Frequently asked questions
Is voice cloning the same as TTS?
No. TTS generates speech from text using fixed presets, while cloning creates a unique voice model from specific audio samples. We offer eight fixed presets like Aria and Felix, but do not provide custom voice cloning for customers. You cannot upload your own voice to create a clone.
Do I need a GPU to clone voices?
This applies to cloning, not our API. Our TTS API processes requests on our cloud servers, so you do not need a GPU. You simply send text via HTTP, and we return the audio file. Your local machine only handles the API request and audio playback.
Can I clone a celebrity's voice?
We do not offer voice cloning, so you cannot clone celebrities through our API. Our voices are based on licensed recordings from LibriTTS-R and LibriVox. We do not use real people's voices or imitate celebrities directly. You get synthetic speech from our eight preset voices.
What is the LibriTTS-R license?
LibriTTS-R is a dataset licensed under CC BY 4.0. Our voices are based on recordings from this dataset, which includes public-domain audiobook readings from LibriVox. This provides a clear legal basis for our synthetic voices, avoiding the copyright headaches of traditional voice acting for your game.
Are GitHub cloning projects free?
GitHub projects vary in cost. Our TTS API is not free; it costs $0.01 per 1,000 characters. You get a $0.50 free trial for 7 days. Failed requests are free, but successful generations incur charges. There are no monthly plans or subscriptions.
Do synthetic voices sound robotic?
Our voices are based on licensed recordings and sound natural, not robotic. They include presets like Aria (warm narrator) and Felix (energetic). Each generated file carries an inaudible watermark marking it as synthetic speech, but the audio quality is designed for clear NPC dialogue.
How long does cloning take?
Cloning is not offered, so there is no training time. Our TTS API generates audio in real-time. Long texts take roughly one second per 25 characters. Short phrases take a few seconds. Requests are processed sequentially, so high load may cause slight delays.
Can I use cloned voices in games?
We do not offer cloning, but you can use our eight preset voices in games. The audio is synthetic and carries an inaudible watermark. Input text is not used for training, and generated audio is returned in the response without being stored by us.
What is a voice preset?
A preset is a fixed voice model provided by the API. We offer eight presets: Aria, Nova, Luna, Sage, Felix, Milo, Atlas, and Orion. Each has a distinct tone, like calm instruction or upbeat energy. You select one via the voice parameter in your request.
Do I need to train a model?
No. Our API uses pre-trained models with eight fixed presets. You do not need to train anything. Just send text via the API endpoint, and receive the audio file. This simplifies integration for freelance developers who need immediate results without complex setup.
Is API speech copyrighted?
Our voices are based on LibriTTS-R (CC BY 4.0) and LibriVox public-domain readings. The generated audio is synthetic. We do not claim copyright over the output. Input text is not used for training. You can use the generated audio in your projects without legal headaches.
Can I modify cloned voices?
We do not offer cloning, so you cannot modify a cloned voice. However, you can adjust the speed of our presets between 0.25 and 4.0. You can also choose from different response formats like MP3, WAV, or FLAC. Emotion tags are not supported.
What is the difference between TTS and cloning?
TTS converts text to speech using fixed models like our presets. Cloning creates a unique voice from specific audio samples. We provide TTS with eight presets. We do not offer custom cloning, voice design, or uploading your own voice for cloning purposes.
Do I need consent for synthetic voices?
Our voices are based on licensed data (LibriTTS-R) and public-domain readings. You do not need additional consent for using the synthetic output in your game. The voices are not imitations of specific real people or celebrities. Each file has an inaudible synthetic watermark.
Can I use GitHub code for commercial games?
This question refers to external projects. Our API is commercial: $0.01 per 1,000 characters. You pay via USDT. There are no minimum spends. You can use the generated audio commercially. We do not store the audio, only request metadata for billing.
How do I integrate cloning into my game?
Since we do not offer cloning, you integrate our TTS API. Send a POST request to the speech endpoint with your text and voice preset. Receive the audio file. Use the OpenAI SDK or plain HTTP. No custom SDK or mobile SDK is needed.
What are the limitations of cloning?
We do not offer cloning, so these limitations apply to our TTS. You cannot clone voices, add emotion tags, or use SSML. Text is limited to 4096 characters per request. No streaming or real-time partial audio is available. Support is email only.
Is cloning more expensive than TTS?
