Frequently asked questions
What is the best voice for historical tours?
Atlas offers a neutral, news-style tone ideal for factual historical tours. Orion provides a deep, documentary voice suitable for immersive heritage experiences. You can choose any of the eight presets based on your content’s tone, as all voices support all 23 languages with equal clarity.
How do I ensure accurate pronunciation?
Set the language parameter explicitly using its ISO code. If omitted, Latin script defaults to English, which may mispronounce non-English terms. The API guesses language from script, so specifying it ensures correct articulation for your specific historical or tour context.
Can I mix languages in one audio file?
No. Each API request generates a single audio file for one language. To create a mixed-language audioguide, you must send separate requests for each language segment and join the resulting audio files yourself using standard audio editing tools.
What is the character limit for tours?
Each API request accepts up to 4096 characters. For longer tour scripts, split the text into parts of this size. You can then generate each part individually and join the audio files to create a complete, continuous audioguide experience.
How long does it take to generate audio?
Generation is roughly real-time, averaging 25 characters per second. Short phrases take a few seconds, while longer texts take longer. Requests are processed sequentially, so under load, you may experience slight delays in a queue before synthesis begins.
Do I need to pay for each language?
No. You pay per character generated, regardless of language. All 23 supported languages, including Arabic, Chinese, and Swahili, use the same voice engine and pricing structure. There are no extra fees for switching languages between different audio segments.
Is the audio quality consistent across languages?
Yes. The same eight voice presets are available in all 23 languages. Each voice maintains its specific tonal characteristics, such as Luna’s soft storytelling or Felix’s energetic promo style, ensuring consistent brand audio across your multilingual museum collections.
Can I update the audio after publishing?
Yes. Since audio is generated on demand via API, you can update your script and regenerate the audio file. The new file will replace the old one in your system. Generated audio is not stored by the API, so you manage the files.
What is SSML and do I need it?
SSML is a markup language for controlling speech synthesis. Our API does not support SSML, phoneme dictionaries, or custom pronunciation lexicons. You must rely on standard text formatting and explicit language codes to control pacing and pronunciation within the 4096-character limit.
How do I handle special characters?
Special characters count toward the 4096-character limit per request. Spaces and punctuation are treated as characters. Ensure your input text fits within this limit, splitting longer scripts as needed. The API processes standard Unicode text without additional formatting requirements.
Is the API suitable for large museums?
Yes. The API supports 60 requests per minute per key. For higher volumes, you can issue a new key, which immediately invalidates the old one. This allows you to manage multiple concurrent audioguide generation pipelines without relying on an SLA.
Can I use the audio on radio?
Yes. The API generates synthetic speech in formats like MP3, WAV, or FLAC. Each file includes an inaudible watermark marking it as synthetic. While the watermark is designed to be unnoticeable, you should verify it meets your specific station's technical standards before distributing the audio files.
What formats are best for QR codes?
MP3 is recommended for QR code links due to its small file size and universal compatibility. WAV and FLAC offer higher quality but larger files. Opus and PCM are also available. Choose MP3 for efficient loading when users scan codes in the field.
How do I track usage?
Log in to your account dashboard to view your balance, characters used, and usage history. The API keeps request metadata for billing purposes. You can monitor your character consumption directly on the /get-api-key/ page to manage your budget and top-ups effectively.
Why Multilingual Audio Matters
Multilingual audio transforms passive exhibits into engaging experiences for global visitors. By offering clear narration in multiple languages, museums can effectively reach international tourists who prefer listening over reading dense placards. This approach also improves accessibility for visitors with visual impairments or reading difficulties, ensuring inclusive engagement with cultural heritage.
Attracting international tourists
Tourists often struggle with language barriers in unfamiliar settings. Providing high-quality audio guides in their native language removes this friction, allowing them to fully appreciate the context and history of exhibits without needing physical guidebooks or expensive human guides.
- Support for 23 languages covers major tourist demographics.
- Consistent voice quality across all languages builds brand trust.
- No need for multiple physical guidebook prints.
Improving accessibility
Audio content serves visitors who find reading small text challenging. Synthetic speech offers a reliable, on-demand alternative to human readers, ensuring that accessibility needs are met efficiently without scheduling constraints.
- Clear, natural-sounding voices aid comprehension.
- Adjustable speed allows users to control pacing.
- Invisible watermarks mark content as synthetic speech.
