Generate Realistic NPC Dialogue with Our Text-to-Speech API

By EchoQuest editorial team · Updated 2026-11-18

This guide helps freelance game developers integrate EchoQuest’s text-to-speech API for NPC dialogue. You will learn how to generate audio files that fit directly into your existing audio workflow with minimal code changes. The API uses the same request shape as OpenAI, so you can reuse your existing SDKs by simply updating the base URL.

We focus on practical steps for localization, voice consistency, and cost management. You will see how to handle long texts, choose the right audio format, and manage rate limits effectively. This page is designed for engineers who need reliable, character-based pricing and clear technical constraints without unnecessary complexity.

Generate Realistic NPC Dialogue with Our Text-to-Speech API

Frequently asked questions

How do I choose a voice?

Select from eight presets based on your NPC’s role. Female options include Aria, Nova, Luna, and Sage. Male options include Felix, Milo, Atlas, and Orion. Each voice has a distinct tone, from warm narration to deep documentary styles. All voices support 23 languages, so choose the one that best fits the character’s personality and context.

What is the best format for games?

We support mp3, wav, opus, flac, and pcm. For most games, mp3 or opus offers a good balance of quality and file size. Use wav or flac for lossless audio during post-processing. Choose pcm for raw data streams. Your format choice depends on your engine’s audio loader and memory constraints, so select the option that fits your specific pipeline requirements.

How do I split long texts?

The API limits input to 4096 characters per request. Split your script into chunks of this size. Generate audio for each chunk individually, then join the resulting audio files in your game engine or post-production pipeline. This ensures reliable synthesis and avoids truncation errors in your dialogue lines.

Why is my request queued?

Requests are processed sequentially. Under load, your request may wait in a queue before synthesis begins. This is normal during high traffic. Longer texts also take time to generate, roughly real-time speed. Short phrases may take a few seconds. Patience ensures your audio is generated correctly without errors.

How do I pay with USDT?

Top up using USDT on the TRON network (TRC-20). The minimum deposit is $5. Credits are added automatically within minutes of your transfer. We do not accept cards, PayPal, or subscriptions. Your balance does not expire. Send funds to the address provided in your account dashboard to start generating audio.

Can I use voices in commercial games?

Yes, you can use the generated audio in commercial games. The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain readings. Each file includes an inaudible watermark marking it as synthetic speech. There are no additional licensing fees for commercial use beyond the API usage cost.

Do voices speak all languages?

Voices support 23 languages, including English, Spanish, German, Japanese, and Chinese. Any voice can speak any of these languages. If you do not specify a language, the system guesses it, defaulting to English for Latin script. For non-English Latin text, explicitly set the language parameter to ensure accuracy.

Is the API fast enough for real-time?

Generation speed is roughly real-time for long texts, processing about 25 characters per second. Short phrases may take a few seconds. Requests are processed sequentially, so under load, you may experience queuing delays. For strict real-time needs, consider pre-generating audio assets rather than relying on on-the-fly synthesis.

How do I detect language correctly?

Set the optional language field in your request to an ISO code. If omitted, the system guesses the language, defaulting to English for Latin script. Explicitly setting the language ensures correct pronunciation for non-English texts, especially those using Latin characters but spoken in other languages like Dutch or Malay.

What happens if I fail a request?

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. Check your request parameters, such as text length and voice availability, if failures persist. Failed requests do not consume your character quota.

Do I need a custom SDK?

No, you do not need a custom SDK. The API endpoint matches the OpenAI speech endpoint structure. You can use the official OpenAI SDKs by changing the base_url to https://api.play33ok.com/v1. Plain HTTP requests also work perfectly. This allows easy integration into existing workflows without learning a new library.

Can I clone my own voice?

No, we do not offer custom voice cloning. You cannot upload your own voice or design a custom voice. You must choose from the eight available presets: Aria, Nova, Luna, Sage, Felix, Milo, Atlas, or Orion. These voices are based on licensed recordings and are ready for immediate use.

