Frequently asked questions
What is the base URL for the API?
The API base URL is https://api.bme688.info/v1. All requests, including speech synthesis and voice retrieval, are directed to this endpoint. You can use the official OpenAI SDKs by configuring the base_url parameter to this value, ensuring compatibility with existing integration patterns.
Which SDKs are compatible?
The API is compatible with the official OpenAI SDKs because the request shape matches the OpenAI speech endpoint. You can also use plain HTTP requests. There is no dedicated LinguaLink SDK, so you must rely on the OpenAI client libraries or standard HTTP libraries.
How do I set the base URL?
Set the base URL to https://api.bme688.info/v1 in your SDK configuration. For the OpenAI SDK, update the base_url parameter. For plain HTTP, direct POST requests to https://api.bme688.info/v1/audio/speech. Ensure your Authorization header includes your Bearer token.
What is the maximum input length?
The maximum input length is 4096 characters per request. If your text exceeds this limit, you must split it into smaller parts. Generate audio for each part separately and then join the resulting audio files to create the final output.
Can I use the OpenAI SDK directly?
Yes, you can use the official OpenAI SDK directly. Simply change the base_url to https://api.bme688.info/v1. The request structure for the speech endpoint is identical, allowing you to use the same code with your LinguaLink API key.
What happens if I exceed the rate limit?
The rate limit is 60 requests per minute per key. If you exceed this, requests may be queued or rejected. New accounts start with a 10 requests per minute limit. Ensure your integration handles retries or pauses to stay within these limits.
How do I handle errors?
Errors return standard HTTP status codes. Failed synthesis requests are free, and the charge is returned in full. Check the response body for error details. Ensure your text is under 4096 characters and your API key is valid before sending requests.
Is the API synchronous?
Yes, the API is synchronous. You send a POST request and receive the audio file in the response. Requests are processed sequentially, so under load, you may experience delays. There is no asynchronous webhook or polling mechanism for completion.
Can I stream the audio?
No, the API does not support streaming or WebSocket connections. You receive the complete audio file after synthesis. For real-time needs, you must manage client-side buffering or pre-generate audio files. There is no partial audio delivery.
How do I revoke my API key?
You can revoke your API key by issuing a new one from your account dashboard. The old key stops working immediately. This is the only way to invalidate a key. There is no separate revoke endpoint; generating a new key replaces the old one.
What is the default voice?
The default voice is not specified in the API documentation; you should explicitly set the voice parameter. You can specify any of the eight available voices in your request. If you do not specify a voice, the system behavior is not explicitly defined in the provided facts; it is recommended to specify a voice. Select a voice that best fits your call center or IVR tone.
Can I specify a language?
Yes, you can specify the language using an ISO code. If omitted, the language is guessed from the script, defaulting to English for Latin text. For non-English Latin text, always set the language parameter to ensure accurate synthesis.
Does the API support SSML?
No, the API does not support SSML. You cannot use phoneme dictionaries or custom pronunciation lexicons. Input is plain text up to 4096 characters. Adjust speed and voice presets to control the output, as there are no SSML tags available.
How are characters counted?
Characters are counted exactly as provided, including spaces and punctuation. The price is $0.01 per 1,000 characters. Failed requests are free. Ensure your input text is accurate, as every character contributes to the usage count and billing.
First request
curl
curl https://api.bme688.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.bme688.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.bme688.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
Authentication for the LinguaLink TTS API relies on a single active Bearer token per account. Include this token in the Authorization header of every request to verify your identity and authorize usage against your account balance.
Your API key is issued upon account creation or via the dashboard. It serves as the sole credential for accessing all endpoints, including speech generation and voice listing, ensuring secure access to your synthesis credits.
Bearer Token
Include the token in the Authorization: Bearer <key> header. The API validates this token against your account to process requests and deduct characters from your balance.
Ensure the token is transmitted securely over HTTPS. It grants full access to your account’s features and billing data.
