Dubbing Agencies: TTS API Voices, Pricing & Integration

By Dubbing Voice Metrics editorial team · Updated 2026-11-19

This guide helps localization agencies evaluate our text-to-speech API for global dubbing workflows. You will learn how to select from eight licensed voice presets across 23 languages, manage USDT billing for high-volume character usage, and integrate the REST API into your production pipeline.

We provide concrete metrics on latency, rate limits, and audio quality for post-production. The content covers technical integration details, legal licensing under CC BY 4.0, and practical strategies for maintaining voice consistency and handling long scripts, ensuring you can make informed decisions about synthetic voice adoption.

Dubbing Agencies: TTS API Voices, Pricing & Integration

Frequently asked questions

Which voice is best for news dubbing?

Atlas is a suitable voice for news dubbing. It provides a neutral, clear delivery suitable for professional broadcast standards. Its tone is neutral, making it suitable for factual content.

Which voice is best for storytelling?

Luna is a voice preset suitable for storytelling. Her soft, narrative tone creates an engaging atmosphere for fiction or audiobooks. Alternatively, Sage offers a calm, instructional style suitable for educational content. Both voices are based on licensed recordings, ensuring high-quality synthetic speech that maintains listener interest throughout long scripts.

How do I handle long scripts?

Split your script into segments of up to 4096 characters. Send each segment as a separate POST request to the /v1/audio/speech endpoint. Receive the resulting audio files and join them using standard audio editing software. This approach ensures compliance with API limits while maintaining seamless playback.

What format should I use for editing?

Use WAV or FLAC for editing. These formats preserve audio quality without compression artifacts, allowing precise cuts and adjustments. MP3 and OPUS are suitable for final distribution due to smaller file sizes. PCM is available for raw data processing. Select the format that best fits your post-production workflow.

Can I use this API for YouTube dubbing?

Yes, you can use this API for YouTube dubbing. Generated audio files are returned directly in the response and not stored by the API. You can download the files and upload them to your video platform. The synthetic watermark is inaudible, so viewers will not notice the artificial origin of the voice.

Do I need to attribute the voices?

Attribution is not strictly required by the API for usage, but the voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain readings. While the API does not enforce specific attribution, acknowledging the source is good practice. The synthetic nature is marked by an inaudible watermark.

How do I detect synthetic speech?

Synthetic speech is marked by an inaudible watermark embedded in the audio file. This watermark is designed to be transparent to human listeners but detectable by specialized software. It serves as a digital signature confirming the audio was generated by this API, aiding in content verification and transparency.

What is the character limit per request?

Each API request supports up to 4096 characters of input text. If your script exceeds this limit, you must split it into smaller chunks. Each chunk is processed independently, and the resulting audio files are then joined. Spaces and punctuation count toward the character total.

Is there a rate limit for agencies?

Yes, the rate limit is 60 requests per minute per API key. Before the first top-up, the limit is 10 requests per minute. This applies to all accounts, including agencies. If you exceed the limit, requests will be queued or rejected until the window resets. Monitor your usage via the dashboard.

Can I use multiple voices in one project?

Yes, you can use multiple voices in one project. Simply specify a different voice parameter for each API request. Generate each segment with the desired voice, then combine the audio files in your editing software. This allows for dynamic dialogue and varied character voices within a single dubbing project.

How do I split audio files manually?

Split audio files manually using any standard audio editor. Import the generated WAV or FLAC files, cut at the desired points, and export the segments. Ensure the cuts align with natural pauses in the speech. You can then reassemble the segments or use them independently as needed for your project.

Is the API suitable for live dubbing?

The API is not optimized for live dubbing. Requests are processed sequentially, and generation takes roughly real-time (25 characters per second). Short phrases may take a few seconds. For live use, pre-generate audio files and trigger them during playback. The queue under load may introduce latency.

What languages are best supported?

The API supports 23 languages, including English, Spanish, German, French, Japanese, and Chinese. Any voice can speak any supported language. If the language field is omitted, the system guesses the language, defaulting to English for Latin script. Explicitly setting the language ensures accuracy for non-English texts.

Do I need a developer to use this?