Is there an SSML option?

No, we do not support SSML. The API accepts plain text input. You cannot use SSML tags for pronunciation, pauses, or emphasis. If you need specific timing or pronunciation, you must split your text into smaller chunks and manage the timing in your game engine or post-production pipeline.

How do I manage rate limits?

The limit is 60 requests per minute per key. Before your first top-up, the limit is 10 requests per minute. Monitor your usage in the dashboard. If you exceed the limit, requests may be delayed or rejected. Implement exponential backoff in your client to handle rate limit errors gracefully.

Do you store my audio?

No, generated audio is returned in the response and not stored. Input text is used only to produce the audio and is not used for training. We keep request metadata, such as time, characters, and status, for billing purposes. Your audio files are transient unless you save them from the response.

Can I change my API key?

Yes, you can issue a new key at any time from your account dashboard. When you issue a new key, the old one stops working immediately. This allows you to revoke access if a key is compromised. You can have only one active key per account at any given time.

Is there a minimum spend?

No, there is no minimum spend. You can top up any amount, with a minimum of $5 for USDT deposits. There are no monthly plans or subscriptions. You pay only for the characters you generate. This model allows you to scale your spending directly with your game’s needs.

Do balances expire?

No, your balance does not expire. Once you top up with USDT, the credit remains in your account indefinitely. You can generate audio whenever you need it. There is no usage deadline or expiration date for your funds. This allows you to plan your audio production schedule flexibly.

How do I get a free trial?

Sign in with Google to receive $0.50 of trial credit, valid for 7 days. This covers about 50,000 characters. Accounts created with email and password do not receive trial credit. Use the trial to test the API, voices, and formats before committing to a top-up for your game development.

Is support available 24/7?

No, we do not offer 24/7 support or live chat. Support is available via email only. There is no phone number for direct contact. For urgent issues, check the documentation first. Email support is the primary channel for resolving account, billing, or technical questions related to the API.

Can I use this for VR games?

Yes, the API is suitable for VR games. You can generate NPC dialogue and environmental audio cues. Use the speed parameter to adjust pacing. The generated audio files can be integrated into your VR engine. Ensure you handle audio streaming or pre-loading appropriately for your VR platform’s requirements.

Do you support mobile platforms?

We do not have a dedicated mobile SDK, but the API works on any platform that supports HTTP requests. You can use the OpenAI SDK or plain HTTP from iOS, Android, or other mobile environments. Integrate the API call into your game’s backend or client-side logic as needed.

Is there a bulk discount?

No, there are no bulk discounts or volume-based pricing tiers. The price is fixed at $10 per 1 million characters. This rate applies regardless of how many characters you generate. You can optimize costs by choosing efficient formats like mp3 or opus to reduce file sizes and storage needs.

Can I get an invoice?

No, we do not provide invoices or VAT receipts. Payments are made via USDT on the TRON network. You can view your transaction history and usage details in the dashboard. For accounting purposes, you may use the dashboard logs as proof of payment and usage for your records.

Do you have a dashboard?

Yes, the dashboard is available at /get-api-key/. It shows your API key, balance, trial status, characters left, usage history, and a ready-to-use curl example. You can also issue a new API key from this page. The dashboard provides all the essential information for managing your account and usage.

How do I track usage?

Track usage via the dashboard, which displays characters generated and balance remaining. Request metadata, including time, characters, and status, is kept for billing. You can also monitor your own request logs in your application. The dashboard provides a clear overview of your consumption and helps you manage your budget effectively.

Can I use multiple keys?

No, each account has only one active API key. You can issue a new key at any time, but the old one stops working immediately. This ensures that only one key is valid per account. If you need multiple keys for different projects, you may need to create separate accounts.

Is the API secure?

Yes, the API uses standard Bearer token authentication. Send your key in the Authorization: Bearer tts-... header. Ensure you keep your key secure and do not expose it in client-side code if possible. All requests are made over HTTPS, ensuring data transmission security between your application and our servers.