- Format:
Authorization: Bearer tts-xxxxxxxxxx - Required for all POST and GET requests
- Case-sensitive string
Key Management
Each account holds one active API key. You can view and copy your current key from the dashboard at /get-api-key/. The key remains valid until explicitly revoked or replaced.
The key allows up to 60 requests per minute (10 per minute before the first top-up). Monitor usage via the dashboard to track character consumption and remaining balance.
- One active key per account
- View current key in dashboard
- Key format:
tts-prefix
Revocation Process
You can issue a new key at any time from the dashboard. Upon generating a new key, the previous key becomes invalid immediately. This ensures rapid response to potential security concerns.
Revocation is instant. Any requests using the old key after the new key’s creation will receive an authentication error.
- Generate new key via dashboard
- Old key stops instantly
- No delay in revocation
Endpoints
The API exposes three core endpoints for integration. The speech endpoint accepts text and configuration to generate audio files. The voices endpoint lists available presets, while the models endpoint provides engine details.
Speech Generation
Send a POST request to /v1/audio/speech with your text and voice selection. The service returns an audio file in the specified format. Text is processed character by character, respecting the 4096-character limit per request.
Voice Listing
Query GET /v1/voices to retrieve available voices, languages, formats, and price. This endpoint helps you verify supported options before generating audio. It confirms which voices are active and their associated costs.
Model Information
Use GET /v1/models to check the available synthesis engines. Currently, the API offers tts-1 and tts-1-hd models. Both models use the same underlying engine and pricing structure, allowing flexible selection based on quality needs.
Request Parameters
The API requires specific parameters to generate synthetic speech. The model field accepts either "tts-1" or "tts-1-hd", both using the same engine and pricing structure. The input field contains the text to be synthesized, limited to 4096 characters per request. Spaces and punctuation count toward this limit. For texts exceeding this length, split the input into multiple requests and join the resulting audio files.
The voice parameter selects one of eight presets, such as aria, nova, felix, or atlas, based on your tonal requirements. Every voice can speak any of the 23 supported languages. Use the optional language field to specify an ISO code if the script is ambiguous. Without it, Latin text defaults to English.
Model Selection
The model parameter determines the synthesis engine. Both "tts-1" and "tts-1-hd" utilize the same underlying engine and incur identical charges. The choice between them does not affect the price or the processing speed. Select the model based on your preference for audio quality versus file size, though the API treats them functionally equivalent for standard IVR applications.
- Accepts "tts-1" or "tts-1-hd"
- Same engine and price for both models
- No performance difference in processing time
Input Text
The input field accepts plain text strings up to 4096 characters. Each character, including spaces and punctuation, is counted toward the limit and the billing total. If your script exceeds this length, you must split the text into smaller chunks. Generate audio for each chunk separately and concatenate the files in your application logic.
The API processes requests sequentially under load, which may introduce queuing delays. Ensure your application handles these delays gracefully when constructing long IVR menus or call center scripts.
- Maximum 4096 characters per request
- Spaces and punctuation count as characters
- Split longer texts into multiple requests
Voice Selection
The voice parameter selects one of eight presets, such as aria, nova, felix, or atlas, based on your tonal requirements. Every voice can speak any of the 23 supported languages. Use the optional language field to specify an ISO code if the script is ambiguous. Without it, Latin text defaults to English.
- Select from 8 preset voices
- All voices support all 23 languages
- Set
languagefor non-English Latin text - Default language is English for Latin script
- Voices include aria, nova, luna, sage, felix, milo, atlas, orion
- No custom voice cloning available
Response Formats
Selecting the correct response format ensures optimal integration with your IVR system. The API returns audio data directly in the response body, allowing you to handle the file stream or save it to disk immediately after synthesis completes.
MP3 Output
MP3 is the default and most widely supported format for telephony systems. It offers a good balance between file size and compatibility, making it suitable for standard voice prompts in call centers.
