Text-to-Speech API for Indie Game NPC Dialogue

By EchoMouth editorial team · Updated 2026-11-19

This guide evaluates text-to-speech APIs for indie game NPC dialogue, focusing on deterministic character-limited synthesis. We analyze cost efficiency, integration complexity, and voice consistency across major cloud providers and specialized services to help you choose the right engine for your project.

Designed for freelance audio engineers and indie developers, this page provides concrete data on pricing models, latency tolerance, and batch processing requirements. You will find specific comparisons of features like streaming and SSML, alongside practical advice on keeping asset production costs predictable with transparent per-character billing. (Note: Streaming and SSML are not supported by this API, but are compared in the alternatives section.)

Text-to-Speech API for Indie Game NPC Dialogue

Frequently asked questions

Which TTS API is the cheapest?

EchoMouth charges $0.01 per 1,000 characters, including spaces and punctuation. Failed requests are free, and billing is transparent per character. Unlike subscription models, you pay only for what you use via USDT top-ups, making it cost-effective for high-volume NPC dialogue generation without monthly commitments.

Is ElevenLabs better than GCP TTS?

EchoMouth uses licensed LibriTTS-R and LibriVox voices, offering deterministic character-limited synthesis. While ElevenLabs offers advanced features, EchoMouth provides an OpenAI-compatible endpoint with explicit pricing. We do not claim superiority, but our API is optimized for game developers needing predictable asset production costs and simple integration.

Does Azure support SSML?

This is a question about Azure, not EchoMouth. EchoMouth does not support SSML. We provide a straightforward JSON API with fields for model, input, voice, response_format, and speed. If you need SSML, you must use a different provider, as our endpoint relies on simple text input.

Can I clone voices in Amazon Polly?

This question concerns Amazon Polly, not EchoMouth. EchoMouth does not offer custom voice cloning. You can choose from eight presets like Aria or Felix. These are based on licensed recordings. We do not allow uploading your own voice or designing custom voices, keeping the API simple and deterministic.

Which API has the lowest latency?

EchoMouth processes requests sequentially, so latency depends on queue load. Typical speed is roughly real-time (25 characters per second). Short phrases take a few seconds. Under heavy load, requests may wait. We do not guarantee specific latency or offer streaming, so plan for variable generation times.

Is Google Cloud TTS free?

This question refers to Google Cloud. EchoMouth offers a free trial of $0.50 credit (approx. 50,000 characters) valid for 7 days when signing in with Google. Email accounts start without trial credit. After the trial, you pay $0.01 per 1,000 characters via USDT top-ups.

Do I need a subscription for ElevenLabs?

This question concerns ElevenLabs. EchoMouth requires no subscription, seats, or minimum spend. You top up with USDT on the TRON network. Balances do not expire. You only pay for the characters you generate. This model suits indie developers who want predictable costs without recurring monthly fees or commitments.

Which API supports the most languages?

EchoMouth supports 23 languages, including English, Chinese, Spanish, and Japanese. Any voice can speak any of these languages. If you need more than 23 languages, you may need a different API. We support standard ISO codes, and the language is guessed from the script if not specified.

Can I use these APIs for mobile games?

EchoMouth is ideal for mobile games. You send text via POST requests and receive audio files. Since audio is generated server-side, you do not need local processing power on the device. This keeps your mobile app lightweight. You can generate NPC dialogue assets during development or at runtime.

Is voice cloning accurate?

EchoMouth does not offer voice cloning. We provide eight preset voices like Aria and Felix, based on licensed recordings. These are deterministic and consistent. If you need cloning, you must use a different provider. Our voices are synthetic and carry an inaudible watermark to mark them as such.

Which API is easiest to integrate?

EchoMouth uses an OpenAI-compatible endpoint with the same request shape as OpenAI. You can use official OpenAI SDKs by changing the base_url. This makes integration straightforward for developers already familiar with standard TTS APIs. We provide a ready curl example and clear documentation for quick setup.

Do I get a free trial for each API?

