Frequently asked questions
Is this API suitable for generating large-scale synthetic speech datasets for machine learning?
Yes. The API supports bulk generation by splitting texts into 4096-character segments and joining the audio files. With 23 languages and eight voice presets, it provides the volume and variety needed for synthetic speech datasets. Note that requests are processed sequentially, so under load, synthesis may queue. Failed requests are free, and pricing is strictly per character at $0.01 per 1,000 characters.
How does the character-based pricing work for bulk data generation?
Our pricing is $10 per 1 million characters ($0.01 per 1,000 characters), with spaces and punctuation counting toward the total. Failed requests are free, with charges returned in full. There are no monthly plans, seats, or minimum spend requirements, making it cost-effective for variable workloads.
Which programming languages can I use to call this TTS API?
Any language supporting HTTP requests works. Because the endpoint matches the OpenAI speech API shape, official OpenAI SDKs function seamlessly by updating the base URL to https://api.playvn500.com/v1. You may also use plain HTTP or any language capable of sending JSON POST requests, making integration straightforward for your synthetic speech dataset workflows.
Does the API support multiple languages for multilingual dataset creation?
Yes. The API supports 23 languages, including English, Chinese, Spanish, Arabic, Hindi, and Japanese. Any voice can speak any of these languages. While the system guesses the language from the script if the parameter is omitted, we recommend setting it explicitly for non-English Latin-script text to ensure accuracy in your multilingual datasets.
Are the generated audio files watermarked to indicate they are synthetic?
Yes. Every generated audio file includes an inaudible watermark identifying it as synthetic speech. This metadata helps you track the origin of audio data within your machine learning datasets without affecting audio quality or usability for model training pipelines. The watermark is embedded directly into the generated files, ensuring clear distinction between synthetic and human-recorded audio in your training sets.
What happens if a request fails during audio generation?
If synthesis fails, the charge is returned in full. You are billed only for successful audio generation. This ensures your dataset generation costs remain accurate, as you do not pay for unusable audio files resulting from transient errors or processing issues.
Can I use the OpenAI Python SDK with this API?
Yes. Our API accepts the same request shape as the OpenAI speech endpoint. You can use the official OpenAI Python SDK by setting base_url to https://api.playvn500.com/v1 and providing your SynthCode API key. This enables seamless integration into your existing data ingestion and model training workflows without requiring code rewrites.
Is my input text used to train your models?
No. Input text is used solely to generate audio and is not used for model training. The generated audio is returned in the response and not stored. We retain only request metadata—such as time, character count, and status—for billing purposes, ensuring your input data remains private and secure.
Do I need to pay a monthly fee to use the API?
No. SynthCode uses a pay-as-you-go model with no monthly plans, subscriptions, or minimum spend. You pay $0.01 per 1,000 characters generated. Balances never expire, and you can top up via USTRC-20 whenever needed. This structure aligns with the variable costs of large-scale synthetic speech dataset projects, ensuring you only pay for actual usage without recurring fees.
How fast is the API for generating long audio files?
Generation runs at roughly real time, producing about 25 characters of text per second of audio. Since requests are processed sequentially, long texts scale linearly with duration. Under load, requests may wait in a queue. While short phrases take a few seconds, longer files generate at a consistent rate relative to their length, ensuring predictable throughput for large-scale dataset creation.
What audio formats are supported for dataset storage?
The API supports mp3, wav, opus, flac, and pcm formats via the response_format parameter. These options allow you to balance file size and quality for your synthetic speech datasets. Note that generated audio is returned in the response and not stored by us, so you control the storage format directly in your ingestion pipeline.
Can I change my API key if it gets compromised?
Yes. You can issue a new key at any time from your account dashboard. When a new key is issued, the old one stops working immediately. This allows you to quickly revoke access if a key is compromised without needing to contact support.
Is there a free trial available for testing the API?
Yes. Signing in with Google provides $0.50 of credit, valid for 7 days, covering approximately 50,000 characters. Accounts created with email and password start without trial credit. This allows you to test the API’s quality and integration before committing to a top-up.
How do I top up my account balance?
Top-ups use USDT (TRC-20) on the TRON network only. The minimum amount is $5, with credits added automatically within minutes. We do not accept cards, PayPal, or invoices. This direct method suits global product teams generating large-scale synthetic speech datasets, ensuring simple, uninterrupted integration into your ML workflows without administrative overhead.
First request
curl
curl https://api.playvn500.com/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 SynthCode for ML Data Pipelines
SynthCode delivers a reliable character-based pricing model tailored for scalable data collection, charging $0.01 per 1,000 characters. This structure eliminates monthly commitments and minimum spend requirements, ensuring costs scale linearly with your dataset generation volume. Failed requests incur zero charges, protecting your budget during large-scale ingestion pipelines.