Reducing translation costs
Traditional production involves hiring voice actors, studios, and editors for each language, which is time-consuming and expensive. Our API allows you to generate audio in 23 languages rapidly. While requests are processed sequentially and may queue under load, the cost is simply $0.01 per 1,000 characters, making it significantly cheaper than traditional methods.
Choosing the Right Voices
Selecting the appropriate voice preset ensures your museum content resonates with visitors. MuseVoice offers eight distinct presets derived from licensed recordings, allowing you to match the tone of your exhibits precisely. Whether you need a warm narrator for historical artifacts or a neutral tone for daily announcements, the right voice enhances the visitor experience without requiring external hiring.
Matching tone to exhibit type
Use warm, clear voices like Aria or Luna for storytelling exhibits to create an immersive atmosphere. For instructional areas or guided tours, Sage provides a calm, instructional tone that aids understanding. These presets are based on licensed recordings, ensuring high quality and consistency across all your audio guides.
- Aria: Warm, clear narrator for general exhibits
- Luna: Soft, storytelling voice for historical narratives
- Sage: Calm, instructional tone for guided tours
Neutral voices for news
Atlas delivers a neutral, news-style delivery ideal for daily updates, weather alerts, or factual exhibit descriptions. Its straightforward tone ensures information is conveyed clearly without emotional bias, making it perfect for time-sensitive content or informational plaques where clarity is paramount.
- Atlas: Neutral, news-style delivery for factual content
Warm voices for storytelling
Nova offers a bright, upbeat tone suitable for engaging, energetic content such as children’s exhibits or promotional audio. Felix provides an energetic, promo-style voice for short, dynamic segments. Both voices add personality to your audio guides, making the content more memorable for visitors.
- Nova: Bright, upbeat tone for engaging content
- Felix: Energetic, promo-style voice for dynamic segments

Setting Up the API Request
Configuring the request body correctly ensures your audioguide sounds natural and reaches the right audience. The API accepts text up to 4096 characters per call, requiring you to split longer scripts into manageable segments before joining the audio files.
Specifying language codes
Always include the language field (e.g., "es", "fr", "ja") in your JSON body. While the API guesses language from Latin script (defaulting to English), explicit codes prevent errors for non-English languages or mixed scripts. Supported languages include Arabic, Chinese, German, Korean, and Swahili, among others.
- Use ISO codes for precise control.
- Default to English for unlabeled Latin text.
- Explicitly set codes for non-Latin scripts.
Handling script detection
The system infers language from character scripts when the language field is omitted. If your script contains mixed languages or ambiguous characters, explicitly defining the language prevents mispronunciation. For example, Dutch text without a language tag might be misread as English.
- Omit for simple Latin-only English.
- Set for mixed or non-Latin scripts.
- Prevents default English misinterpretation.
Ensuring correct pronunciation
Pronunciation relies on standard text spelling since custom pronunciation dictionaries and SSML are not supported. Ensure proper capitalization, punctuation, and spacing to guide the TTS engine. For complex names or acronyms, spell them phonetically (e.g., 'N-A-T-O') to guarantee accurate audio output.
- Spell out acronyms phonetically.
- Use clear punctuation for pacing.
- Avoid complex symbols without context.
Managing Audio Length
Long exhibit descriptions often exceed the 4,000-character API limit. You must split your script into smaller chunks and generate audio files for each part independently.
The API processes requests sequentially, so under load, generation may queue. Ensure your integration handles this timing to avoid bottlenecks when creating large audioguide sets.
Splitting long scripts
Divide your text into segments of up to 4,096 characters. Each request generates one audio file. This ensures every chunk is processed within the character limit, regardless of the total script length.
Keep each segment logically complete to avoid awkward pauses. Break at sentence or paragraph boundaries to maintain natural phrasing for the synthetic voice.
- Limit input to 4,096 characters per request.
- Break text at natural pause points.
- Treat spaces and punctuation as characters.
Joining audio segments
Receive each audio file separately via the API response. Your application must concatenate these files into a single track. The API does not join them automatically.
Ensure your code handles the different response formats (mp3, wav, opus, flac, pcm) consistently. Proper concatenation requires compatible formats and sample rates.
- Concatenate audio client-side.
- Use consistent response formats.
- Handle file metadata correctly.
Maintaining flow
Since requests are processed one after another, long scripts may take time. Plan your generation workflow to avoid delays for users waiting for the full audioguide.
Test the joined audio for continuity. Synthetic speech may have slight timing variations between segments. Adjust speed settings (0.25–4.0) to ensure a consistent listening experience across all parts.
