Frequently asked questions
What is SpeechSynthesisUtterance voice?
SpeechSynthesisUtterance is a browser API that uses the device’s built-in voice engine. It relies on the operating system’s available voices, which vary by platform and language. It offers no consistent quality across different browsers or devices, making it unsuitable for branded customer experiences where audio consistency is critical for professional presentation.
How does browser TTS differ from cloud TTS?
Browser TTS uses local system voices with inconsistent quality, while our cloud API delivers consistent, high-fidelity audio from 8 licensed presets. Browser voices depend on the user’s device OS, whereas our API ensures the same voice quality regardless of the client’s browser or operating system, using a simple POST request.
Can cloud TTS replace browser voice in web apps?
Yes, our API provides a reliable alternative to the browser’s SpeechSynthesisUtterance. You send text via POST and receive an audio file. This ensures consistent voice quality across all devices, unlike browser TTS which varies by OS. It is ideal for e-commerce product descriptions where brand consistency matters more than offline capability.
Which voices are best for e-commerce product descriptions?
A: Any of our eight presets works well. Aria offers a warm, clear narration for general descriptions, while Felix provides an energetic tone for promotional highlights. Milo is suitable for a friendly, conversational style, and Atlas delivers a neutral, trustworthy voice for technical specifications. All voices support 23 languages and maintain consistent audio quality for your online store.
Does the cloud API support multiple languages?
Yes, the API supports 23 languages including English, Spanish, German, French, Japanese, and Chinese. Any voice can speak any of these languages. You can specify the language via an ISO code or let the system guess it from the script, though explicit setting is recommended for non-English Latin text.
Is the cloud TTS API compatible with OpenAI SDKs?
Yes, the API uses the same request shape as the OpenAI speech endpoint. You can use official OpenAI SDKs by changing the base_url to https://api.kiriya.info/v1 and using your Kiriya API key. This allows for easy integration into existing workflows without needing a custom SDK.
How much does the cloud TTS API cost?
Pricing is $10 per 1 million characters, or $0.01 per 1,000 characters. Spaces and punctuation count as characters. Failed requests are free. There are no monthly plans, seats, or minimum spends. You pay only for what you use, with a free trial of $0.50 credit for new Google sign-ins.
Do I need a custom voice for my brand?
No, custom voice cloning is not offered. We provide 8 high-quality presets based on licensed recordings from LibriTTS-R. These voices are designed to sound professional and consistent. For most e-commerce use cases, choosing the right preset is sufficient to establish a distinct and reliable audio brand identity without needing cloning.
Can I use the API for audio books?
Yes, the API supports long texts by splitting them into parts of up to 4,096 characters and joining the audio files. Voices like Luna (soft storytelling) or Orion (deep, documentary) are well-suited for audiobooks. The typical speed is roughly real-time, making it efficient for generating full-length content from your source text.
Is there a limit on text length per request?
Each request supports up to 4,096 characters of input text. For longer texts, you must split the content into chunks of this size and join the resulting audio files. This ensures consistent processing speed and quality. The API does not support streaming, so you receive the complete audio file for each chunk.
Do voices sound different across different languages?
The same eight voice presets are used across all 23 supported languages. While the accent and intonation adapt to the language, the core voice character remains consistent. This ensures that your brand audio sounds uniform whether you are generating content in English, Spanish, Japanese, or any other supported language in your catalog.
What happens if a request fails?
If synthesis fails, the charge is returned in full. You are not billed for failed requests. This ensures you only pay for successfully generated audio files. Failed requests are free, providing a safe environment for testing and integration with your e-commerce systems.
Can I clone a specific voice for my store?
No, custom voice cloning is not offered. You cannot upload your own voice or design a custom voice. The API uses 8 fixed presets based on licensed recordings. This keeps the service simple and reliable. If you need a specific voice, you must choose the closest match from the available eight options.
Is the audio generated watermarked?
