Frequently asked questions
Is this a direct replacement for Amazon Polly login?
Q: Is this a direct replacement for Amazon Polly? A: Yes, if you use the OpenAI SDK. Our endpoint https://api.ek33.info/v1/audio/speech accepts the same JSON format. Simply change the base_url to https://api.ek33.info/v1 and use your API key. This lets you swap in high-quality voices with a single line of code, making migration from legacy providers simple for indie developers.
How does the pricing compare to AWS Polly?
We charge $10 per 1 million characters, or $0.01 per 1,000 characters. This flat rate includes all voice models and formats. There are no monthly plans, seats, or minimum spends. Failed requests are free, ensuring you only pay for successful audio generation.
Can I use this API for automated phone calls?
Yes. The API returns standard audio formats like MP3, WAV, Opus, FLAC, and PCM. You can send text via POST requests and receive audio files. With speeds up to 4.0x, you can optimize generation time for IVR flows.
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. The system processes requests sequentially, so under load, a request may wait in a queue before processing.
Do I need to retrain voices for my brand?
No voice training is required. We provide eight preset voices (Aria, Nova, Luna, Sage, Felix, Milo, Atlas, Orion) based on licensed recordings. These voices are ready to use immediately. You can select any voice for any language without additional setup or training time.
Is there a free tier for developers?
Yes, new accounts signed in with Google receive $0.50 of trial credit, valid for 7 days. This covers about 50,000 characters. Accounts created with email and password start without this trial credit. There is no free tier beyond this initial trial credit.
How many languages are supported?
We support 23 languages: Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. Any voice can speak any of these languages. Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, and Bengali are not supported.
Can I change my API key if it gets compromised?
Yes. You can issue a new API key at any time from your dashboard. When a new key is issued, the old one stops working immediately. This allows you to secure your account quickly if a key is exposed or compromised by a third party.
Do you offer SSML support for pauses?
No, we do not support SSML. The API accepts plain text input up to 4096 characters. You cannot add phoneme dictionaries or custom pronunciation lexicons. For precise control, you can adjust the speed parameter between 0.25 and 4.0 to modify the playback rate.
Is the audio watermarked for detection?
Yes. Every generated audio file carries an inaudible watermark that marks it as synthetic speech. This helps identify the audio as AI-generated. The watermark is embedded in the audio file itself and does not affect the playback quality or listening experience for end users.
What is the maximum text length per request?
The maximum input text length is 4096 characters per request. For longer texts, you must split the text into parts of up to 4096 characters and join the resulting audio files. Spaces and punctuation count toward the character limit and pricing.
How long are my audio files stored?
Generated audio is returned in the response and is not stored by us. We keep request metadata (time, characters, status) for billing purposes only. The input text is used only to produce the audio and is not used for training models.
Can I upload my own voice for cloning?
No, we do not offer custom voice cloning. You cannot upload your own voice or design new voices. We provide eight preset voices based on licensed recordings. If you need a specific voice, you must choose from our existing library of eight options.
Is there a limit on characters per month?
No, there is no limit on characters per month. You pay per character used, with no monthly plans or minimum spend. Your balance does not expire. You can top up with USDT on the TRON network whenever you need more credit.
First request
curl
curl https://api.ek33.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.mp3Why Switch Your IVR Voice Engine?
Legacy TTS providers often tie you to complex, tiered volume discounts that complicate billing. Our platform simplifies this with a straightforward per-character cost. You pay $0.01 per 1,000 characters, making it easy to predict expenses for your IVR scripts without guessing future call volumes.
Simplify your billing with per-character costs
Stop managing multiple rate tiers. Every character, including spaces and punctuation, counts toward your usage. This flat rate structure removes the ambiguity of volume discounts, allowing your team to forecast costs accurately based on actual text length.
No more complex tiered volume discounts
Unlike traditional providers that require you to predict usage for better rates, our model scales linearly. You only pay for what you generate. Failed requests are free, so you never pay for errors.
Instant API key activation for dev teams
Sign up with Google or email to receive your API key immediately. Start testing with a $0.50 free trial credit (valid for 7 days) without entering a credit card. Your key is active instantly, allowing developers to begin integration without waiting for approval.
Voice Quality for Customer Calls
Switching your IVR engine means replacing robotic monotony with voices that actually sound like humans. Our engine uses licensed recordings from LibriTTS-R, ensuring every syllable carries natural intonation rather than the flat, predictable cadence of legacy providers.
This quality reduces customer confusion and hold times. When callers hear clear voices, they understand prompts faster, leading to smoother interactions and a more professional brand image.
Natural sounding female voices like Aria and Luna
Aria delivers a warm, clear narrative tone perfect for guided menus and welcome messages. Luna offers a soft, storytelling quality that makes waiting feel less tedious. Both presets handle conversational flow naturally, reducing the cognitive load on callers.