This question asks about other APIs. EchoMouth offers a $0.50 free trial (50,000 characters) valid for 7 days when signing in with Google. Email accounts do not receive trial credit. This allows you to test the API quality and integration before committing to USDT top-ups for production.

Which API has the best voice quality?

Voice quality is subjective. EchoMouth uses licensed LibriTTS-R and LibriVox voices. We do not claim to be the best. Our voices are clear and deterministic. If you prefer different quality, you can test our eight presets. We provide samples so you can decide what fits your game's aesthetic.

Can I use these APIs for audiobooks?

EchoMouth supports long texts by splitting them into 4096-character parts and joining the audio. This works for audiobooks. However, we do not offer SSML or custom pronunciation. If your audiobook needs precise control, you might prefer a different API. Our pricing is transparent per character generated.

Is there a limit on characters per minute?

EchoMouth has a rate limit of 60 requests per minute per key. Each request can handle up to 4096 characters. Before the first top-up, the limit is 10 requests per minute. You can split longer texts into multiple requests to stay within these limits while generating audio.

Do I need to credit the voices?

EchoMouth voices are based on LibriTTS-R (CC BY 4.0) and LibriVox public-domain readings. While we do not mandate specific credit, attribution is good practice. Each generated file carries an inaudible watermark marking it as synthetic. You can check the specific license terms for LibriVox if required by your project.

Which API is best for indie developers?

EchoMouth is optimized for indie developers. We offer transparent per-character pricing, no subscriptions, and USDT payments. Our API is deterministic and easy to integrate. We support common game languages and provide a free trial. This model keeps asset production costs predictable without the overhead of enterprise contracts or SLAs.

Can I switch APIs easily?

EchoMouth uses an OpenAI-compatible endpoint with the same request shape as OpenAI. If you are already using OpenAI SDKs, you only need to change the base_url. This makes switching from other compatible APIs easy. We do not lock you into a proprietary format, ensuring flexibility for your workflow.

Do these APIs support real-time streaming?

This question refers to other APIs. EchoMouth does not support real-time streaming or WebSocket. You send a text request and receive the full audio file. This is suitable for pre-generated NPC dialogue. If you need low-latency streaming for interactive conversations, you may need a different API solution.

Which API has the best documentation?

EchoMouth provides clear documentation at /docs/. We offer a ready curl example and explain the request fields. We do not claim to have the best documentation, but we ensure it is specific and factual. You can find details on endpoints, voices, and pricing directly on our site.

Evaluation Criteria

Sound wave illustration for: Evaluation Criteria

Evaluating a TTS API for NPC dialogue requires balancing deterministic output with production overhead. Since game assets are static, the primary metric is cost per character, ensuring that thousands of lines of dialogue remain within budget without hidden fees. Voice variety must support distinct character archetypes without requiring custom cloning.

Integration complexity dictates development velocity. An OpenAI-compatible endpoint with standard OpenAI SDKs allows immediate implementation into existing game engines. Minimal boilerplate code reduces the time spent on infrastructure management, letting developers focus on narrative design and asset creation.

Cost Per Character

Transparent per-character pricing eliminates surprise invoices. With a flat rate of $0.01 per 1,000 characters, you can precisely calculate the cost of entire dialogue trees. Failed requests incur no charge, protecting your budget during testing phases.

This model scales linearly with content volume. Whether you have 5,000 characters for a short quest or 500,000 for a full RPG, the cost remains predictable and directly proportional to the text length.

  • $0.01 per 1,000 characters
  • Failed requests are free
  • No monthly base fees

Voice Quality and Variety

Eight distinct presets cover common NPC roles: from the calm instructional voice of Sage to the energetic promo style of Felix. Each voice is based on licensed recordings, ensuring legal clarity for commercial use.

Consistency is key for character identity. The same voice preset produces deterministic audio for identical inputs, allowing you to reuse audio files across different scenes without re-synthesis.