Yes, every generated audio file carries an inaudible watermark that marks it as synthetic speech. This helps identify AI-generated content without affecting the listening experience. The watermark is embedded in the audio file itself, providing a layer of transparency for your customers and helping distinguish your TTS audio from human-recorded tracks.
First request
curl
curl https://api.kiriya.info/v1/audio/speech \
-H "Authorization: Bearer $TTS_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"tts-1","voice":"aria","input":"Hello from the API.","response_format":"mp3"}' \
--output hello.mp3The Limitation of Browser-Based Voice
Browser-based text-to-speech relies on the Web Speech API, which varies significantly depending on the user’s device and operating system. An agency cannot guarantee that a voice heard on an iPhone will match the experience on Android or desktop browsers.
This inconsistency undermines brand cohesion for clients who need uniform audio across all touchpoints.
Inconsistent voices across devices
The same text may sound like a different speaker or even a different language accent depending on the client’s browser. This makes it difficult for digital agencies to maintain a consistent brand voice in voice-enabled e-commerce features.
Without a cloud-based standard, the audio experience becomes unpredictable and dependent on local system resources.
- Different voices on iOS vs. Android
- Varying quality on desktop browsers
- Unpredictable accent shifts
No SSML or pronunciation control
Browser APIs lack support for SSML, meaning agencies cannot control pacing, emphasis, or pauses. There is no way to ensure proper pronunciation of product names or technical terms.
This limits the professionalism of automated announcements or product stories.
Dependent on user's OS
The voice engine is tied to the operating system’s built-in speech synthesis. If the OS updates or changes default voices, the client’s experience changes without the agency’s control.
This adds risk for long-term projects where consistency is key.
- OS-specific voice engines
- No control over system updates
- Variable performance on low-end devices
Why Agencies Choose Cloud TTS
Agencies building voice-enabled features for online stores need reliability that browser-based solutions cannot guarantee. Client devices vary widely, causing inconsistent audio quality and rendering behaviors depending on the user’s operating system. A cloud API ensures that the audio experience remains uniform regardless of whether the customer is on a desktop, mobile, or tablet.
The consistent quality allows agencies to deliver professional-grade audio for product stories, promos, and updates without worrying about device-specific limitations.
Consistent audio quality everywhere
Browser-based TTS relies on the user's installed engines, leading to variable voice quality and rendering inconsistencies across different devices and operating systems. A cloud API decouples the audio generation from the client’s environment, ensuring that every customer hears the same high-quality synthetic speech, whether they are using an iPhone, Android device, or desktop browser. This consistency is critical for maintaining brand voice across all touchpoints in an online store.
Unlike browser APIs, our cloud solution does not depend on local OS updates or device-specific voice packs. This predictability simplifies testing and quality assurance for digital agencies, as the output remains identical across all user environments.
- Uniform audio quality across all devices and operating systems
- No dependency on local browser voice engines
- Simplified QA with consistent rendering results
23 languages with unified voices
Our API supports 23 languages, including Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. Crucially, every voice preset can speak any of these languages, allowing agencies to maintain a unified audio identity across diverse international markets. This eliminates the need to manage multiple voice libraries for different regions.
If the language is not explicitly set, the system guesses from the script, defaulting to English for Latin text. For non-English Latin-script text, agencies should explicitly set the language parameter to ensure accurate pronunciation. This flexibility supports global e-commerce strategies without requiring separate voice assets for each language.
- Support for 23 major languages
- All voices can speak all supported languages
- Unified audio identity for global stores
Programmatic control via API
Agencies gain precise programmatic control through a standard REST API compatible with the OpenAI SDK. By changing the base URL to https://api.kiriya.info/v1, existing integrations can be adapted quickly. Requests are sent as JSON bodies, allowing for automated generation of audio for product descriptions, promotional banners, and news updates directly from the store’s backend.
The API supports parameters for voice selection, response format, and speed (0.25–4.0), enabling dynamic audio customization based on context. For example, a faster speed can be used for quick promo alerts, while a slower speed enhances clarity for detailed product stories. This programmability integrates seamlessly into existing e-commerce workflows.