A developer is helpful but not strictly required. The API uses standard JSON and HTTP requests, compatible with tools like Postman or cURL. For complex integrations, such as automatic script splitting or voice switching, programming knowledge is needed. The documentation provides examples to assist with implementation.

Can I integrate this with my CMS?

Yes, you can integrate this API with your CMS. Use the standard HTTP endpoints to generate audio from text fields. Store the returned audio URLs in your database. Most CMS platforms can be extended with custom scripts or plugins to automate the TTS workflow, though no official plugin is provided.

What happens if I exceed the rate limit?

If you exceed the rate limit, requests may be queued or rejected. The API processes requests sequentially, so high volume can cause delays. Failed requests are free. Monitor your usage in the dashboard to adjust your request rate. Top-ups do not increase the rate limit, only the character allowance.

Is the watermarked audio detectable?

The watermark is inaudible to human listeners but detectable by specialized software. It marks the audio as synthetic speech. This feature helps verify the origin of the content. The watermark does not affect audio quality or playback, ensuring a natural listening experience while providing a digital signature for authenticity.

Selecting the Right Voice Presets

Sound wave illustration for: Selecting the Right Voice Presets

Dubbing agencies must align voice presets with content tone to ensure synthetic narration matches the original media’s emotional intent. The eight available presets span female and male ranges, each engineered for specific delivery styles.

Choosing correctly reduces post-production editing time. Agencies should evaluate each voice against sample scripts before committing to a full localization pipeline.

Choosing Between Female and Male Voices

Female voices include Aria (warm narrator), Nova (upbeat), Luna (soft storytelling), and Sage (calm instruction). Male options are Felix (energetic promo), Milo (conversational), Atlas (neutral news), and Orion (deep documentary). Select based on the target audience’s preference for the source material.

Matching Tone to Content Type

Match the preset to the genre: use Atlas for news segments or Orion for documentaries. For promotional content, Felix or Nova provide necessary energy. Luna suits narrative-driven pieces, while Sage works for instructional or explainer videos.

  • News: Atlas
  • Documentary: Orion
  • Promo: Felix or Nova
  • Storytelling: Luna or Aria

Voice Consistency Across Episodes

Maintain a single voice preset across all episodes of a series to preserve brand identity. The API does not support voice cloning, so consistency relies on selecting the correct preset initially. Use the same voice parameter in every POST request for a given series.

Language Support for Global Localization

Sound wave illustration for: Language Support for Global Localization

For localization agencies, consistent linguistic accuracy is as critical as vocal quality. Our API supports 23 languages, covering major markets from English and Spanish to Arabic, Hindi, and Japanese. This range allows agencies to produce content for diverse global audiences without needing multiple specialized providers.

The system automatically detects language from the input script, defaulting to English for Latin characters. For non-English Latin texts, setting the language parameter ensures correct pronunciation, preventing awkward accent artifacts that can undermine professional dubbing quality.

23 Supported Languages List

The API supports 23 languages: Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. Any voice can speak any of these languages, allowing flexible casting across episodes.

Not supported are Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, Bengali, and others. Verify your target market is on this list before committing to a production pipeline.

  • Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Turkish

Language Detection Logic

If the language field is omitted, the API guesses the language from the script. Latin script text defaults to English. For non-English Latin-script text, you must explicitly set the language parameter to avoid incorrect pronunciation.

Manual Language Override

To ensure precise localization, always set the language field with the ISO code. This overrides automatic detection, guaranteeing the voice uses the correct phonetic rules for your target audience, even if the script contains mixed scripts.

Audio Format Selection for Post-Production

Sound wave illustration for: Audio Format Selection for Post-Production

Dubbing agencies must choose audio formats that align with their post-production pipeline. The API supports MP3, WAV, FLAC, and OPUS, allowing you to balance file size, editing flexibility, and archival quality.

MP3 for Web Delivery

MP3 is ideal for final web delivery due to its small file size and universal compatibility. It is suitable for streaming platforms where bandwidth efficiency matters more than bit-perfect fidelity.