  • 8 distinct voice presets
  • Licensed audio sources
  • Deterministic output for identical inputs

Integration Complexity

The API mimics the OpenAI speech endpoint structure. You can use the official OpenAI SDKs by simply changing the base_url to https://api.ao999.info/v1. This reduces integration time significantly.

Standard JSON requests with fields like model, input, and voice are all that is needed. No custom wrappers or proprietary libraries are required, making it easy to integrate into Unity, Unreal, or Godot.

  • OpenAI SDK compatible
  • Standard JSON format
  • No custom libraries needed

Our API Features

Sound wave illustration for: Our API Features

EchoMouth provides a deterministic TTS API optimized for game developers needing consistent NPC dialogue. We offer eight preset voices covering diverse tones, ensuring your audio assets remain coherent across thousands of lines without the variability of custom cloning services.

Payments are handled exclusively via USDT on the TRON network, keeping production costs predictable and transparent. The API endpoint is compatible with the official OpenAI SDKs by changing the base_url, allowing you to integrate TTS into your existing pipelines with minimal code changes.

Preset Voice Library

Our eight voice presets range from warm narrators to energetic promo styles, all supporting 23 languages. Each voice is based on licensed recordings, providing clear, synthetic speech suitable for NPC interactions. Every generated file includes an inaudible watermark marking it as synthetic.

  • Eight distinct voice presets (e.g., Aria, Nova, Felix, Atlas)
  • Support for 23 languages, including English, Japanese, and German
  • Inaudible synthetic watermark in every audio file

USDT Payment Model

We charge $10 per 1 million characters, with failed requests refunded in full. Top-ups via USDT (TRC-20) are credited automatically within minutes. There are no monthly plans, seats, or hidden fees, making it ideal for indie budgets.

  • $0.01 per 1,000 characters
  • Minimum $5 top-up via USDT on TRON
  • No monthly subscriptions or credit expiration

OpenAI SDK Compatibility

The endpoint POST https://api.ao999.info/v1/audio/speech mirrors the OpenAI speech structure. You can use the official OpenAI SDKs by simply changing the base_url. This allows for quick integration into existing Python or Node.js game projects without custom parsers.

  • OpenAI SDK compatible with base_url override
  • Standard JSON request body
  • Supports tts-1 and tts-1-hd models

Alternative Cloud Providers

Sound wave illustration for: Alternative Cloud Providers

When evaluating TTS for game NPC dialogue, established cloud providers offer robust infrastructure but often at a higher complexity and cost per character. Their enterprise-grade features are powerful but may exceed the needs of indie developers seeking predictable, per-character billing for asset production.

Below is a factual comparison of the three major cloud TTS APIs against our specific requirements for deterministic, character-limited synthesis.

Azure Cognitive Services

Azure offers a wide range of neural voices with strong language support. It provides SSML capabilities and streaming endpoints, which are useful for dynamic dialogue but add integration complexity compared to our simple REST API. Pricing is based on character tiers and may include hidden costs for premium neural voices or long-text processing.

Unlike our model, Azure requires a monthly Azure subscription and potentially higher minimum spend commitments. Voice cloning is available but at a premium, and the API does not support USDT payments, creating friction for freelance audio engineers managing crypto-friendly budgets.

  • Supports SSML for fine-grained control
  • Requires Azure subscription management
  • Higher base complexity for indie devs

Amazon Polly

Amazon Polly is a mature TTS service with a strong focus on text-to-speech conversion and voice management. It offers a variety of neural voices and supports SSML for pronunciation control. However, its pricing structure is complex, with different rates for standard and neural voices, and it lacks a simple per-character metering that aligns with game asset production workflows.

Polly does not natively support USDT payments, requiring credit card or AWS billing integration. While reliable, the API requires more setup than our OpenAI-compatible endpoint, and it does not offer the same transparent, per-character pricing model that simplifies budgeting for indie game developers.

  • Mature neural voice technology
  • Complex pricing tiers
  • No USDT payment option

Google Cloud TTS

Google Cloud TTS provides high-quality neural voices and supports SSML. It is a robust option for large-scale applications but often involves higher costs and a more complex integration process compared to our streamlined API. The service requires Google Cloud Platform accounts and billing setups, which can be overkill for small indie teams.