- OpenAI SDK compatible with simple base URL change
- JSON request structure for easy integration
- Dynamic control over voice, format, and speed
Voice Options for E-commerce
Selecting the right voice ensures your client’s store audio matches the brand’s tone. The API offers eight distinct presets, each optimized for specific e-commerce contexts.
These voices are based on licensed recordings from LibriVox, ensuring high quality without the variability of browser-based defaults.
Narrator voices for product stories
Aria provides a warm, clear narration ideal for detailed product descriptions. Luna offers a soft, storytelling tone that engages customers during longer reads.
These voices maintain consistency across devices, ensuring the narrative experience remains uniform whether accessed on mobile or desktop.
- Aria: warm, clear narrator
- Luna: soft, storytelling
Energetic voices for promos
Felix delivers an energetic tone perfect for promotional announcements and limited-time offers. Nova adds a bright, upbeat quality that drives attention.
Use these voices to highlight sales or new arrivals, capturing customer interest with dynamic audio that stands out in a quiet browsing environment.
- Felix: energetic, promo
- Nova: bright, upbeat
Neutral voices for news updates
Atlas provides a neutral, news-style delivery suitable for objective updates or policy changes. Milo offers a friendly, conversational tone for general communications.
These presets ensure clarity and professionalism, avoiding emotional bias while delivering essential information effectively to your audience.
- Atlas: neutral, news
- Milo: friendly, conversational
Technical Implementation
The API endpoint mirrors the OpenAI speech structure, allowing agencies to reuse existing client code by simply adjusting the base URL. This compatibility eliminates the need for custom middleware or complex translation layers when integrating voice features into online store frontends.
Requests are sent as standard JSON payloads, making the integration straightforward for developers familiar with RESTful services. The consistent request shape ensures that switching between development and production environments requires minimal code changes.
OpenAI SDK compatibility
By pointing the official OpenAI SDK’s base_url to https://api.kiriya.info/v1, teams can use familiar libraries without rewriting integration logic. This reduces development time for agencies adding voice capabilities to e-commerce platforms.
- Reuse existing OpenAI SDK clients
- Minimal code refactoring required
- Familiar authentication patterns
Simple JSON request structure
Each request contains only the necessary fields: model, input, voice, and response_format. This simplicity reduces parsing errors and speeds up debugging during initial integration phases.
The structure supports optional parameters like speed and language, giving agencies precise control over audio output without requiring complex configuration objects.
model: tts-1 or tts-1-hdinput: text up to 4096 charactersresponse_format: mp3, wav, opus, flac, or pcm
Fast generation speeds
Long texts are generated at roughly real-time speed, producing approximately 25 characters of text per second of audio. Short phrases may take a few seconds to process, with requests handled sequentially.
Under heavy load, requests may queue, but the consistent processing order ensures reliable delivery of audio files to the client application.
- ~25 chars/sec for long texts
- Sequential request processing
- Predictable latency for short phrases
Cost Efficiency for Scale
For agencies scaling voice features for online stores, costs scale linearly with usage. The API charges $0.01 per 1,000 characters, making it economical for both short product descriptions and long audio catalogs. Spaces and punctuation count as characters, ensuring predictable billing.
Pay-per-character pricing
Calculate exact costs per project. At $10 per million characters, large catalogs remain affordable. No hidden fees for high volume.
No monthly subscriptions
Eliminate wasted spend on unused seats or tier limits. Pay only for what you generate. Top up via USDT (TRC-20) when needed.
Free failed requests
If synthesis fails, the charge is returned in full. You never pay for broken audio. This protects margins during integration testing.

Data Privacy and Usage
Your input text is used strictly for generating the audio file and is not retained for model training. This ensures that your product descriptions and marketing copy remain proprietary to your agency and clients.
