Frequently asked questions
What is the API key format?
API keys begin with the prefix tts-. Each account has one active key. You can issue a new key at any time via the dashboard, and the old key stops working immediately.
Can I use the OpenAI SDK?
Yes. The request shape matches the OpenAI speech endpoint. You can use official OpenAI SDKs by changing the base_url to https://api.ux45.info/v1 and providing your VerbaSync API key for authentication.
What models are available?
Two models are available: tts-1 and tts-1-hd. Both use the same engine and price. You can list available models by sending a GET request to the /v1/models endpoint.
How do I change the base URL?
Configure your SDK or HTTP client to use https://api.ux45.info/v1 as the base URL. This allows you to use standard OpenAI SDK methods while routing requests through VerbaSync's infrastructure.
What is the character limit?
Each request supports up to 4096 characters of input text. For longer scripts, split the text into parts of this size, generate audio for each part, and join the resulting audio files.
Can I stream audio?
No. The API does not support streaming, WebSockets, or real-time partial audio. You send a POST request and receive the complete audio file in the response. Use formats like mp3, wav, opus, flac, or pcm.
Does it support SSML?
No. The API does not support SSML, phoneme dictionaries, or custom pronunciation lexicons. You can control speech speed using the speed parameter (0.25–4.0) and select from eight voice presets.
How do I set the language?
Include the language field in your request body with an ISO code. If omitted, the system guesses the language from the script, defaulting to English for Latin text. Explicitly setting it is recommended for non-English languages.
What if I exceed rate limits?
The limit is 60 requests per minute per key (10 per minute before the first top-up). Requests are processed sequentially, so exceeding limits may cause queuing delays. Monitor your usage via the dashboard.
Are failed requests refunded?
Yes. If synthesis fails, the charge is returned in full. You are only billed for successful requests. Failed requests do not consume your balance.
Is there a WebSocket endpoint?
No. There is no WebSocket endpoint. The API operates on standard HTTP POST requests, returning the full audio file upon completion. Use the /v1/audio/speech endpoint.
How do I rotate my API key?
Generate a new key from your account dashboard. The new key becomes active immediately, and the old key stops working at once. You can issue a new key anytime without downtime.
What is the default language?
If the language field is not set, the system guesses the language from the script. For Latin-script text, it defaults to English. For other languages, explicitly set the language parameter.
Can I send audio back?
Yes. The API returns the generated audio file directly in the response body. You can choose formats including mp3, wav, opus, flac, and pcm. The audio is not stored on our servers.
First request
curl
curl https://api.ux45.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.mp3Python (openai)
from openai import OpenAI
client = OpenAI(base_url="https://api.ux45.info/v1", api_key="tts-...")
with client.audio.speech.with_streaming_response.create(
model="tts-1", voice="nova", input="Hello from Python.", response_format="mp3"
) as response:
response.stream_to_file("hello.mp3")Node.js (openai)
import OpenAI from "openai";
import fs from "node:fs";
const client = new OpenAI({ baseURL: "https://api.ux45.info/v1", apiKey: process.env.TTS_API_KEY });
const res = await client.audio.speech.create({ model: "tts-1", voice: "milo", input: "Hello from Node.js." });
fs.writeFileSync("hello.mp3", Buffer.from(await res.arrayBuffer()));| voice | |
|---|---|
aria | female |
nova | female |
luna | female |
sage | female |
felix | male |
milo | male |
atlas | male |
orion | male |
| language | |
|---|---|
ar | العربية |
zh | 中文 |
da | Dansk |
nl | Nederlands |
en | English |
fi | Suomi |
fr | Français |
de | Deutsch |
el | Ελληνικά |
he | עברית |
hi | हिन्दी |
it | Italiano |
ja | 日本語 |
ko | 한국어 |
ms | Bahasa Melayu |
no | Norsk |
pl | Polski |
pt | Português |
ru | Русский |
es | Español |
sw | Kiswahili |
sv | Svenska |
tr | Türkçe |
Authentication
Access the VerbaSync TTS API using a single active bearer token issued to your account. You can generate this key via the /get-api-key/ dashboard. Each account holds exactly one active key at a time, ensuring straightforward authentication for your localization pipelines.
API Key Format
Your API key is a string starting with the prefix tts- followed by alphanumeric characters. This key serves as your sole credential for all API requests. Keep it secure, as it provides direct access to your billing and usage data.
- Prefix:
tts- - Single active key per account
- Viewable in the dashboard
Bearer Token Usage
Include the key in the HTTP Authorization header of every request. Use the format Authorization: Bearer tts-.... The API validates this token against your account to authorize speech synthesis and bill character usage accordingly.