Like other major cloud providers, Google Cloud TTS does not support USDT payments and relies on traditional credit card billing. The pricing is not strictly per-character in a simple flat rate, making it harder to predict costs for high-volume NPC dialogue generation in games.

  • High-quality neural voices
  • Requires Google Cloud account
  • Traditional billing only

Specialized TTS Services

Sound wave illustration for: Specialized TTS Services

Specialized TTS services often prioritize emotional nuance and voice cloning over the deterministic, character-limited synthesis that indie game developers require for NPC dialogue. While they offer polished results, their pricing models and feature sets may not align with the strict budget constraints of small-scale game production.

ElevenLabs

ElevenLabs is known for high-quality, emotive voice synthesis and excels in realism. However, it lacks deterministic character limits per request, which can complicate asset budgeting for developers who need precise cost-per-dialogue-line calculations.

  • High emotional range
  • Voice cloning available
  • Higher per-character cost

PlayHT

PlayHT offers a wide range of voices with good clarity but focuses more on content creation workflows than game asset integration. Its pricing structure is less transparent for high-volume, short-duration NPC lines compared to per-character models.

  • Large voice library
  • Podcast-focused features
  • Complex pricing tiers

Murf.ai

Murf.ai is designed primarily for studio-style voiceovers and presentations. It lacks the API-first simplicity needed for automated game dialogue pipelines, often requiring manual exports rather than direct programmatic access for batch NPC generation.

  • Studio interface
  • Video sync tools
  • Less API-driven

Feature Comparison

Sound wave illustration for: Feature Comparison

Understanding feature parity helps indie developers choose the right tool for NPC dialogue pipelines. We compare our API against major cloud providers and specialized services based on concrete technical capabilities relevant to game audio production.

SSML Support

Our API does not support SSML. You control prosody via the speed parameter (0.25–4.0) and voice selection. This simplifies integration but lacks fine-grained phoneme control.

Voice Cloning Availability

We do not offer custom voice cloning. You choose from eight licensed presets (e.g., aria, atlas). (Fact check: Correct.)

Streaming Capability

We do not provide streaming or WebSocket endpoints. Audio is returned as a complete file after synthesis. For NPC dialogue, this fits batch processing; for real-time interaction, you must handle client-side buffering.

Pricing Analysis

Sound wave illustration for: Pricing Analysis

Pricing is strictly per-character, not per-second or per-minute, making costs predictable regardless of audio length. At $0.01 per 1,000 characters, long NPC dialogue lines cost pennies. Failed requests are free, so you only pay for successful synthesis.

Subscription vs. Pay-Per-Use

We offer no monthly plans or seats. You top up once and spend down. This model suits indie devs who generate audio in bursts during development sprints rather than steady daily production.

  • $10 per 1 million characters
  • No monthly subscription
  • No minimum spend

Hidden Fees Comparison

There are no hidden fees. The character count includes spaces and punctuation. You know the exact cost before sending the request. No VAT invoices or currency conversion fees apply.

Free Tier Availability

Google sign-ups receive $0.50 credit (approx. 50,000 characters) valid for 7 days. Email/password accounts start with zero balance. This allows testing integration without payment commitment.

Technical Implementation

Sound wave illustration for: Technical Implementation

Integrate EchoMouth using the standard OpenAI Python or Node.js SDK by pointing base_url to https://api.ao999.info/v1. This allows you to reuse existing code structures for NPC dialogue generation without learning a new client library. Simply swap the API key and base URL.

SDK Availability

Use the official OpenAI SDKs (Python, Node.js, etc.) by setting base_url to https://api.ao999.info/v1. This ensures compatibility with existing wrappers and simplifies integration for developers already familiar with the OpenAI speech endpoint structure.

API Rate Limits

Each API key allows 60 requests per minute. New accounts start with a 10 requests/minute limit until the first top-up is processed. Exceeding this limit returns a 429 error, so implement exponential backoff for batch NPC line generation.