Character-based pricing for scalable data collection
Pricing is strictly $0.01 per 1,000 characters, including spaces and punctuation. This predictable cost model allows product teams to forecast expenses accurately without paying for unused monthly seats or enterprise contracts. You only pay for the audio generated, making it ideal for variable workloads.
Direct Python SDK compatibility via OpenAI base_url
The API follows the standard OpenAI request shape, allowing you to use the official OpenAI Python SDK by simply changing the base_url to https://api.playvn500.com/v1. This approach works for any language supporting the standard HTTP client, not just Python. It requires no proprietary library or custom wrapper.
Inaudible watermarking for synthetic speech identification
Every generated audio file carries an inaudible watermark that marks it as synthetic speech. This feature helps data teams distinguish generated content from human recordings in training sets without altering the audio quality. The watermark is embedded directly into the output file during synthesis.
Voices and Languages for Diverse Datasets
We provide eight distinct voice presets—four female (Aria, Nova, Luna, Sage) and four male (Felix, Milo, Atlas, Orion)—designed for robust synthetic speech datasets. These voices, based on licensed recordings from LibriTTS-R and LibriVox, offer varied tonal qualities suitable for diverse model training scenarios, from calm instructional content to energetic promotional clips.
8 distinct voice presets (male and female)
Each preset delivers consistent audio quality, ensuring uniformity across your generated datasets. The voices are engineered to handle various narrative styles, allowing product teams to select the optimal tone for specific ML tasks without compromising on clarity or naturalness.
- Aria: Warm, clear narrator
- Nova: Bright, upbeat
- Luna: Soft, storytelling
- Sage: Calm, instructional
- Felix: Energetic, promo
- Milo: Friendly, conversational
- Atlas: Neutral, news
- Orion: Deep, documentary
Support for 23 languages including English, Chinese, Arabic
Our API supports 23 languages, enabling the creation of truly multilingual datasets. Any voice can speak any supported language, 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. This flexibility ensures consistent voice characteristics across linguistic boundaries, a critical factor for training robust speech recognition and NLP models.
Consistent audio quality across all voice types
Regardless of the selected voice or language, the output maintains high fidelity. This consistency simplifies data preprocessing and model training, as variations in audio quality are minimized. Each generated file also includes an inaudible watermark, allowing you to easily identify synthetic speech within your datasets.

Technical Specifications for Engineers
SynthCode adopts the standard OpenAI-compatible request structure, allowing teams to integrate seamlessly into existing Python workflows. By simply updating the base_url to https://api.playvn500.com/v1, developers can utilize the official OpenAI SDKs without rewriting core logic. The API accepts a JSON body with fields for model, input, voice, response_format, and speed, ensuring compatibility with established data ingestion pipelines.
Standard OpenAI-compatible request structure
The endpoint POST https://api.playvn500.com/v1/audio/speech processes text inputs up to 4096 characters. For longer texts, split the input and join the resulting audio files. The model field accepts "tts-1" or "tts-1-hd", which share the same engine and pricing. Authentication is handled via the Authorization: Bearer header using your unique API key.
Output formats: MP3, WAV, OPUS, FLAC, PCM
Choose from MP3, WAV, OPUS, FLAC, or PCM using the response_format parameter. These formats cover most machine learning dataset requirements, balancing compression efficiency with raw audio fidelity. PCM provides uncompressed data ideal for immediate processing, while OPUS and MP3 offer smaller file sizes for storage efficiency.
Rate limits: 60 requests per minute per key
New accounts without a top-up are limited to 10 requests per minute. Once the first top-up is completed, the limit increases to 60 requests per minute per key. Requests are processed sequentially, so under heavy load, generation may queue slightly behind real-time speed.
Integration Workflow
SynthCode integrates directly into existing ML data pipelines using standard HTTP protocols. Authentication relies on a Bearer token passed in the Authorization header, ensuring secure access without the overhead of complex SDK dependencies. This approach allows product teams to embed generation logic into any Python script or data ingestion framework with minimal configuration.
Authenticate with Bearer token in Python scripts
Include your API key in the request header as Authorization: Bearer tts-.... This standard mechanism ensures that only authorized accounts can consume characters from your balance. Keys can be rotated instantly via the dashboard if compromised, providing secure control over your dataset generation pipeline.