- Process requests sequentially.
- Test joined audio for gaps.
- Adjust speed for consistency.
Quality Control
Quality control ensures your audioguides meet professional standards before deployment. While the API handles synthesis, you must verify audio fidelity, timing, and user experience to maintain engagement in museum and tourism contexts.
Listening for errors
Review generated files for unnatural pauses or mispronounced proper nouns. Each file carries an inaudible watermark marking it as synthetic, ensuring compliance without auditory distraction. Check that the audio flows naturally within your specific exhibit context.
Adjusting speed
The speed parameter accepts values from 0.25 to 4.0. For complex historical explanations, slow down the voice (e.g., 0.8) to aid comprehension. For quick transit updates, increase speed (e.g., 1.5) to keep visitors moving. Test multiple speeds to find the optimal balance between clarity and pacing.
Testing with users
Play the audio in the actual museum environment to assess background noise interference. Verify that the voice tone matches the exhibit’s mood—warm for storytelling, neutral for news. Gather feedback from a small group of visitors to identify any clarity issues before full rollout.
Distribution Methods
Once you have generated the audio files, the next step is making them accessible to visitors within your museum or tourism environment. The API returns audio directly in the response, allowing you to store these files locally or on a CDN for fast, reliable playback during peak hours.
Web players for museums
Embed audio players directly into your museum’s website or digital guide. Since generated audio is returned in the response, you can cache these files to ensure fast loading times for tourists browsing on-site.
Mobile app integration
Integrate audio generation directly into your visitor app using the OpenAI SDK. This allows for on-demand audioguide creation, ensuring content remains fresh without pre-downloading large audio libraries.
QR code audio links
Generate static URLs for each audio track and print them as QR codes next to exhibits. This method requires no app installation and works with any smartphone camera, providing instant access to multilingual content.
Cost Management
Museums and tourism teams need precise budget control when generating multilingual audioguides. Our API charges $0.01 per 1,000 characters, meaning a 10,000-character script costs just $0.10. This predictable pricing helps product teams plan audio content for hundreds of exhibits without unexpected costs.
Estimating character counts
Every character, including spaces and punctuation, counts toward your usage. A standard museum guide of 500 words contains roughly 2,500 characters. With a free trial credit of $0.50, you can test about 50,000 characters across multiple languages before spending real funds.
- Count spaces and punctuation as characters
- Test scripts within the 40,960 character limit per request
- Use the $0.50 trial credit to verify costs
Optimizing script length
Keep scripts concise to reduce generation time and cost. Long texts should be split into segments under 4,096 characters. This ensures faster processing and avoids queue delays during peak hours. Each segment generates a separate audio file, which you can join for a continuous experience.
- Split long scripts at logical pause points
- Stay under 4,096 characters per request
- Join segments for seamless playback
Budgeting for updates
Content updates are cost-effective. Top-ups via USDT on the TRON network (TRC-20) are processed automatically within minutes. Since balance does not expire, you can accumulate credits over time. Failed requests are free, so you can refine scripts without financial penalty.
- Top up with USDT (TRC-20)
- Balance never expires
- Failed requests are refunded

Technical Integration
MuseVoice works with the standard OpenAI SDK by simply pointing it to our API base URL. This allows you to use existing code libraries without installing custom packages, ensuring smooth integration into your museum’s digital infrastructure.
Using the OpenAI SDK
Set the base URL to https://api.playbdkk.com/v1 and use your Bearer token. The request shape matches the OpenAI speech endpoint, so tts-1 and tts-1-hd models work exactly as documented.
Caching audio files
Since audio generation takes time, store the resulting files on your server. Request metadata is kept for billing, but the audio itself is returned in the response and not stored by us.
Handling retries
Requests are processed sequentially, so under load they may queue. If synthesis fails, the charge is returned in full. Implement retry logic for failed requests, but avoid redundant calls for successful ones.
Common Pitfalls
Museums often assume the API detects language automatically, but Latin-script text defaults to English. If your script is in French or German, omitting the language parameter results in an English pronunciation, confusing visitors. Always explicitly set the language code for non-English texts.
Another frequent issue is sending large texts in a single request. The API limits input to 4096 characters. Exceeding this causes errors or truncation. Split your audioguide scripts into smaller segments to ensure reliable generation and smooth audio joining.
Ignoring language codes
The API guesses language from script, defaulting to English for Latin text. For French, German, or Spanish scripts, you must set the language field. Without this, the voice speaks English, even if the text is foreign.
