watermarking and Authenticity
Every audio file generated by our API carries an inaudible watermark that clearly marks it as synthetic speech. This technical marker allows IVR systems and downstream platforms to automatically identify content that was machine-generated rather than recorded by a human voice actor.
Synthetic Watermark
The watermark is embedded directly into the audio stream during synthesis, ensuring it remains intact across all supported formats including MP3, WAV, and Opus. It serves as a reliable technical signal for content management systems without altering the audible quality experienced by callers.
Brand Trust
Using synthetic voices requires transparency with your users. By acknowledging that the voice is AI-driven, you manage expectations regarding the naturalness of the interaction. This honesty prevents user confusion and builds long-term trust in your automated support channels.
Transparency
We do not hide the synthetic nature of the output. The watermark provides verifiable proof of origin, which is increasingly important for regulatory compliance and content auditing in high-volume call centers where AI voices are used for billing or legal notices.
Final Recommendations
For IVR systems prioritizing a balance of clarity, warmth, and natural flow, Aria is the most reliable choice. Her clear narration style handles complex menu structures without sounding robotic, making her ideal for general-purpose greeting lines where listener comfort matters.
Aria also performs exceptionally well for German content, maintaining high intelligibility and natural rhythm even when switching between languages. Her neutral yet warm tone ensures that German speakers feel the system is responsive and professional, reducing friction during initial interactions.
Best Overall Voice
Aria is a reliable choice for general IVR use. Her warm, clear narration style minimizes listener fatigue during long menus. She handles complex instructions naturally, making her the safest default choice for startups deploying voice menus.
Best for German
Aria is a strong choice for German. She maintains high intelligibility and natural rhythm in German, ensuring that German-speaking callers feel the system is responsive and professional. Her performance reduces friction during initial interactions, which is critical for first impressions.
Best for Cost
All voices share the same transparent pricing of $0.01 per 1,000 characters. There are no volume discounts or per-voice premiums. Use Aria for quality or switch to Felix or Milo if a more energetic or conversational tone is needed for specific prompts without affecting your budget.
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. For longer scripts, split the text into parts, generate audio files, and join them programmatically.
23 languages
Any voice speaks any of these languages.