- Aria: Warm, clear narrator for standard IVR menus
- Luna: Soft, storytelling tone for wait messages
Clear male voices for news and prompts
Felix brings energetic enthusiasm ideal for promotional calls or urgent alerts. Milo provides a friendly, conversational style that feels personal rather than automated. Atlas remains neutral and steady, perfect for standard informational prompts where clarity is key.
Orion adds a deep, documentary-style gravity to your calls. Whether delivering news updates or critical system information, these male voices ensure authority and clarity without sounding stiff or overly scripted.
- Felix: Energetic for promos and alerts
- Milo: Friendly and conversational for general inquiries
- Atlas: Neutral and clear for standard info
- Orion: Deep and authoritative for news
8 distinct presets for different brand tones
You have eight unique voices to match your specific brand identity. From the bright energy of Nova to the calm instruction of Sage, you can select the exact tone that fits your customer service style.
Each voice is distinct, allowing you to segment different parts of your call flow. Use one voice for billing, another for support, and another for sales, creating a more engaging and organized caller experience.
- 8 total presets: Aria, Nova, Luna, Sage, Felix, Milo, Atlas, Orion
- Mix and match voices across different call flows
- Consistent quality across all supported languages
Language Support for Global Teams
Our API supports 23 languages, covering major markets like English, Spanish, and French. You can use any of our 8 voice presets with any of these languages, giving you consistent quality across global customer interactions without needing separate voice licenses for each region.
23 languages including English, Spanish, and French
We support Arabic, Chinese, Danish, Dutch, Finnish, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Swahili, Swedish, and Turkish. This covers most standard IVR needs, though we do not currently support Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, or Bengali.
Auto-detection for Latin script texts
If you do not specify a language, the system guesses based on the text script. Latin script text defaults to English. This works well for simple English prompts but may fail for other languages using Latin characters.
Manual language setting for mixed content
For non-English Latin-script text, you should set the language field explicitly. This ensures the voice pronounces words correctly. We recommend always defining the language for mixed-language IVR scripts to avoid mispronunciation.
Developer Friendly Integration
Switching your IVR voice engine is straightforward because our API mirrors the OpenAI speech endpoint structure. You only need to change the base_url in your existing OpenAI SDK configuration to https://api.ek33.info/v1 and swap in your API key. This compatibility means your current code logic remains intact, reducing migration friction.
The request body uses standard JSON fields like model, input, and voice. Since the shape matches OpenAI’s, you can often copy-paste examples directly. Just ensure you use the tts-1 or tts-1-hd model values and provide your tts-... bearer token in the authorization header.
OpenAI SDK compatible with base_url change
Our endpoint https://api.ek33.info/v1 accepts the same JSON structure as OpenAI. By updating your SDK’s base URL, you leverage the official OpenAI client libraries without rewriting your integration logic. This means existing scripts using the OpenAI SDK can point to our infrastructure with minimal code changes.
You simply replace the default base URL with https://api.ek33.info/v1 and update the API key. The request shape remains identical, so your error handling and retry logic continue to work as expected.
- Use official OpenAI SDKs with a single base_url configuration change
- No custom client library required; use standard HTTP or SDK wrappers
- API key format:
Authorization: Bearer tts-...
Standard JSON request body structure
Send a POST request to /v1/audio/speech with a JSON body containing the text, voice, and format. The input field accepts up to 4096 characters per request. If your IVR script exceeds this, split the text into chunks and join the resulting audio files.
Key fields include model (select tts-1 or tts-1-hd), voice (choose from 8 presets like Aria or Felix), and response_format (mp3, wav, opus, flac, or pcm). The speed parameter allows adjustments between 0.25 and 4.0.
- Endpoint: POST /v1/audio/speech
- Max input: 4096 characters per request
- Supported formats: mp3, wav, opus, flac, pcm
Easy migration from existing TTS libraries
Migrating from legacy providers often requires significant code rewrites. With our compatible structure, you can swap the provider in hours, not days. Update your base URL and API key, then adjust the voice names to our 8 presets.
For long IVR prompts, implement a simple client-side splitter to handle texts over 4096 characters. Concatenate the returned audio files to play a seamless prompt in your telephony system.
- Change base_url and API key to migrate in hours
- Split text >4096 chars and join audio files client-side
- Use standard JSON for predictable integration
Transparent Pricing Model
Switching to IVR Switch simplifies your billing by replacing complex tiered volume discounts with a single, transparent rate. You pay $10 per 1 million characters, which translates to just $0.01 per 1,000 characters. Spaces and punctuation count as characters, making it easy to predict costs for your IVR scripts before generation begins.