- Use
Authorization: Bearer tts-...header - Rotate keys instantly via dashboard
- Standard HTTP authentication
Send POST requests to /v1/audio/speech endpoint
Direct your requests to POST https://api.playvn500.com/v1/audio/speech with a JSON body containing the model, input text, voice, and format. The endpoint accepts up to 4096 characters per request, allowing you to batch process text segments efficiently. You can specify parameters like speed and language to tailor the output for your specific training data needs.
- Endpoint:
POST https://api.playvn500.com/v1/audio/speech - JSON body with model, input, voice, format
- Max 4096 characters per request
Receive audio files directly in response body
The API returns the synthesized audio file directly in the response body, eliminating the need for separate download steps. Supported formats include MP3, WAV, OPUS, FLAC, and PCM, allowing you to store data in the format best suited for your machine learning models. This direct delivery streamlines the integration of synthetic speech into your training datasets.
- Audio returned in response body
- Formats: MP3, WAV, OPUS, FLAC, PCM
- No separate download step required
Data Privacy and Usage
Our API processes input text exclusively to produce the requested audio file. Unlike many competitors, we do not use your source text to train or improve our underlying speech synthesis models, ensuring your proprietary data remains distinct from our training corpus.
Input text used only for synthesis, not for training
Your text is transmitted to our engines solely for the purpose of generating the audio response. Once the synthesis is complete, the text is not retained for model improvement or future training cycles, preserving the integrity of your specific dataset.
Generated audio not stored on our servers
The synthesized audio files are returned directly in the response body and are not permanently stored on our infrastructure. This ephemeral handling minimizes data exposure and aligns with strict privacy requirements for sensitive voice data in ML pipelines.
Metadata kept for billing and usage tracking
We retain only request metadata, including timestamp, character count, and status, to accurately calculate charges and monitor usage. This limited retention ensures transparent billing without storing the actual content of your generated audio or source text.
Pricing Model for Data Teams
SynthCode uses a straightforward character-based pricing model, charging $0.01 per 1,000 characters. Spaces and punctuation count as characters, making costs predictable for large-scale dataset generation. Failed requests are free, ensuring you only pay for successful synthesis.
There are no monthly plans, seats, or minimum spend requirements. You pay only for what you use, with no expiration on your balance. This flexibility supports variable workloads typical in ML data pipelines.
$0.01 per 1,000 characters generated
The cost is calculated based on the input text length, including spaces and punctuation. This linear pricing scales directly with your dataset size, allowing precise budget forecasting for training runs.
- Characters include spaces and punctuation
- $0.01 per 1,000 characters
- No hidden fees
No monthly plans or minimum spend requirements
Unlike traditional SaaS models, we do not lock you into subscriptions or require minimum commitments. Your balance does not expire, allowing you to top up only when needed.
- No monthly plans
- No minimum spend
- No expiration on balance
Failed requests are charged zero credits
If synthesis fails, the charge is returned in full. You are only billed for successfully generated audio files, reducing risk during experimental phases.
- Failed requests are free
- Full credit return
- Only pay for success

Use Cases for Synthetic Data
Synthetic speech APIs enable product teams to generate massive, labeled audio datasets for machine learning without human recording bottlenecks. By leveraging 8 distinct voice presets and 23 languages, teams can create diverse training data for speech recognition models and voice assistants. The character-based pricing model ensures cost-effective scaling for large-scale dataset generation.
Training voice assistants with diverse accents
Use eight voice presets like Aria or Felix to simulate varied vocal styles for voice assistant training. Each voice offers a distinct tone, from warm narration to energetic promo, allowing you to build robust models that handle different vocal characteristics without recording hundreds of humans.
- 8 distinct voice presets with unique tones
- Consistent audio quality across all voices
- Support for 23 languages for global coverage
Generating labeled audio for speech recognition models
Generate labeled audio datasets at scale by sending text inputs and receiving synchronized audio files. With character-based pricing at $0.01 per 1,000 characters, you can produce millions of audio samples for training speech recognition models without monthly fees or minimum spend requirements.
- Character-based pricing for scalable data collection
- No monthly plans or minimum spend
- Failed requests are free
Creating multilingual audio datasets for NLP
Create multilingual audio datasets for NLP by specifying the language parameter in your API requests. Support for 23 languages, including English, Chinese, and Arabic, allows you to generate diverse training data for natural language processing models without needing native speakers for each language.
- Support for 23 languages
- Optional language parameter for precise control
- Inaudible watermarking for synthetic speech identification
Getting Started with Python
Create an account on playvn500.com using your Google credentials or an email and password combination. This registration process provisions your unique API key, which serves as the authentication credential for all subsequent requests. You can manage your key directly from the dashboard, where you can also monitor your usage metrics and balance.