WAV Output
WAV provides uncompressed, high-quality audio suitable for legacy IVR hardware or systems requiring lossless playback. Use this format when audio fidelity is prioritized over bandwidth efficiency.
Opus and FLAC
Opus delivers high compression for low-bandwidth environments, suitable for VoIP integrations. FLAC provides lossless compression for archival purposes, ensuring the original audio quality is preserved.
SDK Integration
LinguaLink’s API is fully compatible with the official OpenAI SDKs. By simply redirecting the client’s base URL, developers can utilize existing Python or Node.js libraries without rewriting request logic. This ensures immediate integration for teams already familiar with the standard speech endpoint structure.
OpenAI Python SDK
Install the standard OpenAI library and initialize the client. The SDK handles JSON serialization and retry logic automatically, reducing boilerplate code for IVR bot implementations.
- Use the official
openaiPython package. - No custom client library required.
Base URL Modification
Set the base_url parameter to https://api.bme688.info/v1. This redirects all subsequent calls to the LinguaLink infrastructure while maintaining the standard request shape. Ensure the API key is passed via the standard authorization header.
JSON Body Structure
Pass model, input, voice, and response_format in the JSON body. The API accepts standard parameters identical to the OpenAI speech endpoint. Unsupported fields are ignored, ensuring backward compatibility with existing SDK implementations.

Rate Limits
Synthesis requests are processed sequentially, meaning your application may experience queueing delays during periods of high load. The system enforces a cap of 60 requests per minute for active keys.
Requests Per Minute
The API processes requests one after another. Under heavy load, a request may wait in a queue before synthesis begins. Typical speed is roughly real time on long texts (about 25 characters of text per second of generation); short phrases take a few seconds.
Initial Limit
New accounts start with a restricted rate of 10 requests per minute. This limit applies immediately upon key creation and remains in effect until the first top-up is processed. This ensures stable service during the trial phase.
Post Top-Up Limit
After your first USDT top-up, the rate limit increases to 60 requests per minute. Top-ups are processed automatically within minutes via the TRON network. The limit applies per API key, not per account.
Error Handling
Reliable error handling ensures your IVR bots continue functioning during transient network issues or temporary service disruptions. Understanding how the API responds to failures allows you to implement robust retry mechanisms and manage billing accurately.
HTTP Status Codes
The API returns standard HTTP status codes to indicate request success or failure. A 429 Too Many Requests signals that you have exceeded your rate limit. Other 4xx codes indicate client errors, such as invalid parameters, while 5xx codes represent server-side issues.
Retry Logic
Implement exponential backoff for 5xx errors and rate limit (429) responses. Short delays between retries prevent overwhelming the service during recovery. For 4xx errors, verify your request payload before retrying, as these errors are not transient.
Failure Refunds
If synthesis fails due to a server error, the charge is returned in full. Failed requests do not consume your character allowance. This ensures you are only billed for successfully generated audio files.
Curl Example
The following example demonstrates a standard synthesis request using the cURL command-line tool. It sends a short text string to the speech endpoint with the default configuration.
This snippet serves as a starting point for integration into IVR systems or call center bots. Adjust the voice and speed parameters as needed for your specific audio requirements.
Basic Request
Send a POST request to the API endpoint with your authentication key and the text payload. The system returns the audio file directly in the response body.
- Endpoint: https://api.bme688.info/v1/audio/speech
- Method: POST
- Body: JSON with model, input, voice
Headers
Include the Authorization header with your Bearer token. Set the Content-Type to application/json to ensure proper parsing of the request body.
- Authorization: Bearer YOUR_API_KEY
- Content-Type: application/json
Response Body
On success, the response contains the raw audio binary data. The format matches the requested response_format, such as mp3 or wav. Failed requests return an error JSON object.
- Audio binary stream
- Error JSON on failure
- HTTP 200 on success
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. For projects needing only basic IVR menus without custom cloning, this preset range is sufficient and reliable.
23 languages
Any voice speaks any of these languages.