Cloning is not offered. Our TTS costs $0.01 per 1,000 characters. There are no separate costs for different voices. Failed requests are free. You pay only for successful generations. There are no monthly plans or subscriptions, making it affordable for freelancers.
Do I need to store voice data?
No. We do not store your input text or generated audio. Text is used only to produce audio and not for training. Request metadata (time, characters, status) is kept for billing. You receive the audio in the response and handle storage yourself.
Can I clone voices in real-time?
We do not offer cloning. Our TTS is not real-time streaming; it returns the full audio file after generation. Typical speed is roughly real-time for long texts. Short phrases take a few seconds. Requests are processed sequentially, so load may cause delays.
Defining Voice Cloning
Voice cloning creates a unique synthetic voice by analyzing a specific person's speech patterns, requiring substantial training data and computational power to replicate their exact timbre and cadence. In contrast, our API presets are pre-trained models that deliver consistent, high-quality synthetic speech without the need for custom model training.
What Cloning Actually Is
Cloning involves training a new model on a targeted voice sample, resulting in a highly personalized audio output. This process typically demands significant technical expertise and hardware resources, making it less accessible for solo game developers compared to using established presets.
- Requires dedicated training time
- High computational overhead
Training vs. Presets
Our presets like Aria, Nova, or Atlas are already optimized for naturalness and clarity. You receive immediate, consistent results with zero setup, whereas cloning requires managing the training pipeline and validating the output quality before integration.
- Instant integration
- No training wait time
Data Requirements
Effective cloning usually needs hours of clean, diverse audio from the target voice. Our API uses licensed recordings from LibriTTS-R and LibriVox, ensuring legal clarity and immediate usability without you needing to gather and process large datasets.
GitHub Open Source Projects
Open-source cloning tools like Coqui TTS or XTTS are popular choices for developers seeking custom voices. These projects typically require downloading pre-trained models and managing dependencies, which can vary significantly between repositories.
While some projects offer cloud inference endpoints, most rely on local processing. This approach gives you full control over the voice model but demands your own hardware resources to run the synthesis.
Setting up these repositories often involves navigating complex dependency trees, configuring environment variables, and potentially compiling audio libraries. The learning curve is steeper than simply calling an API endpoint, requiring more initial time investment.
Popular Cloning Repos
Projects like XTTS and Coqui TTS dominate the open-source space for voice cloning. They allow you to generate a voice profile from an audio sample, offering high customization for specific NPC characters.
These repositories are actively maintained by communities but vary in documentation quality and ease of use. You often need to manually install dependencies like PyTorch or TensorFlow to get the inference running.
- XTTS: High-quality multi-language support
- Coqui TTS: Modular and flexible architecture
- VALL-E X: Neural codec language models
Local vs. Cloud Processing
Local processing keeps your audio files and voice data on your machine, which is ideal for privacy. However, it requires a capable GPU to generate audio in a reasonable time frame.
Cloud-based open-source options exist but often charge for inference time. They remove the hardware burden but introduce latency and dependency on third-party uptime, similar to our API but with less predictable pricing.
Complexity of Setup
Installing a cloning repo often means dealing with version conflicts between Python packages and CUDA drivers. You might spend hours just getting the environment ready before generating your first voice.
You also need to manage the voice profiles yourself. Unlike our preset system, you are responsible for ensuring the cloned voice remains consistent across different dialogue lines and game states.
Debugging audio artifacts or format issues requires a deeper understanding of the underlying model architecture compared to using our straightforward API.
Our TTS API Approach
Our TTS API relies on eight pre-trained voice presets rather than complex cloning workflows. Each voice, such as the warm narration of Aria or the energetic promo style of Felix, is derived from licensed recordings. This approach eliminates the need for you to gather and process specific audio samples to create a unique voice profile.
We prioritize consistency and speed, ensuring that every NPC line sounds like the same character, even when generated hours apart. By using a standardized engine, you avoid the variability often found in locally cloned models, keeping your game’s audio production streamlined and predictable.
Pre-Trained Presets
We offer eight distinct voice presets designed for specific roles. You can choose from female voices like Aria for clear narration or Nova for upbeat delivery, to male voices like Atlas for neutral news or Orion for deep documentary styles. These voices are built on the LibriTTS-R dataset, ensuring high-quality base audio.
Each preset is optimized for natural speech patterns without requiring additional training data from you. This means you can immediately assign a voice to a character and start generating dialogue without spending time on voice design or sample collection.