There are no monthly minimums, subscriptions, or seats to manage. Your balance never expires, and you can top up with USDT (TRC-20) starting at just $5. This model suits indie developers and small call center teams who want to avoid unexpected overage charges or wasted subscription fees during low-traffic periods.
$10 per 1 million characters flat rate
The cost is strictly proportional to input length. At $0.01 per 1,000 characters, generating a 60-second IVR prompt typically costs fractions of a cent. This flat rate applies to both tts-1 and tts-1-hd models, ensuring consistent budgeting regardless of the audio quality tier you choose.
- $10 per 1,000,000 characters
- $0.01 per 1,000 characters
- Spaces and punctuation count
No monthly minimums or subscriptions
Unlike legacy providers, we do not charge for idle time. You only pay when you send a request. Accounts start with a $0.50 free trial (valid for 7 days) when signing in with Google, allowing you to test integration without financial commitment. No credit card is required to begin.
- No monthly fees
- No minimum spend
- Trial credit available
Pay only for successful generations
If a synthesis fails, you are not charged. The credit is returned to your balance immediately. This ensures you only pay for usable audio files delivered to your IVR system, removing the risk of paying for truncated or corrupted audio responses.
- Failed requests are free
- Instant credit return
- Pay for usable audio only

Technical Limits You Should Know
Before integrating, ensure your IVR scripts respect the 4096-character limit per API call. Longer dialogues should be split into segments and stitched together after generation to maintain audio continuity without hitting the boundary.
4096 character limit per request
Each API call accepts up to 4096 characters, including spaces and punctuation. For lengthy prompts, divide the text into chunks, generate each part, and concatenate the resulting audio files.
60 requests per minute standard rate limit
Your API key supports 60 requests per minute once funded. New accounts start with a lower limit of 10 requests per minute until the first top-up is processed.
Real-time processing with potential queuing
Synthesis typically runs at real-time speed, generating roughly 25 characters of text per second. Short phrases take a few seconds, but under load, requests may queue and wait for processing.
Audio Formats for IVR Systems
Choose the right format for your IVR system’s playback engine. MP3 offers broad compatibility for standard web players, while WAV ensures high-fidelity delivery for telephony systems that require uncompressed audio.
For archival or specialized needs, OPUS and FLAC provide efficient compression or lossless quality respectively. All generated files include an inaudible watermark to identify synthetic speech.
MP3 for standard web players
MP3 is the default choice for most web-based IVR interfaces. It balances file size and quality, ensuring smooth playback across various devices and browsers without consuming excessive bandwidth.
WAV for high quality telephony
WAV provides uncompressed audio, ideal for telephony systems that demand consistent, high-quality sound. This format avoids potential compression artifacts that might affect voice clarity in phone calls.
OPUS and FLAC for archival needs
OPUS offers excellent compression for storage, while FLAC provides lossless quality for future-proofing. Both formats are supported alongside MP3 and WAV for flexible storage strategies.
Data Privacy and Usage
Your input text is used strictly to produce the audio file and is not added to our training data. Once the synthesis is complete, the generated audio is returned directly in the response and is not stored on our servers, ensuring your IVR scripts remain yours.
Input text not used for training models
We treat your script content as ephemeral. It is processed solely for the purpose of generating the requested audio file and does not contribute to improving the underlying model.
Generated audio returned in response
The resulting audio file is delivered immediately in the API response. We do not retain a copy of the generated voice file, so your call center recordings depend entirely on your own storage infrastructure.
Metadata kept for billing accuracy
To ensure accurate invoicing, we keep request metadata including the timestamp, character count, and status. This data helps us verify that you are charged correctly for the volume of text processed.
Start Your Free Trial Today
Stop guessing your TTS costs. Start with $0.50 in credit when you sign in with Google. That is about 50,000 characters to test your IVR flows without any upfront commitment.
Create your API key in seconds. You get immediate access to the dashboard, your key, and a ready-to-use curl example. It is a quick way to verify voice quality for your call center scripts.
Get $0.50 credit with Google sign-in
Sign up on /get-api-key/ using your Google account. You receive $0.50 of trial credit valid for 7 days.
Email/password accounts do not start with trial credit but can top up anytime with USTRC-20.
- Automatic credit application
- 7-day validity window
Test voices with sample IVR scripts
Use your free credit to generate audio with all 8 voices. Test Aria, Luna, or Atlas with your actual IVR prompts.
Verify clarity and tone before committing to volume.
Check the /docs/ page for request examples.
No credit card required to start
No subscriptions, no minimum spend, no invoices.
Pay only for successful generations at $10 per 1 million characters.
Top up with USDT on TRON when you need more.
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. Her gentle cadence helps complex information land clearly, reducing the need for callers to ask for repetition during onboarding or support calls.
23 languages
Any voice speaks any of these languages.