Error Handling

Failed synthesis requests are fully refunded, so you can retry without worrying about credit loss. Handle HTTP 429 (rate limit) and 400/401 errors gracefully. For long dialogues, split text into chunks under 4096 characters and concatenate the resulting audio files.

Technical Implementation

Use Case: NPC Dialogue

Sound wave illustration for: Use Case: NPC Dialogue

NPC dialogue requires deterministic synthesis where character voices remain consistent across thousands of unique lines. Our API ensures that the same voice preset produces uniform timbre regardless of input variation, critical for maintaining immersion in open-world games.

Voice Consistency Needs

Each of the 8 presets delivers stable acoustic traits, allowing developers to assign specific voices to character archetypes. The deterministic nature of the engine means repeated calls with identical text and parameters yield consistent results, reducing the need for extensive post-production cleanup.

  • 8 distinct voice presets for NPC archetypes
  • Consistent timbre across varying input lengths
  • No random variance in generated audio

Batch Processing Requirements

Game assets often require generating hundreds of lines simultaneously. The API accepts standard JSON bodies, allowing you to script batch requests via the OpenAI SDK. Split texts exceeding 4096 characters into manageable chunks to ensure reliable synthesis without token limits.

  • SDK-compatible batch requests
  • 4096 character limit per request
  • Automated splitting for long dialogues

Latency Tolerance

Pre-baked NPC lines do not require real-time generation, making our queue-based processing ideal. Requests are processed sequentially, which may introduce delays under load, but this is negligible for asset production pipelines where audio is generated offline before build packaging.

  • Optimized for offline asset generation
  • Sequential processing queue
  • Real-time speed approximation for long texts

Use Case: Audiobooks

Sound wave illustration for: Use Case: Audiobooks

EchoMouth supports long-form narration by splitting input text into 4096-character segments, synthesizing each part individually, and joining the audio files. This approach ensures deterministic output without exceeding API limits, making it suitable for full-length books.

Long-Text Handling

Texts are processed sequentially, so generation speed aligns with real-time playback (approx. 25 characters per second). For lengthy manuscripts, batch processing allows you to pre-generate all chapters asynchronously, keeping your pipeline efficient without blocking.

Voice Variety

You can assign different voices to different characters or chapters using the 8 preset options. Each voice supports all 23 languages, allowing consistent multilingual narration or distinct character voices within a single project without additional cloning costs.

Cost Efficiency

At $0.01 per 1,000 characters, audiobook production remains predictable. Failed requests are free, and your balance never expires. With a flat rate and no hidden fees, you can accurately forecast costs for projects ranging from short stories to full novels.

Use Case: Mobile Games

Sound wave illustration for: Use Case: Mobile Games

Mobile games demand efficient asset pipelines where network conditions vary. Our API supports WAV, MP3, Opus, FLAC, and PCM formats, allowing you to balance file size and quality. For dialogue-heavy titles, Opus offers significant bandwidth savings without perceptible loss in voice clarity.

Offline playback is a primary use case for NPC interactions. By downloading voice lines during menu screens or loading states, you eliminate runtime latency. The deterministic character limit ensures consistent audio lengths, simplifying animation synchronization across devices.

Bandwidth Considerations

Selecting the right output format directly impacts download sizes. Opus and MP3 provide smaller footprints compared to WAV, preserving mobile data for other game assets. FLAC remains an option for lossless needs, though it increases storage requirements.

  • Opus for optimal compression
  • MP3 for broad compatibility
  • WAV for uncompressed quality
  • FLAC for lossless archival

Offline Playback

Pre-generating dialogue assets during idle periods ensures instant playback. This approach removes network dependency for critical game events. Audio files are generated once and cached locally, guaranteeing consistent performance regardless of connectivity.

Device Compatibility

Standard audio formats ensure compatibility across iOS and Android devices. No proprietary codecs or SDKs are required for playback. The API returns standard audio streams that integrate seamlessly with existing game engines.