Once registered, access your API key from the dashboard to begin integration. The key allows you to authenticate via the Authorization: Bearer header. You can test connectivity immediately using the provided curl example or by configuring the OpenAI Python SDK with the base_url pointing to https://api.playvn500.com/v1.
Sign up via Google or email for API access
Visit the /get-api-key/ page to register. Google sign-ups include a 7-day, $0.50 trial credit, while email-based accounts start without trial credit. This simple setup ensures you have immediate access to the API for testing your data pipelines.
Retrieve your API key from the dashboard
After logging in, the dashboard displays your active API key, current balance, and character usage. You can issue a new key at any time, which immediately invalidates the old one. This ensures secure access control for your production environments.
Use the provided curl example or Python SDK
Use the standard OpenAI Python SDK by setting the base_url to https://api.playvn500.com/v1. Send a POST request to /v1/audio/speech with your text, voice selection, and desired format. The API returns the audio file directly in the response body, ready for immediate ingestion into your ML dataset.
- Set
base_urltohttps://api.playvn500.com/v1 - Authenticate with
Authorization: Bearer <your-key> - Send
POSTrequests to/v1/audio/speech
Support and Documentation
SynthCode prioritizes clarity for engineering teams integrating synthetic speech into ML pipelines. Our documentation provides comprehensive references for all API endpoints, ensuring you can programmatically access voice configurations, model details, and audio generation parameters without guesswork.
Email support for technical inquiries
For direct assistance, reach out via email. We do not offer phone lines or live chat, ensuring focused, asynchronous support for integration questions and billing details.
Detailed documentation for API endpoints
Our docs detail every endpoint, including POST /v1/audio/speech for generation and GET /v1/voices for cataloging available assets. This allows seamless integration with standard HTTP clients or the OpenAI Python SDK by simply adjusting the base URL.
Clear limits and error handling guidelines
We provide explicit rate limits (60 requests per minute) and character constraints (4096 per request) to help you design robust retry logic. Error responses clearly indicate failures, ensuring failed requests incur no charge and your data pipeline remains cost-effective.

Frequently Asked Questions
This section clarifies common operational questions for teams integrating synthetic speech into ML pipelines. It covers pricing mechanics, language support, technical compatibility, and data handling specifics to ensure smooth dataset generation workflows.
Understanding rate limits, format options, and trial conditions helps engineering teams plan their data ingestion strategies accurately without unexpected costs or downtime.
Is this API suitable for generating large-scale synthetic speech datasets for machine learning?
Yes. The API is designed for scalable data collection with character-based pricing and direct Python SDK compatibility. It integrates seamlessly into data ingestion workflows for model training.
How does the character-based pricing work for bulk data generation?
Pricing is $0.01 per 1,000 characters, including spaces and punctuation. Failed requests are free, ensuring you only pay for successful synthesis.
Which programming languages can I use to call this TTS API?
You can use any language that supports HTTP requests. The official OpenAI SDKs work by changing the base_url to https://api.playvn500.com/v1.
Does the API support multiple languages for multilingual dataset creation?
Yes. It supports 23 languages including English, Chinese, and Arabic. Any voice can speak any supported language, and the language can be explicitly set via the ISO code.
Are the generated audio files watermarked to indicate they are synthetic?
Every generated file carries an inaudible watermark that marks it as synthetic speech, aiding in dataset identification and compliance.
What happens if a request fails during audio generation?
If synthesis fails, the charge is returned in full. You are not billed for failed requests.
Can I use the OpenAI Python SDK with this API?
Yes. The request shape matches the OpenAI speech endpoint. You can use the OpenAI Python SDK by setting the base_url to https://api.playvn500.com/v1.
Is my input text used to train your models?
No. Input text is used only to produce the audio and is not used for training. Generated audio is returned in the response and not stored.
Do I need to pay a monthly fee to use the API?
No. There are no monthly plans, seats, or minimum spend requirements. Balance does not expire.
How fast is the API for generating long audio files?
Typical speed is roughly real time (25 characters per second). Requests are processed sequentially, so under load, a request may wait in a queue.
What audio formats are supported for dataset storage?
Supported formats are mp3, wav, opus, flac, and pcm.
Can I change my API key if it gets compromised?
Yes. One active key is issued per account. A new key can be issued any time, and the old one stops immediately.
Is there a free trial available for testing the API?
Yes. Google sign-ups receive $0.50 credit (50,000 characters) valid for 7 days. Email sign-ups start without trial credit.
How do I top up my account balance?
Top up using USDT on the TRON network (TRC-20). Minimum $5, credited automatically within minutes. No cards or PayPal.
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. High quality synthetic speech data enhances the accuracy and naturalness of our machine learning models.
23 languages
Any voice speaks any of these languages.