- Header:
Authorization: Bearer <key> - Required for all endpoints
- Case-sensitive prefix
Key Rotation
If needed, you can issue a new key from your dashboard at any time. The old key becomes invalid immediately upon generation of the new one. Ensure your integration updates to the new key before the old one expires to avoid service interruption.
- Issue new key anytime
- Old key stops instantly
- No downtime if updated
Endpoints
The VerbaSync API exposes three core endpoints for generating and managing speech assets. The primary endpoint handles audio synthesis, while the metadata endpoints provide details on available voices and models to ensure your localization pipeline remains consistent across diverse language scripts.
All endpoints require a Bearer token in the Authorization header. You can retrieve your active key from the dashboard and rotate it instantly if needed. Each request is processed independently, returning standardized JSON responses that integrate seamlessly with standard HTTP clients.
POST /v1/audio/speech
Send text and receive synthesized audio. The endpoint accepts JSON with fields for model, input text, voice, response format, and speed. It returns the audio file directly in the response body, allowing you to save it locally or stream it to your dubbing tools.
Supports up to 4096 characters per request. For longer scripts, split the text into segments, generate each part, and concatenate the resulting audio files. Failed requests are free, ensuring you only pay for successful synthesis.
- Endpoint: https://api.ux45.info/v1/audio/speech
- Method: POST
- Auth: Bearer token required
GET /v1/voices
Retrieve a list of all available voices, their associated languages, and supported audio formats. This endpoint helps you map the correct voice ID to each language in your localization workflow. Each voice is designed for specific tones, from warm narrators to energetic promos.
The response includes metadata such as voice ID, gender, and language support. Use this data to dynamically select the appropriate voice for each script segment without hardcoding values.
- Endpoint: https://api.ux45.info/v1/voices
- Method: GET
- Returns: Voice list with language support
GET /v1/models
Fetch available TTS models, currently including 'tts-1' and 'tts-1-hd'. Both models use the same engine and pricing. Select the model that best fits your storage and bandwidth constraints for large-scale dubbing projects.
The response provides model IDs and descriptions. Use these IDs in the POST /v1/audio/speech endpoint to specify which quality level to use for your generated audio files.
- Endpoint: https://api.ux45.info/v1/models
- Method: GET
- Models: tts-1, tts-1-hd
Request Parameters
Configure the core synthesis parameters to match your dubbing pipeline requirements. Each request defines the voice model, the text content, and the audio characteristics needed for professional localization.
Model Selection
Choose between tts-1 for standard quality or tts-1-hd for higher fidelity. Both models use the same engine and pricing structure, allowing you to balance cost and audio quality based on project needs.
Input Text Limits
The input field accepts up to 4096 characters per request, including spaces and punctuation. For longer scripts, split the text into chunks and join the resulting audio files sequentially to maintain consistency.
Voice Selection
Select one of eight licensed presets, such as aria for warm narration or atlas for neutral news delivery. Each voice can speak any of the 23 supported languages.
- Female: aria, nova, luna, sage
- Male: felix, milo, atlas, orion
Response Formats
VerbaSync returns synthesized audio in the format specified by the response_format parameter. The API delivers the audio file directly in the response body, allowing your localization pipeline to ingest the asset immediately without intermediate conversion steps.
MP3 Output
MP3 is the default format, optimized for broad compatibility and efficient bandwidth usage. It is ideal for web-based playback and standard audio workflows where file size matters more than absolute fidelity.
WAV Output
WAV provides uncompressed, lossless audio data. This format is preferred for professional editing pipelines where you need to apply further effects or ensure zero quality loss before final mastering.
OPUS, FLAC, PCM Options
OPUS offers high-quality compression suitable for storage-constrained environments. FLAC provides lossless compression. PCM delivers raw uncompressed audio, useful for specific hardware integrations or precise sample-level analysis.
- OPUS: High efficiency, good for streaming
- FLAC: Lossless compression, smaller than WAV
- PCM: Raw data, no compression
Speed and Language
Speed and language control are critical for maintaining consistent dubbing quality across diverse scripts. Adjusting playback speed allows your team to match the pacing of original source material precisely, while explicit language codes ensure accurate phonetic rendering for non-English languages.
Speed Range (0.25–4.0)
The speed parameter accepts values from 0.25 to 4.0, allowing fine-grained control over audio pacing. Slower speeds (e.g., 0.75) help synchronize voiceovers with slower-paced source videos, while faster settings (e.g., 1.5) compress dialogue for quicker delivery without sacrificing clarity. Note that generation time scales with the speed factor; slower speech takes longer to synthesize.
Language ISO Codes
For optimal accuracy, always provide the language parameter using standard ISO 639-1 codes (e.g., fr, de, ja). While our models support 23 languages including Arabic, Chinese, Hindi, and Swahili, explicit specification prevents mispronunciation. This is especially important for technical scripts or proper nouns where context alone might not guide the model correctly.
