Frequently asked questions
Is the Hindi text-to-speech free for commercial use?
There is no free tier; you receive a $0.50 free trial valid for 7 days when signing in with Google. You receive a $0.50 free trial (about 50,000 characters) valid for 7 days when signing in with Google. Standard pricing is $10 per 1 million characters. Failed requests are free, but successful generations incur charges based on character count.
How many characters can I process per request?
You can process up to 4,096 characters per request. For longer texts, you must split the input into parts of this size and join the resulting audio files. Spaces and punctuation count toward this character limit.
Do you support SSML for dataset control?
No, we do not support SSML. The API accepts plain text input. If you need specific control over pronunciation or pacing, you must adjust the text content or use the speed parameter, which ranges from 0.25 to 4.0.
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 without any risk of being charged for errors.
Are the voices cloned from real humans?
Yes, the voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and public-domain audiobook readings from LibriVox. Every generated file carries an inaudible watermark marking it as synthetic speech.
Can I use the audio files in my ML training set?
Yes, the 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 by us, making the output suitable for your synthetic speech datasets.
Do you store the input text I send?
No, input text is used only to produce the audio and is not used for training. We keep request metadata (time, characters, status) for billing purposes, but the actual text content is not stored after generation.
How do I top up my balance?
Top up using USDT on the TRON network (TRC-20) only. The minimum amount is $5. Credits are automatically added within minutes after the transfer. We do not accept cards, PayPal, or subscriptions, and balance does not expire.
What is the speed of generation?
Generation is roughly real-time on long texts, processing about 25 characters of text per second of audio. Short phrases may take a few seconds. Requests are processed sequentially, so under load, a request may wait in a queue.
Is there a limit on the number of API keys?
Yes, there is one active key per account. You can issue a new key at any time, but the old one stops working immediately. This ensures secure access control for your team's API usage.
Do you offer enterprise contracts?
No, we do not offer enterprise contracts, SLAs, or uptime guarantees. We provide a simple API with no monthly plans, seats, or minimum spend. Support is available by email only, with no phone number or live chat.
Can I customize the voice pitch?
No, you cannot customize the voice pitch. You can only adjust the speed (0.25–4.0) and select from eight voice presets. There are no emotion tags, SSML, or custom pronunciation lexicons available for fine-tuning the audio output.
Is there a free tier or only a trial?
There is no free tier, only a free trial. You receive $0.50 of credit (about 50,000 characters) valid for 7 days when signing in with Google. Accounts created with email and password start without this trial credit.
Which languages are supported for Hindi TTS?
Hindi is one of our 23 supported languages. Other supported languages include Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. Any voice can speak any of these languages.
First request
curl
curl https://api.playzbzb.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.mp3Generate Hindi TTS for ML Training Data
Generating synthetic speech datasets requires reliable infrastructure that handles volume without compromising data consistency. Our API accepts text via a simple POST request, processing inputs up to 4,096 characters per call. This structure allows teams to batch-process large corpora efficiently, ensuring every audio file aligns precisely with the source text for accurate model training.
High-volume character generation
Teams can generate audio at scale by splitting longer texts into 4,096-character chunks and joining the resulting files. With a base rate of $0.01 per 1,000 characters, you pay only for what you use, making it economical to produce thousands of hours of training data without fixed monthly overhead or minimum spend requirements.
Consistent synthetic voices
Maintaining voice consistency across your dataset is critical for model performance. You can select from eight distinct presets, such as Aria’s warm narration or Felix’s energetic tone. Since every generated file includes an inaudible watermark marking it as synthetic, you can easily filter or identify synthetic data during your training pipeline.
Direct API response
Audio files are returned directly in the API response, allowing immediate integration into your data collection scripts. While typical generation speed is roughly real-time, short phrases may take a few seconds to process. Requests are handled sequentially, so high loads may introduce queue delays, but failed requests incur no charge.
Supported Languages for Datasets
Our API supports Hindi alongside 22 other languages, ensuring your synthetic speech datasets cover a broad linguistic range. You can generate consistent audio in any of these languages using any of the eight available voices, which simplifies data collection for multilingual models.
Hindi support details
Hindi is fully supported with all eight voice presets available. This allows you to generate diverse synthetic speech data for training NLP models in Indian languages without needing separate voice engines.
Other 22 languages
Beyond Hindi, we support Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. If a language is not on this list, it is not supported.
- Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Turkish
Language guessing logic
If you do not specify the language field, the API guesses the language from the text script. Latin script defaults to English, so you should explicitly set the language parameter for non-English Latin text to ensure accurate synthesis.
Voice Selection for Synthetic Data
Choosing the right voice is critical for generating consistent synthetic speech datasets. Linguicore offers eight distinct presets trained on licensed recordings, ensuring your ML models receive uniform audio characteristics across large-scale training runs.
Female presets (Aria, Nova)
Aria provides a warm, clear narrator tone ideal for standard instructional content. Nova offers a bright, upbeat delivery suitable for engaging promotional materials or dynamic dialogue datasets.
Male presets (Felix, Milo)
Felix delivers energetic, promo-style audio perfect for advertising datasets. Milo provides a friendly, conversational tone that mimics natural human interaction, excellent for chatbot or assistant training data.
Neutral tones (Atlas, Orion)
Atlas offers a neutral, news-style voice for clean, objective speech data. Orion provides a deep, documentary tone suitable for authoritative or narrative datasets requiring distinct vocal depth.