- Eight distinct voice presets available
- Based on licensed LibriTTS-R recordings
- No custom training data needed
Consistency Across Requests
Because we use a single engine for all generations, the voice remains identical across different requests. This is crucial for game development, where a character must sound consistent whether the line is generated during testing or in production. There is no drift in tone or timbre between sessions.
This consistency extends to pronunciation and pacing. Since the model is static, you do not need to retrain a cloned voice every time you add new dialogue, ensuring that your NPC’s voice remains reliable as your game grows.
Speed of Generation
Our API processes requests sequentially, typically generating audio at a speed close to real-time for long texts. Short phrases may take a few seconds, but the process is straightforward. You simply send a POST request and receive the audio file without waiting for a complex training phase.
This speed allows for rapid iteration during development. You can generate dozens of dialogue variations quickly to test how a voice fits a scene, without the computational overhead of running a local cloning model on your own hardware.
Comparison: Cloning vs. Presets
Choosing between cloning and presets often comes down to balancing unique character needs against project scale. Cloning offers distinct voices but requires more effort, while presets provide immediate consistency.
This section compares cost, legal clarity, and integration speed to help you decide which path fits your game's production pipeline.
Cost Efficiency
Presets cost $10 per million characters with no upfront GPU or infrastructure fees. Cloning tools often require renting GPUs or paying for training data preparation, adding hidden expenses.
For large dialogue trees, presets remain cheaper per line since you avoid the one-time training cost of cloning models.
- No hardware rental needed for presets
- Predictable pay-per-character billing
- Cloning adds GPU and storage overhead
Legal Simplicity
Our presets are built on LibriTTS-R (CC BY 4.0) and public domain sources, making commercial use straightforward. Cloning often involves navigating complex rights for the source voice.
You get clear, consistent licensing without worrying about the specific source material of a cloned voice.
- CC BY 4.0 and public domain sources
- Clear commercial rights
- No source-voice rights ambiguity
Ease of Integration
Presets integrate instantly via a simple API call with 8 established voices. Cloning requires uploading audio, waiting for training, and managing versioned models.
For freelancers, the speed of generating audio directly from text without a training phase saves significant development time.
- Immediate generation via API
- No training wait time
- Simpler version management
Consent and Rights
Your NPC dialogue inherits the clean legal foundation of our voice presets, avoiding the consent ambiguities that plague custom cloning solutions. We handle the licensing so you can focus on game development without worrying about voice rights disputes for your generated audio assets.
LibriTTS-R License
Our voices are derived from the LibriTTS-R dataset, released under the CC BY 4.0 license. This allows commercial use with proper attribution, ensuring your game’s audio assets are safe for distribution.
- CC BY 4.0 compatible
- Clear commercial rights
- No hidden restrictions
Public Domain Sources
The underlying recordings originate from LibriVox, where volunteer readers have donated their performances to the public domain. This means the source material is free from copyright claims, providing a stable legal base for your NPC voices.
Commercial Use
You can integrate these voices into any game, commercial or indie, without additional licensing fees. The synthetic nature of the output means you do not need to negotiate individual performer rights for each generated line of dialogue.
When to Clone vs. Use Presets
When you have dozens of NPCs with distinct personalities, presets like Aria or Atlas offer consistency but lack deep individuality. Cloning captures specific quirks, though it requires careful data selection to avoid capturing unwanted artifacts.
For massive volume, presets are faster and cheaper. Cloning adds setup time per voice, making it better for unique, high-value characters rather than background noise or generic crowd dialogue.
Unique Character Needs
Presets provide reliable, consistent tones ideal for generic roles or background characters. Cloning excels when a character’s voice is central to the narrative, allowing for nuanced emotional range that presets might miss.
Use presets for standard NPCs to save time. Reserve cloning for main protagonists or memorable side characters where vocal uniqueness drives player immersion.
- Presets: Best for generic or background roles
- Cloning: Ideal for main protagonists with specific vocal traits
Volume Considerations
Generating hundreds of lines with presets is immediate and cost-effective. Each cloned voice adds a one-time training overhead, which is negligible for small projects but scales up for large rosters.
If you need 100+ voices for a dense world, presets are more efficient. Cloning is worth the effort only if the voice’s uniqueness justifies the initial setup cost.
- Presets: High volume, low setup cost
- Cloning: Low volume, high uniqueness value
Technical Resources
Presets require no GPU power on your end, just API calls. Cloning models can run locally or in the cloud, requiring more computational resources and expertise to manage training data.