- Supported languages: Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Turkish.
Auto-Detection Fallback
If you omit the language field, the system attempts to detect the language based on the script's characters. Latin script defaults to English, which may cause errors for other Latin-based languages like Spanish or Dutch. Always set the language explicitly for non-English content to avoid incorrect accents or phonetic choices.

Rate Limits
VerbaSync enforces a strict rate limit of 60 requests per minute per API key. This cap ensures stable processing for large localization teams generating audio from scripts. If you exceed this threshold, requests are queued and processed sequentially, which may introduce slight delays during peak generation hours.
60 Requests Per Minute
Each API key allows a maximum of 60 concurrent or sequential requests within a one-minute window. This limit applies uniformly across all endpoints. For high-volume dubbing pipelines, consider distributing requests across multiple keys if your workflow demands higher throughput.
- Limit is per key, not per IP or account.
- Counts all requests.
- Resets every 60 seconds.
Initial 10 RPM Limit
New accounts without a top-up start with a reduced limit of 10 requests per minute. This lower threshold applies before the first top-up. Once you add USDT (TRC-20) to your balance, the limit automatically increases to the full 60 RPM.
- Applies only before the first top-up.
- Increases automatically after funding.
Queue Behavior
Requests are processed one after another. If you send more than 60 requests in a minute, the excess wait in a queue. Longer texts take more time to generate, so under heavy load, your requests may experience noticeable delays. Plan your generation batches accordingly to avoid bottlenecks.
Error Handling
Handling errors correctly ensures your dubbing pipeline remains stable during high-volume generation. Our API provides clear, standard HTTP status codes to help you distinguish between client mistakes, such as invalid parameters, and transient service issues like temporary queue congestion.
Understanding these responses allows you to implement robust retry logic without wasting credits on duplicate requests or losing sync in your localization workflow.
Common Error Codes
We return standard HTTP status codes to indicate the result of your request. A 400 Bad Request indicates invalid JSON, missing fields, or an input text exceeding the 4096-character limit. A 401 Unauthorized error means your API key is missing, invalid, or has been rotated. A 429 Too Many Requests signal indicates you have exceeded the rate limit.
- 400: Invalid request body or character limit exceeded
- 401: Invalid or missing Bearer token
- 429: Rate limit exceeded (queueing may occur)
Failed Request Refunds
If synthesis fails due to a server-side error (5xx status codes), the character count is not deducted from your balance. The charge is returned in full automatically. You do not need to submit a support ticket for these refunds; they appear in your dashboard balance within minutes.
- 5xx errors trigger automatic credit refunds
- Refunds appear in your dashboard balance within minutes
- No manual claim process required
Retry Logic
For transient 429 errors, implement an exponential backoff strategy. Start with a short delay (e.g., 100ms) and double the wait time with each retry, up to a maximum of 10 seconds. This prevents overwhelming the queue during high traffic, which is common when processing long scripts for multiple languages.
- Use exponential backoff for 429 responses
- Start delay at 100ms, doubling up to 10s
- Retry for 5xx server errors
SDK Integration
The API mimics the OpenAI speech endpoint structure, allowing you to reuse existing OpenAI SDKs with minimal configuration. By updating the base URL to point to our infrastructure, you can leverage familiar libraries for authentication and request handling.
This approach simplifies integration for teams already accustomed to the OpenAI ecosystem, requiring only a base URL adjustment and the VerbaSync API key for authorization.
OpenAI SDK Setup
Install your preferred OpenAI-compatible SDK. Initialize the client by providing your API key and the VerbaSync base URL. This setup ensures compatibility with standard TTS requests without needing a custom library.
Base URL Configuration
Set the base URL to https://api.ux45.info/v1. This directs all requests to the correct infrastructure while maintaining the standard request shape expected by OpenAI SDKs.
Code Examples
Use the standard POST /v1/audio/speech endpoint. Pass your text, selected voice, and desired format. The response will return the audio file directly, ready for insertion into your dubbing pipeline.
- Supports standard OpenAI SDK methods
- Compatible with Python, Node.js, and other major languages
- No custom client library required
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. Failed requests are free, ensuring you do not pay for errors.
23 languages
Any voice speaks any of these languages.
Sources and further reading
- Wikipedia
- Python documentation
- openai-python on GitHub
- LibriTTS-R corpus (OpenSLR 141)
- Creative Commons Attribution 4.0
- LibriVox public-domain audiobooks
- Tether (USDT)
- curl documentation
- LibriTTS-R paper (arXiv 2305.18802)
- W3C WAI: making audio and video accessible
- FFmpeg documentation
- TRON developer hub: TRC-20