API Integration for Teams
Teams building synthetic speech datasets for machine learning can integrate Linguicore’s API using standard tools. The endpoint mirrors the OpenAI speech structure, allowing you to keep your existing codebase and simply swap the base URL to https://api.playzbzb.com/v1. This means official OpenAI SDKs work out of the box, reducing integration friction for large engineering teams.
OpenAI SDK compatibility
You can use the official OpenAI SDKs by updating the base_url to our API endpoint. This ensures consistent behavior with libraries your team already knows, avoiding the need to learn a new client library from scratch.
Simple POST requests
Send text via a standard POST request to /v1/audio/speech. Provide your model (tts-1 or tts-1-hd), input text, voice selection, and desired format in the JSON body. The API processes requests sequentially, so under load, requests may queue, but typical speed is roughly real-time for long texts.
JSON response format
The API returns audio files directly in the response body, formatted according to your specified response_format (mp3, wav, opus, flac, or pcm). Metadata such as request time and character count is tracked for billing but not stored in the audio file itself, keeping your pipeline clean and efficient.
Audio Formats for ML Pipelines
Synthetic speech datasets require specific audio containers to match downstream ML training pipelines. Our API supports multiple formats, allowing you to choose the balance between compression efficiency and data fidelity without reprocessing.
MP3 compression
MP3 is ideal for reducing storage costs in large-scale datasets. It offers high compression ratios while maintaining sufficient audio quality for most speech recognition models, keeping your training data storage lean.
WAV for raw data
WAV provides uncompressed, lossless audio files. This format is preferred for raw feature extraction or when your model requires exact waveform data without any encoding artifacts, ensuring maximum fidelity.
FLAC and OPUS options
FLAC offers lossless compression, preserving full audio quality while reducing file size compared to WAV. OPUS provides efficient compression suitable for various models, balancing bitrate and audio clarity effectively.

Pricing for Large Teams
Linguicore bills strictly on usage, charging $0.01 per 1,000 characters. Failed requests are free, so you only pay for successful synthesis. This model scales efficiently with your dataset size.
Per character billing
We charge $10 per million characters, including spaces and punctuation. You can generate large volumes of synthetic speech without worrying about per-request overhead or hidden fees.
No monthly plans
There are no monthly subscriptions, seats, or minimum spend requirements. You can pause usage at any time without losing access or paying for idle capacity.
USDT top-ups
Top up your balance using USDT on the TRON network (TRC-20). Minimum deposits are $5, and credits are available within minutes. No cards or invoices required.
- Minimum top-up: $5
- Network: TRON (TRC-20)
- Credits valid indefinitely
Free Trial for Evaluation
Evaluate our Hindi TTS API for synthetic speech datasets with a free trial. Sign up with Google to receive $0.50 in credit, valid for 7 days, which covers approximately 50,000 characters. This allows your team to test audio quality and consistency without immediate payment setup.
If you prefer a traditional account, you can register using an email and password. Note that email sign-ups start without trial credit, so you will need to add funds via USDT on TRON to begin generating audio files for your machine learning models.
Google sign-up bonus
Linking your Google account instantly activates a $0.50 credit. This bonus is valid for 7 days and helps you verify that our synthetic voices meet your dataset requirements before committing to larger volumes.
- Instant activation upon Google sign-in
- $0.50 credit added automatically
- Valid for 7 days
50,000 characters free
The free credit provides roughly 50,000 characters of generation time. This volume is sufficient to generate multiple short phrases or a few longer segments, allowing you to assess latency and audio clarity across different voices.
Email sign-up options
You can also create an account using just an email address and password. This method is ideal for teams that prefer standard credentials, though it does not include the initial trial credit found with Google sign-ups.
Data Usage and Privacy
We prioritize data integrity for your machine learning pipelines. Input text is used strictly to generate the corresponding audio file and is never repurposed for training our own models. This ensures your proprietary prompts remain isolated and secure within our processing environment.
No training on input text
Your source text is consumed solely for synthesis. We do not ingest your inputs into any training dataset, preserving the uniqueness of your synthetic speech data.
Metadata retention
We retain request metadata, including timestamps, character counts, and status codes, exclusively for accurate billing and usage tracking. This minimal data ensures transparent accounting without storing your actual content.
Synthetic watermarking
Every generated audio file contains an inaudible watermark marking it as synthetic speech. This allows downstream models to easily distinguish between human-recorded and API-generated data during training.
Why Choose This API for Datasets
Linguicore is built for teams generating synthetic speech datasets at scale. We provide a straightforward API that integrates easily into your ML pipelines, allowing you to focus on model training rather than infrastructure management.
Cost-effective scaling
Generate consistent, high-volume audio without complex infrastructure. Our API handles large requests efficiently, delivering synthetic voices in 23 languages with predictable performance metrics.
- Simple POST requests for bulk generation
- Consistent voice quality across languages
- No minimum spend requirements
No voice cloning limits
We offer eight fixed presets. Each voice is based on licensed recordings, ensuring legal clarity and consistent synthetic characteristics for your training data.
- Eight licensed voice presets
- No per-voice cloning fees
- Inaudible watermarking for provenance
Transparent billing
Pay only for what you use. Billing is strictly per character, with failed requests refunded instantly. Top-ups via USTRC-20 are processed automatically, keeping your workflow uninterrupted.
- $0.01 per 1,000 characters
- No monthly subscriptions
- Instant refund on failure
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. This ensures data provenance and prevents accidental mixing of human and synthetic samples during model training.
23 languages
Any voice speaks any of these languages.