Text not used for training
The synthesis engine processes your characters solely to produce the output audio. We do not analyze or store the textual content for improving the TTS models, keeping your creative assets distinct from the training data.
Audio returned in response
Generated audio files are delivered directly in the API response and are not stored on our servers after delivery. This minimizes data retention and ensures you maintain full control over the final audio assets.
Metadata kept for billing
We retain request metadata, including timestamp, character count, and status, strictly for accurate billing and usage tracking. This data supports transparent invoicing without storing the actual voice content.
Supported Languages
Our API supports 23 languages, allowing agencies to deploy consistent voice experiences across diverse markets. Each voice can speak any of the supported languages, ensuring brand consistency whether you are launching in Berlin or Tokyo. This flexibility simplifies localization workflows for digital agencies managing multi-regional online stores.
Major European languages
We cover the core markets for European e-commerce, including English, German, French, Spanish, Italian, and Portuguese. Additional support is available for Danish, Dutch, Finnish, Greek, Norwegian, Polish, Russian, Swedish, and Turkish. This range ensures that most major Western markets are covered without needing multiple voice providers.
Asian languages included
For Asian markets, we support Chinese, Japanese, Korean, Hindi, and Malay. These languages are fully integrated into the same voice presets, meaning you can maintain a consistent audio brand across your global inventory. The API automatically detects language from the script, though explicit language codes are recommended for Latin-script texts.
Limited language support
While we cover 23 languages, several common ones are not supported. Specifically, Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, and Bengali are excluded. If your store targets these regions, you may need to fallback to browser-based synthesis or a different provider for those specific language segments.
- Not supported: Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, Bengali
Audio Formats and Delivery
Audio delivery formats vary by use case. MP3 offers high compression for web playback, while WAV provides uncompressed quality for local processing. Both are standard choices for e-commerce audio features.
For bandwidth efficiency, Opus delivers excellent quality at low bitrates, ideal for mobile stores. FLAC provides lossless compression for archival needs, ensuring no audio quality is lost during transfer.
MP3 and WAV support
MP3 is the default for web integration, balancing file size and compatibility. WAV is available when uncompressed audio is required for specific editing workflows or local playback systems.
- Standard for web playback
- Uncompressed quality option
Opus and FLAC options
Opus reduces data usage significantly, perfect for high-traffic mobile stores. FLAC ensures lossless storage for premium audio assets without increasing file size as much as WAV.
- Low bandwidth usage
- Lossless compression
PCM for raw processing
PCM provides raw, uncompressed audio data. This format is ideal for immediate processing in voice pipelines or when integrating with third-party audio engines that require raw input.
- Raw audio data
- Immediate processing ready
Getting Started Quickly
Create your API key in seconds at /get-api-key/. Choose Google sign-in for an instant $0.50 trial credit, valid for 7 days and roughly 50,000 characters. Email accounts start without trial credit. Your key allows immediate integration with the OpenAI SDK by adjusting the base URL.
The API key is unique to kiriya.info. You can issue a new key at any time, which immediately invalidates the old one. This ensures secure access control for your agency’s client projects without complex subscription management.
Instant API key generation
Sign up via Google or email on the /get-api-key/ page to receive your unique API key instantly. The dashboard displays your key, balance, and usage metrics. Issue a new key anytime to revoke access to the previous one immediately.
- Google or email authentication
- Immediate key activation
- Revocable via new key issuance
Trial credit available
Google sign-ups receive $0.50 in trial credit, valid for 7 days. This covers approximately 50,000 characters of synthesis. Email-based accounts do not receive trial credit, allowing agencies to test the API with real billing from the start.
Simple documentation
Documentation at /docs/ provides ready-to-use curl examples and SDK instructions. The API follows the standard OpenAI request shape, requiring only a base URL change. Rate limits are 60 requests per minute per key, scaling with your top-up balance.
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. The following examples illustrate how to implement each approach using standard tools available to modern web developers.
23 languages
Any voice speaks any of these languages.