Do you offer SLA?

No, we do not offer an SLA or uptime guarantee. The API is provided as-is without enterprise contracts. There are no on-premise options or data-center choices. For critical production use, consider implementing retry logic and fallback mechanisms in your application to handle potential service interruptions or latency spikes.

Can I use this for AI assistants?

Yes, the API is suitable for AI assistants. Generate voice responses for chatbots or voice interfaces. Use the speed parameter to adjust the pacing of responses. The API supports multiple languages, allowing you to localize your AI assistant. Ensure you handle text splitting and audio joining for longer responses.

How do I handle errors?

Check the API response for error codes. Common issues include rate limits, invalid parameters, or network errors. Implement retry logic with exponential backoff for transient errors. If a request fails, the charge is returned. Review your input text length and voice selection to ensure they meet the API requirements.

What is the generation speed?

Generation speed is roughly real-time for long texts, processing about 25 characters per second. Short phrases may take a few seconds. Requests are processed sequentially, so under load, you may experience queuing delays. This speed is suitable for pre-generated audio but may require buffering for real-time applications.

Can I adjust volume?

The API does not support volume adjustment directly. You can adjust the volume in your game engine or audio player after receiving the audio file. The speed parameter controls the playback speed, not the volume. Use your engine’s audio mixer to normalize or adjust volume levels for consistent user experience.

Do you support DTMF?

No, we do not support DTMF (Dual-Tone Multi-Frequency) tones. The API generates speech audio only. If you need DTMF tones for IVR systems, generate them separately or use your engine’s audio capabilities. The API focuses on text-to-speech conversion for natural language dialogue and narration.

Is there a demo?

No, there is no dedicated demo page. You can test the API directly using the curl example provided in the dashboard. Sign in with Google to get trial credit and test the endpoint. Use the /v1/voices endpoint to list available voices and test different parameters with your own text.

Can I test with my own text?

Yes, you can test with your own text. Use the input field in the API request to send your script. The text limit is 4096 characters per request. You can test different voices, languages, and formats to find the best fit for your project. Use the trial credit to test without cost.

Do you have a GitHub repo?

No, we do not have an official GitHub repository. The API is accessed via standard HTTP requests. You can use the OpenAI SDK or plain HTTP libraries. Documentation is available at /docs/. There are no code samples or libraries hosted on GitHub specific to our API.

Can I use this for podcasts?

Yes, you can use the API for podcasts. Generate narration or dialogue for your episodes. Use the speed parameter to control pacing. The audio is suitable for podcast distribution. You can generate multiple episodes or segments and join them for final production. Ensure you comply with any licensing requirements for the voices.

Is the audio high quality?

The audio is generated from licensed recordings. Voices are based on LibriTTS-R and LibriVox data. The API supports high-quality formats like wav and flac. The audio includes an inaudible watermark marking it as synthetic.

Do voices sound robotic?

No, the voices are based on licensed recordings. Each voice has a distinct personality, from warm narration to energetic promo styles. The quality is comparable to modern commercial TTS services, providing a realistic listening experience for your audience.

Can I use this for ads?

Yes, you can use the API for advertisements. Generate voiceovers for radio, video, or online ads. Use voices like Felix for energetic promo or Aria for clear narration. The API supports multiple languages, allowing you to localize your ads. Ensure you comply with any licensing requirements for the voices.

Do you support WebSockets?

No, we do not support WebSockets or streaming. The API uses standard HTTP POST requests. You receive the full audio file in the response. There is no real-time partial audio stream. For real-time needs, consider pre-generating audio assets or implementing client-side buffering to handle the latency of HTTP requests.

Can I use this for education?

Yes. Generate narration for e-learning, audiobooks, or language lessons. Our API supports 23 languages, making it suitable for multilingual education. Use different voices to engage students. The API’s reliability and quality make it a good choice for professional educational materials.

Is there a developer forum?