Final Recommendations

Sound wave illustration for: Final Recommendations

EchoMouth’s deterministic per-character pricing and USDT top-ups eliminate billing surprises for tight indie budgets. With no monthly commitments or hidden fees, you only pay for the audio you generate.

For projects demanding specific acoustic textures, the eight distinct voice presets cover narrative, news, and conversational tones without requiring external mixing tools.

Integration is immediate: use the OpenAI SDK, adjust the base URL, and start generating NPC lines in minutes.

For Budget-Conscious Devs

Use the free $0.50 trial to benchmark character costs against your dialogue tree size. USDT payments on TRC-20 ensure low transaction fees, keeping your overhead predictable. Since failed requests are free, you can experiment with voice parameters without financial penalty.

  • $0.01 per 1,000 characters
  • No monthly subscriptions
  • USDT (TRC-20) top-ups only

For High-Quality Needs

Select tts-1-hd for broadcast-ready clarity or tts-1 for standard dialogue. The eight presets (e.g., Aria for narrators, Felix for promos) provide immediate variety. Each file includes an inaudible synthetic watermark, ensuring clear attribution of AI-generated assets.

  • tts-1 and tts-1-hd models
  • 8 distinct voice presets
  • Inaudible synthetic watermark

For Quick Integration

The API mirrors the OpenAI speech endpoint structure. Change the base URL to https://api.ao999.info/v1 and use standard keys. Support is email-only, so keep your integration logic simple and self-documenting to minimize support queries.

  • OpenAI SDK compatible
  • Standard JSON body
  • Email support only

Limitations of Current Options

Sound wave illustration for: Limitations of Current Options

While EchoMouth prioritizes portability through standard formats and OpenAI SDK compatibility, other providers often lock developers into proprietary ecosystems. This section analyzes the specific limitations of current market options that indie developers and audio engineers must consider when selecting a TTS solution for game assets.

Voice Cloning Costs

Many specialized services charge premium rates for custom voice creation, often requiring high minimum spend commitments. Unlike our static preset library, cloning introduces variable costs that can disrupt tight indie game budgets.

  • High upfront fees for custom model training
  • Ongoing costs for cloned voice maintenance

Streaming Limitations

Most competitors lack native WebSocket or real-time partial audio streaming, forcing full request completion before playback. This increases latency for interactive NPC dialogue compared to our deterministic, character-limited synthesis approach.

  • No real-time partial audio support
  • Higher latency for interactive scenes

Platform Lock-in

Proprietary formats or SDKs can restrict portability. EchoMouth uses standard formats and OpenAI SDKs, so this is a general statement about competitors. Relying on vendor-specific integration can make switching engines difficult and costly later.

Future Considerations

Sound wave illustration for: Future Considerations

Roadmap items like model upgrades or new voices are not guaranteed. We do not promise specific release dates or deprecation timelines for the current API version.

Model Updates

We may deploy engine improvements to 'tts-1' or 'tts-1-hd' without changing the endpoint contract. Changes are backward-compatible, but specific quality improvements are not guaranteed.

New Voice Releases

The eight preset voices (Aria, Nova, Luna, Sage, Felix, Milo, Atlas, Orion) are stable. We do not guarantee the addition of new voices or the removal of existing ones.

API Deprecation

The /v1/ endpoint remains the standard. We may deprecate older versions with notice, but no SLA guarantees deprecation windows. Check /v1/models for current status.

  • Backward compatibility is maintained for /v1/ endpoints

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. Unlike our model, Azure requires a monthly Azure subscription and potentially higher minimum spend commitments.

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. LibriTTS-R paper (arXiv 2305.18802)
  3. LibriVox public-domain audiobooks
  4. Python documentation
  5. TRON developer hub: TRC-20
  6. W3C WAI: making audio and video accessible
  7. openai-python on GitHub
  8. Tether (USDT)
  9. Creative Commons Attribution 4.0
  10. Node.js documentation
  11. openai-node on GitHub
  12. IETF RFC 7845: Ogg Opus

More on this site