This leads to confusing audio guides where visitors hear English phonetics applied to their native language text. Explicitly defining the ISO code ensures accurate pronunciation and a professional experience.
- Set
languagefor all non-English Latin scripts. - Default detection works for English but fails for other languages.
- Prevents mixed-language pronunciation errors.
Overloading requests
Each request accepts up to 4096 characters. Sending longer scripts causes failures or unexpected behavior. Split your content into chunks that fit this limit. Use a script to concatenate the resulting audio files for a continuous audioguide experience.
Additionally, respect the rate limit of 60 requests per minute. Bursting too many requests can delay generation, especially during peak hours. Queue your requests to maintain steady processing.
- Max 4096 characters per request.
- Split long scripts into segments.
- Respect 60 requests per minute limit.
- Queue requests to avoid delays.
Missing format options
Choosing the wrong audio format can affect file size and playback compatibility. MP3 is small and web-friendly, while WAV offers high quality for local playback. Select response_format based on your distribution method.
Ensure your web player or mobile app supports the chosen format. Mismatches can lead to playback issues. Test your audio files across different devices to confirm compatibility before deploying to your museum’s systems.
- Use MP3 for web players.
- Use WAV for high-quality local playback.
- Test compatibility across devices.
- Match format to distribution method.
Best Practices
Standardizing your voice profile ensures a cohesive experience across all exhibits and languages. Use warm voices like Aria or Luna for historical storytelling, and neutral tones like Atlas for informational plaques. Consistency helps visitors recognize the brand's audio identity regardless of the language they select.
Consistent voice selection
Assign specific voices to content categories rather than switching randomly. This approach maintains a professional tone and prevents listener fatigue. Since any voice can speak any of the 23 supported languages, you can maintain the same character across English, French, or Japanese guides without losing tonal consistency.
Clear script formatting
Keep input text under 4,096 characters per request to ensure reliable synthesis. Remove special symbols that the TTS engine may misinterpret, as these can disrupt the flow of audioguides. Use simple punctuation to control pacing, allowing the API to generate natural pauses between sentences without requiring complex SSML tags.
Regular content updates
Re-generate audio files whenever exhibit information changes to avoid outdated guidance. Since generated audio is not stored by the API, you must cache the returned files locally or on your server. This allows you to serve stable URLs via QR codes while keeping the underlying text current.
Future Scaling
Scaling your audioguide library requires minimal friction. MuseVoice’s API supports 23 languages, allowing you to expand into new markets like Turkish or Swahili without hiring local voice actors or negotiating new contracts. Each voice can naturally speak any supported language, ensuring consistent quality across your entire collection.
Adding new languages
Since any voice can speak any of the 23 supported languages, you can instantly localize existing scripts. This eliminates the need for separate production pipelines per region, letting you deploy multilingual content simultaneously.
Automating workflows
Integrate the API into your content pipeline to generate audio automatically when new exhibit texts are approved. With 60 requests per minute, you can batch-process large exhibit updates efficiently. Use USDT (TRC-20) for seamless, automatic balance replenishment, ensuring uninterrupted generation during peak tourist seasons.
Integrating with CMS
Connect your existing content management system to the API for direct audio injection. Since the API accepts standard JSON bodies, you can script updates to push new audio files directly to your web players or mobile apps, keeping your digital guide in sync with physical exhibits.

Support and Resources
We provide comprehensive documentation to help museum product teams integrate MuseVoice’s multilingual audioguide API efficiently. Our guides cover endpoint specifications, voice selection, and character limit handling, ensuring your audio content meets heritage and travel experience standards.
Documentation links
Access the full API reference at https://playbdkk.com/docs/ to explore request parameters, response formats, and integration examples. The docs detail how to use the OpenAI SDK or plain HTTP to generate audio in 23 languages, including specific guidance for museum script lengths and voice presets like Aria for storytelling or Atlas for neutral news-style guides.
Contact support
Support is available via email only. We do not offer phone support, live chat, or SLA guarantees. For technical questions regarding integration or billing issues, send a message to our support team. Response times may vary based on volume, but we aim to assist product teams promptly.
Community examples
Explore how other museums and tourism boards use our API to create engaging, multilingual audio experiences. These examples highlight practical workflows for splitting long scripts, caching audio files, and managing costs, helping you optimize your audioguide production process.
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. Our audioguide transforms museum visits into immersive storytelling experiences for visitors.
23 languages
Any voice speaks any of these languages.