No, there is no dedicated developer forum. Support is provided via email. You can find documentation and examples at /docs/. For technical questions, contact support by email. There are no community-driven forums or discussion boards specific to our API. Check the documentation for common issues and solutions.

Do you have case studies?

No, we do not have published case studies. The API is used by freelance game developers and other creators, but specific examples are not highlighted. You can test the API yourself to evaluate its suitability for your project. The documentation and API endpoint details provide all the technical information you need.

Can I use this for e-learning?

Q: Can I use this for e-learning? A: Yes, the API is suitable for e-learning. Generate voiceovers for courses, quizzes, and interactive lessons. Support multiple languages for global reach. Use different voices to distinguish between instructors and characters. The API’s reliability and quality ensure a professional listening experience for students. Pre-generate audio to reduce latency during lessons.

Is the API documented?

Yes, the API is documented at /docs/. The documentation covers endpoints, parameters, and examples. It also includes a ready-to-use curl example in the dashboard. The API follows the OpenAI speech endpoint structure, so existing OpenAI SDK documentation can be helpful. Review the docs for detailed information on voices, formats, and pricing.

Can I use this for social media?

Yes. Generate voiceovers for videos, reels, or posts. Use different voices to match the tone of your content. The API supports multiple languages, allowing you to reach a global audience. Ensure you comply with platform guidelines regarding synthetic voice content.

Do you support batch processing?

The API processes requests sequentially, not in parallel batches. You can send multiple requests, but they are queued. For large-scale generation, consider sending requests in parallel from multiple keys or accounts if available. The rate limit is 60 requests per minute, so plan your batch sizes accordingly to avoid delays.

Can I use this for VR?

Yes, the API is suitable for VR games. Generate NPC dialogue and environmental audio. Use the speed parameter to adjust pacing. The generated audio files can be integrated into your VR engine. Ensure you handle audio streaming or pre-loading appropriately for your VR platform’s requirements to maintain immersion.

Is there a latency issue?

Latency depends on request queuing and processing time. Requests are processed sequentially, so under load, you may experience delays. Generation speed is roughly real-time for long texts. For real-time applications, pre-generate audio assets or implement buffering. Short phrases may take a few seconds to generate, which is acceptable for most use cases.

Can I use this for IoT?

Yes, you can use the API for IoT devices. Generate voice responses for smart speakers or assistants. Use the API via HTTP requests from your IoT backend. The API supports multiple languages, allowing you to localize your IoT products. Ensure your device can handle the latency and network requirements of the API.

Do you support custom models?

No, we do not support custom models or voice cloning. You must use the eight available presets. These voices are based on licensed recordings and are ready for immediate use. There is no option to train a custom model on your data. The fixed set of voices ensures consistency and quality.

Can I use this for AR?

Yes, the API is suitable for AR applications. Generate voiceovers for augmented reality experiences. Use the API via HTTP requests from your AR app or backend. The API supports multiple languages, allowing you to localize your AR content. Ensure you handle audio streaming or pre-loading appropriately for your AR platform’s requirements.

Is there a mobile app?

No, we do not have a dedicated mobile app. The API is accessed via HTTP requests. You can use the API from any mobile application that supports HTTP. The dashboard is web-based and accessible from mobile browsers. Use the API documentation to integrate the TTS service into your mobile game or app.

Can I use this for web apps?

Yes, you can use the API for web applications. Generate voice responses for chatbots, notifications, or interactive elements. Use the API via HTTP requests from your web backend or client-side code. The API supports multiple languages, allowing you to localize your web app. Ensure you handle rate limits and errors appropriately.

Do you support iOS?

We do not have a dedicated iOS SDK, but the API works on iOS. You can use the OpenAI SDK or plain HTTP requests from your iOS app. Integrate the API call into your app’s backend or client-side logic. The API’s standard HTTP interface ensures compatibility with any iOS development environment.

Can I use this for Android?

We do not have a dedicated Android SDK, but the API works on Android. You can use the OpenAI SDK or plain HTTP requests from your Android app. Integrate the API call into your app’s backend or client-side logic. The API’s standard HTTP interface ensures compatibility with any Android development environment.