Freelancers with limited dev time should stick to presets. Teams with dedicated ML engineers can manage cloning pipelines for custom voice assets.
- Presets: No local GPU needed
- Cloning: Requires GPU or cloud compute resources
Technical Implementation
Integrating our TTS API is straightforward for freelance game developers. You send a POST request with your text and desired voice preset to https://api.playmm9.com/v1/audio/speech, receiving an audio file in return. This works with standard HTTP libraries or the OpenAI SDK by simply adjusting the base URL.
Unlike local models that require GPU setup and dependency management, our API handles all processing in the cloud. This means you can focus on game logic rather than maintaining inference servers, ensuring consistent voice quality across all your NPC dialogue without hardware constraints.
API vs. Local Model
Our cloud API abstracts away the complexity of model selection and hardware requirements. You define the voice via the voice parameter (e.g., 'aria' or 'atlas') and receive ready-to-use audio files. This eliminates the need to manage CUDA drivers, model weights, or inference engines within your game build or dedicated server.
- Single API endpoint for all voices
- No local GPU required for generation
- Standard JSON request body
Latency Differences
Generation speed depends on text length and queue status. Typically, processing takes roughly real-time (25 characters per second). Short phrases generate in a few seconds, while longer dialogues may queue if multiple requests are active. This latency is consistent regardless of your local machine's power.
Requests are processed sequentially per key, so high-volume procedural generation might experience slight delays. For immediate feedback, keep individual text chunks under 4096 characters to minimize processing overhead.
- ~25 characters per second processing rate
- Queued under load
- Instant response for short texts
Scalability
Scaling is managed through your API key limits. You can process up to 60 requests per minute once funded. For larger games, simply increase your USDT balance to maintain throughput without changing infrastructure. The API handles the heavy lifting, allowing you to serve thousands of NPCs with minimal code changes.
Since audio is returned directly in the response, you can play it directly into your game engine’s audio player. This reduces memory usage compared to storing large local voice banks, making it ideal for dynamic, on-the-fly dialogue generation.
- 60 requests per minute limit
- Automatic queue management
- Low memory footprint for game clients

Audio Quality
Audio quality in our TTS API is defined by consistency and clarity, rather than the unique but variable character of cloned voices. Our presets deliver reliable, high-fidelity speech suitable for NPC dialogue, ensuring every line sounds professional without the unpredictable artifacts common in local cloning models.
Cloning Artifacts
Open-source cloning tools often introduce subtle digital artifacts, such as metallic timbres or inconsistent breathing, especially when processing complex emotional ranges. These inconsistencies can break immersion in game dialogue, requiring additional post-processing to smooth out glitches that are rare in our pre-trained models.
Synthetic Watermarks
Every generated audio file includes an inaudible watermark identifying it as synthetic speech. This transparent marker helps distinguish our AI-generated NPC voices from human performances, adding a layer of clarity to your asset library without affecting the audible quality or player experience.
Naturalness
Our voices, derived from licensed LibriTTS-R recordings, prioritize natural intonation and pacing over hyper-realistic mimicry. This approach ensures that dialogue flows smoothly during gameplay, maintaining listener engagement without the uncanny valley effect that can plague less refined cloning implementations.
Cost Analysis
Voice cloning often requires expensive GPU infrastructure or complex cloud services, adding significant overhead for indie developers. In contrast, our API uses a simple pay-per-character model, eliminating the need for dedicated hardware or maintenance teams.
Cloning solutions frequently incur hidden costs for storage, compute time, and developer hours spent on tuning. Our transparent pricing ensures you only pay for what you generate, with no surprise charges for idle resources or complex licensing tiers.
GPU Costs for Cloning
Local cloning demands high-end GPUs, which can cost hundreds per month in cloud credits or upfront capital. Training a custom voice requires substantial processing power, often exceeding the cost of generating thousands of lines of dialogue through a standard API.
- High-end GPU rentals: $0.50–$2.00 per hour per node
- Training compute time: 2–10 hours per unique voice
- Storage overhead for model weights and datasets
API Pay-Per-Use
Our API charges $0.01 per 1,000 characters, making it incredibly affordable for large NPC dialogue trees. You only pay for successful generations, with failed requests refunded instantly, ensuring you never pay for wasted compute.