Do you support Windows?

We do not have a dedicated Windows SDK, but the API works on Windows. You can use the OpenAI SDK or plain HTTP requests from your Windows application. Integrate the API call into your app’s backend or client-side logic. The API’s standard HTTP interface ensures compatibility with any Windows development environment.

Can I use this for Mac?

We do not have a dedicated Mac SDK, but the API works on Mac. You can use the OpenAI SDK or plain HTTP requests from your Mac application. Integrate the API call into your app’s backend or client-side logic. The API’s standard HTTP interface ensures compatibility with any Mac development environment.

Do you support Linux?

We do not have a dedicated Linux SDK, but the API works on Linux. You can use the OpenAI SDK or plain HTTP requests from your Linux application. Integrate the API call into your app’s backend or client-side logic. The API’s standard HTTP interface ensures compatibility with any Linux development environment.

Can I use this for cloud?

Yes, the API is cloud-based and accessible from any cloud environment. You can use it from AWS, Azure, Google Cloud, or any other cloud provider. The API is accessed via HTTP requests, making it easy to integrate into your cloud infrastructure. Ensure you manage your API key securely in your cloud environment.

Choosing the Right API

Sound wave illustration for: Choosing the Right API

Selecting the correct API endpoint ensures your NPC dialogue pipeline runs smoothly without unnecessary complexity. EchoQuest offers an endpoint that mirrors the OpenAI speech API structure, allowing you to swap in our service with minimal code changes. This compatibility means you can leverage existing OpenAI SDKs by simply adjusting the base URL to https://api.play33ok.com/v1.

Understanding rate limits and cost structures upfront prevents integration bottlenecks during peak development or localization sprints. Our pricing is straightforward, based on character count, while rate limits ensure fair usage across all freelancers using the shared infrastructure.

OpenAI compatibility

Our endpoint accepts the same request shape as the OpenAI speech endpoint. You can use official OpenAI SDKs by setting the base_url to https://api.play33ok.com/v1. This reduces integration effort, as you only need to update your authentication header with your EchoQuest key.

This approach allows you to maintain your existing code structure while benefiting from our specific voice presets and language support tailored for game audio workflows.

  • Same JSON request body as OpenAI speech endpoint
  • Works with official OpenAI SDKs via base_url change
  • Standard Authorization: Bearer header format

Rate limits consideration

Each API key is limited to 60 requests per minute. Before your first top-up, this limit drops to 10 requests per minute. Plan your batch generation accordingly, especially when localizing large volumes of NPC dialogue.

Requests are processed sequentially. Under load, a request may wait in a queue. For real-time in-game voice generation, consider pre-generating audio files rather than relying on on-the-fly synthesis.

  • 60 requests per minute per key (standard)
  • 10 requests per minute before first top-up
  • Sequential processing may cause queue delays

Cost per character

Pricing is $0.01 per 1,000 characters, or $10 per million characters. Spaces and punctuation count toward this total. Failed requests are free, so you only pay for successful synthesis.

With a free trial credit of $0.50, you can test approximately 50,000 characters over 7 days. This allows you to evaluate voice quality and integration speed before committing to USDT top-ups.

  • $10 per 1 million characters
  • Spaces and punctuation count as characters
  • Failed requests are free

Integration Steps

Sound wave illustration for: Integration Steps

For game developers, integration begins by securing your API credentials. Navigate to the /get-api-key/ dashboard to generate your unique Bearer token. This key authenticates all subsequent requests to the speech endpoint.

Get API key

Create an account using Google or email. The dashboard displays your active key, current balance, and a ready-to-use cURL example. Keep this key secure; issuing a new one instantly invalidates the old one.

Install SDK

The API follows the OpenAI request shape, allowing you to use official OpenAI SDKs by simply changing the base_url to https://api.play33ok.com/v1. Alternatively, use any standard HTTP client to send POST requests.

Make first request

Send a POST request to /v1/audio/speech with your key, model (tts-1 or tts-1-hd), voice preset, and input text. The response returns the audio file directly, ready for immediate integration into your game engine.

Voice Selection

Sound wave illustration for: Voice Selection

Voice selection defines the personality of your NPCs. Our eight presets span warm narrators to deep documentary tones, allowing you to match the audio to your game’s genre without external design tools.

Match tone to genre

Assign specific voices to character archetypes to maintain immersion. For example, use Aria for quest givers or Felix for promotional in-game events. Each voice is distinct, ensuring players instantly recognize character roles through audio cues alone.

  • Aria: Warm, clear narrator for quests
  • Felix: Energetic tone for promos
  • Atlas: Neutral delivery for news updates
  • Orion: Deep voice for documentary-style lore

Test multiple voices

Generate short samples with different voices using the same text. Compare how Sage’s calm instructional tone contrasts with Luna’s soft storytelling style. This helps you select the best fit for specific narrative contexts before bulk generation.

Consistency across scenes

Stick to one voice per character across multiple dialogue lines. Since each request is independent, manually track which voice ID corresponds to each NPC in your script to ensure auditory consistency throughout the game experience.

Handling Long Texts

Sound wave illustration for: Handling Long Texts

For NPCs with extensive dialogue, split input text into chunks of up to 4096 characters. Each chunk triggers a separate synthesis request via the POST /v1/audio/speech endpoint.

Split into 4096 chars

Limit each request payload to 4096 characters to ensure successful processing. Count spaces and punctuation toward this limit. Process segments sequentially or in parallel, respecting your rate limits.

  • Max 4096 characters per request
  • Count all spaces and punctuation

Join audio files

Concatenate the returned audio files in the order of the original text segments. Ensure consistent sample rates and formats (e.g., wav or mp3) across all files before merging.

Manage queue waits

Requests process one after another under load. Expect delays for long texts. Implement retry logic with exponential backoff if the API returns 429 Too Many Requests.

  • Requests process sequentially under load
  • Implement retry logic with backoff

Language Localization

Sound wave illustration for: Language Localization

Localizing NPC dialogue requires precise language control. While the API can guess the language from Latin script (defaulting to English), you should explicitly set the language parameter for non-English text to ensure accurate pronunciation and avoid misinterpretation.

Set language explicitly

Always include the language field in your JSON body. This prevents the default English fallback for Latin-script inputs that aren't English, ensuring your characters sound authentic in their intended market.

Support 23 languages

Our API supports 23 languages, allowing you to localize NPC dialogue for a wide range of game markets. Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Turkish.

Detect non-Latin scripts

The system automatically detects non-Latin scripts like Cyrillic, Arabic, or CJK. You do not need to specify the language code for these scripts; the API identifies the script and applies the correct voice engine.

Audio Format Choice

Sound wave illustration for: Audio Format Choice

The API supports four audio formats—mp3, wav, opus, and flac—allowing you to balance file size, quality, and workflow needs. Choose the format that best fits your game’s audio pipeline and distribution requirements.

MP3 for distribution

MP3 is the standard for most game audio distribution due to its broad compatibility and reasonable file sizes. Use mp3 for in-game dialogue that plays directly in the engine without further processing. It balances quality and storage efficiently for final builds.

WAV for editing

WAV provides uncompressed, lossless audio ideal for post-production workflows. If your audio team needs to apply effects, mix, or adjust levels, WAV preserves the full fidelity. It is larger than MP3 but ensures no quality loss during editing.

Opus for efficiency

Opus provides superior compression efficiency, often smaller than MP3 at equivalent quality. It is excellent for high-volume dialogue trees where bandwidth and storage are critical constraints, such as in large open-world games. High compression ratio saves significant storage space. Low latency suitable for real-time streaming scenarios. Ideal for dense dialogue trees with many lines.

Cost Management

Sound wave illustration for: Cost Management

Track your character consumption and balance directly in your dashboard. The API charges $0.01 per 1,000 characters, so monitoring usage helps you predict costs for large NPC dialogue sets before they generate.