Starting with a $0.50 free trial allows you to test quality before committing. USDT top-ups are instant, with no monthly subscriptions or minimum spend requirements, fitting perfectly into tight indie budgets.
- Rate: $0.01 per 1,000 characters
- No monthly fees or minimum spend
- Failed requests are free
Hidden Expenses
Cloning solutions often hide costs in developer time, infrastructure management, and licensing complexities. Our straightforward pricing includes everything, with no hidden fees for API calls, data egress, or premium feature unlocks.
By avoiding custom voice training, you save on the engineering hours required to maintain your audio pipeline. This allows your team to focus on game design rather than audio infrastructure.
- Developer time for model tuning
- Infrastructure maintenance and updates
- Licensing fees for third-party models
Use Cases in Games
Voice synthesis integrates seamlessly into dialogue trees, allowing NPCs to deliver unique lines without recording sessions. By mapping specific voice presets to character types, you maintain consistency across thousands of interactions while keeping asset sizes minimal.
NPC Dialogue Trees
Generate dynamic responses for branching narratives by sending text snippets via the API. Each request returns audio instantly, enabling reactive conversations that adapt to player choices without pre-recording every possible combination.
Procedural Generation
Pair TTS with procedural text to create infinite unique dialogues. Since each request costs only $0.01 per 1,000 characters, you can afford to generate vast amounts of ambient chatter or random events without bloating your game's audio budget.
Localization
Support 23 languages natively without hiring voice actors for each region. The same text input can be rendered in Spanish, Japanese, or German using the appropriate preset, ensuring consistent character personality across global releases.
Future of Voice Tech
The TTS landscape is moving toward higher fidelity and lower latency, with models becoming more efficient at generating natural speech. While cloning improves, it remains a specialized tool rather than a replacement for standard presets in most game dev workflows.
Cloning technology is becoming more accessible, but it still requires specific audio data and processing power. Our presets offer a stable, consistent alternative that avoids the variability of trained models, ensuring your NPCs sound reliable every time.
Advancements in TTS
Models are getting faster and more expressive, reducing the 'robotic' feel. However, our current engine already delivers high-quality, natural-sounding speech suitable for most NPC interactions without needing complex training pipelines.
Efficiency gains mean lower costs and faster generation times. This allows you to generate large volumes of dialogue quickly, keeping your project timeline on track without waiting for lengthy training sessions.
- Improved naturalness in existing presets
- Faster generation speeds
- Reduced computational overhead
Cloning Trends
Cloning is becoming more common but remains resource-intensive. It requires specific audio samples and often local GPU power, which can be a barrier for solo developers or small teams with limited infrastructure.
While cloning offers unique voices, it introduces variability. Our presets provide a consistent audio identity for your game, ensuring that your NPCs sound exactly as intended across all platforms and languages.
Industry Standards
The industry is standardizing on API-driven solutions for scalability. Cloud-based TTS allows for easy integration into game engines, reducing the need for in-house audio infrastructure.
We adhere to clear licensing with LibriTTS-R, ensuring that your NPC dialogue is legally safe for commercial use. This transparency sets a standard for clarity in the TTS space, avoiding the legal gray areas of some cloning tools.
Summary
VoxQuest offers a straightforward, legal alternative to complex voice cloning workflows. By leveraging pre-trained presets with clear licensing, you avoid the hidden costs of GPU inference and the legal risks of unauthorized voice usage.
Our API delivers consistent, high-quality NPC dialogue with minimal integration effort, perfect for indie developers prioritizing speed and affordability over unique voice generation.
Key Takeaways
Cloning requires significant technical overhead and legal diligence, while our presets provide immediate, commercially safe audio. You get consistent quality without managing training data or inference servers.
- Pre-trained voices cover 23 languages natively.
- No copyright strikes for NPC dialogue.
- Instant integration via standard HTTP requests.
Best Practices
Select the voice that best matches your character’s archetype from our eight presets. Use the speed parameter to adjust pacing for different emotional tones without retraining.
Keep input text under 4096 characters to ensure optimal generation time and audio coherence. Test voice samples early in your development cycle.
Final Recommendation
For most freelance game developers, our TTS API offers a strong balance of cost, legal clarity, and ease of use. Reserve cloning for specific, high-value characters only if you have the technical resources to manage it.
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. This preset is excellent for tutorial systems, helping new users understand controls or rules with a friendly, approachable demeanor.
23 languages
Any voice speaks any of these languages.