Monitor usage

Check your balance and character count on the /get-api-key/ page. Request metadata is kept for billing, allowing you to verify exactly how much each synthesis call consumed.

Top up with USDT

Add funds via USDT on the TRON network (TRC-20). Minimum top-up is $5, and credits appear automatically within minutes. Balance does not expire, and no invoices or VAT apply.

  • TRC-20 network only
  • Minimum $5 top-up
  • Automatic credit within minutes

Avoid overages

Split long texts into 4,096-character chunks to manage generation time and cost predictability. Failed requests are free, but successful ones incur the standard per-character rate regardless of audio length.

Cost Management

Technical Limits

Sound wave illustration for: Technical Limits

The API enforces a strict rate limit of 60 requests per minute per key. This ceiling is hard; exceeding it triggers a 429 error. For projects generating dialogue in batches, plan your request volume accordingly to avoid interruptions during automated workflows.

60 req/min limit

Each API key is capped at 60 requests per minute. New accounts start with a lower limit of 10 requests per minute until the first top-up is processed. Monitor your usage dashboard to stay within these bounds and prevent request rejections during peak generation times.

  • Hard limit of 60 requests/minute per key
  • New accounts start at 10 requests/minute
  • Limit resets every 60 seconds

Queue processing

Requests are processed sequentially. Under heavy load, your request may wait in a queue before synthesis begins. Expect delays for longer texts, as generation speed is roughly real-time (25 characters per second). Short phrases may take a few seconds to complete.

Retry logic

Implement exponential backoff in your client to handle temporary 429 or 503 errors. Do not retry failed requests immediately; wait briefly to avoid overwhelming the queue. Retries for successful requests may result in duplicate audio files, so idempotency checks are recommended.

  • Use exponential backoff for 429 errors
  • Avoid immediate retries to reduce queue pressure
  • Check for duplicate audio on retry

Data Privacy

Sound wave illustration for: Data Privacy

EchoQuest prioritizes developer trust by keeping your data strictly functional. Your input text is used solely for synthesis and is never ingested into training datasets, ensuring your unique NPC dialogue remains proprietary. Generated audio is returned in the response and is not persisted on our servers, minimizing exposure risk.

We retain only the essential metadata required for accurate billing and usage tracking. This includes timestamps, character counts, and request status. No voice recordings or raw text files are stored long-term, keeping your creative assets secure and your data footprint minimal.

Input not trained

All text sent to the API is processed in real-time to generate the audio file. Once the synthesis is complete, the input text is not added to any training corpus or used to improve the model’s future voice outputs. Your dialogue lines remain yours alone.

No audio storage

The generated audio file is returned immediately in the HTTP response and is not saved on our infrastructure. If your application crashes or the connection drops, the audio is lost unless you save it locally. This ensures we do not hold copies of your game’s audio assets.

This approach reduces storage costs and limits data exposure, aligning with secure game development pipelines where assets are version-controlled in your own repository.

Metadata billing

We store request metadata—such as timestamp, character count, and status—to calculate your bill accurately. This data is retained solely for billing verification and usage analytics. It does not include the content of your text or the resulting audio files.

You can review this metadata in your dashboard to monitor usage patterns and ensure your USDT top-ups align with actual consumption.

Troubleshooting

Sound wave illustration for: Troubleshooting

When synthesis fails, check the response body for specific error codes. Failed requests are free, so no charge is applied if the engine cannot process the input.

Understanding common failure modes helps you debug quickly without losing credits or wasting time on unnecessary retries.

Handle errors

If the API returns a 4xx or 5xx error, inspect the JSON response for details. Failed requests are free, ensuring you are not billed for incomplete audio files.

  • Check for invalid model or voice names.
  • Verify input text length does not exceed 4096 characters.
  • Ensure the input text is not empty.

Check rate limits

The API allows 60 requests per minute per key (10 before the first top-up). If you exceed this, requests may queue or fail with a 429 status.

Under load, requests are processed sequentially. Long texts may experience queue waits, so implement exponential backoff in your integration loop.

  • Monitor your requests per minute to avoid throttling.
  • Implement retry logic with jitter for 429 responses.
  • Split large texts into chunks to stay within limits.

Verify authentication

Ensure every request includes the Authorization: Bearer tts-... header with your active key. One key per account; issuing a new key immediately invalidates the old one.

If requests return 401 errors, confirm your key is correct and active. Check the /get-api-key/ page to verify your current key status.

  • Include the full Bearer token in the header.
  • Regenerate your key if it stops working unexpectedly.
  • Check that the base URL matches https://api.play33ok.com/v1.

Best Practices

Sound wave illustration for: Best Practices

Optimizing your NPC dialogue pipeline requires balancing generation speed with audio quality. By pre-generating lines during idle game cycles and caching frequent responses, you minimize latency during gameplay. Always validate voice outputs against actual in-game context to ensure tone consistency across scenes.

Pre-generate dialogue

Generate dialogue files during menu screens or loading states rather than at runtime. This avoids queuing delays, as requests are processed sequentially and may wait under load. Pre-generation ensures instant playback for critical NPC interactions.

  • Trigger generation during non-critical gameplay moments.
  • Batch requests where possible to utilize concurrency limits effectively.

Cache responses

Store generated audio files locally or on a CDN for repeated lines. Since input text up to 4096 characters is processed uniquely, caching identical strings prevents redundant API calls and reduces costs. Use the input hash as a cache key.

  • Cache by exact input string hash.
  • Store WAV or MP3 based on distribution needs.

Test with real text

Validate voices using actual game scripts, including edge cases like punctuation and special characters. Ensure the selected voice matches the character’s persona in the specific language. Verify that speed settings (0.25–4.0) maintain intelligibility in noisy game environments.

  • Test with full 4096-character inputs.
  • Verify audio clarity at various speed settings.

Support Channels

Sound wave illustration for: Support Channels

Our support model is designed to keep costs low for indie developers, relying entirely on asynchronous communication. When you encounter an issue or have a specific question about your integration, you can reach us directly via email. This ensures that every query is logged and addressed without the noise of live queues.

Email support

All technical inquiries and billing questions are handled via email. This is our only support channel, ensuring that your ticket is processed by engineers who understand the API’s specific quirks and limitations. We aim to respond promptly, but volume varies, so please include your API key and error logs for faster resolution.

  • Direct email access for all issues
  • Include API key and request ID for context
  • No phone or ticketing system

Read documentation

Most integration issues can be resolved by reviewing our detailed docs. We provide ready-to-use curl examples, code snippets for the OpenAI SDK, and clear explanations of rate limits and voice characteristics. The documentation is the primary source of truth for implementation details.

No live chat

We do not offer live chat or real-time support. This allows us to maintain a lean team and keep our API pricing competitive. For most developers, email support combined with thorough documentation is sufficient to unblock development and ship NPC dialogue on time.

Listen to the 8 voices

Each sample below was generated by this API from the same sentence.

Avatar of the voice aria
ariafemale
Avatar of the voice nova
novafemale
Avatar of the voice luna
lunafemale
Avatar of the voice sage
sagefemale
Avatar of the voice felix
felixmale
Avatar of the voice milo
milomale
Avatar of the voice atlas
atlasmale
Avatar of the voice orion
orionmale

23 languages

Any voice speaks any of these languages.

Sources and further reading

  1. Wikipedia
  2. MDN: Web Speech API
  3. Python documentation
  4. OpenAI API reference: create speech
  5. LibriTTS-R corpus (OpenSLR 141)
  6. Node.js documentation
  7. LibriTTS-R paper (arXiv 2305.18802)
  8. ISO 639 language codes
  9. IETF RFC 7845: Ogg Opus
  10. TRON developer hub: TRC-20
  11. openai-node on GitHub
  12. Tether (USDT)

More on this site