WAV for Professional Editing

WAV provides uncompressed audio, making it a common choice for professional editing. Use this format when you need to layer voiceovers, apply effects, or perform precise cuts without generational loss.

FLAC for Lossless Archiving

FLAC offers lossless compression, preserving full audio quality while reducing storage needs. It is a good choice for archiving master files that may be re-used or remastered later.

Handling Long Texts in Dubbing Scripts

Sound wave illustration for: Handling Long Texts in Dubbing Scripts

Dubbing scripts often exceed the 4096-character limit per API request. Agencies must split longer texts into manageable chunks before sending them to the endpoint.

Proper splitting ensures that audio generation completes efficiently without hitting payload constraints, allowing for seamless batch processing of entire episodes or segments.

4096 Character Split Limit

Each POST request accepts a maximum of 4096 characters, including spaces and punctuation. Exceeding this limit causes the request to fail, resulting in no charge but requiring a retry.

Split logic should respect word boundaries to avoid cutting mid-sentence. Agencies typically preprocess scripts to ensure each chunk is a complete semantic unit before submission.

  • Limit: 4096 characters per request
  • Includes spaces and punctuation
  • Failed requests are free

Audio Joining Strategies

After receiving individual audio files for each split, the pipeline must join them in the correct order. Use standard audio libraries to concatenate the output files.

Ensure consistent sample rates and formats (e.g., WAV or MP3) across all chunks before merging to prevent playback glitches or format mismatches.

Maintaining Audio Continuity

To preserve natural flow, consider adding slight crossfades or silence buffers between joined segments. This masks potential minor timing differences between generated clips.

Verify that the voice preset and speed settings remain identical across all splits to ensure consistent tone and pacing throughout the entire dub.

Pricing Strategies for High Volume

Sound wave illustration for: Pricing Strategies for High Volume

For dubbing agencies, understanding the unit cost is essential for accurate project quoting. With pricing at $10 per million characters, the expense is directly proportional to text length. Since spaces and punctuation count as characters, precise input formatting ensures predictable billing.

Calculating Cost Per Minute

Audio generation typically runs at a speed of roughly 25 characters per second. This means one second of audio costs approximately $0.00025. For a 60-second voiceover, expect a cost of $0.015. These figures help you estimate total production costs before committing to a script.

  • Rate: $10 per 1,000,000 characters
  • Generation speed: ~25 characters/second
  • Cost per second of audio: ~$0.00025

Budgeting for USDT Top-Ups

Payments are processed via USDT on the TRON network (TRC-20). There are no monthly subscriptions or invoices, allowing flexible cash flow management. The minimum top-up amount is $5, which provides 500,000 characters of usage. Balances do not expire.

Optimizing Character Count

Since every character counts toward the bill, editing scripts for conciseness directly reduces costs. Remove unnecessary pauses or filler words in the source text. However, be mindful that splitting long texts into segments under 4096 characters adds minor processing overhead. Streamlining your input data ensures you pay only for what is actually synthesized.

Technical Integration Workflow

Sound wave illustration for: Technical Integration Workflow

Integration begins with securing a unique API key via the /get-api-key/ page. This key authenticates all requests to the https://api.vs55.info/v1 endpoint. Since only one key is active per account, you can rotate it immediately via the dashboard if compromised.

You can integrate using the official OpenAI SDK by overriding the base_url, or via direct HTTP POST requests. Both methods support the standard JSON body structure. Direct HTTP offers lighter dependencies, while SDKs provide convenient client-side object handling for your localization scripts.

API Key Management

Generate your key on /get-api-key/. Store it securely in your environment variables. Rotate keys instantly from the dashboard; the previous key becomes invalid immediately upon issuance.

  • One active key per account
  • Instant invalidation on rotation
  • Bearer token authentication

SDK vs. Direct HTTP Requests

The API mirrors the OpenAI speech endpoint structure. Using the OpenAI SDK is straightforward: simply set base_url to https://api.vs55.info/v1. For lighter integrations, send standard POST requests with JSON bodies containing model, input, voice, and response_format.

  • Compatible with OpenAI SDK
  • Standard JSON request body
  • No custom SDK required

Error Handling in Pipelines

Implement robust error handling for rate limits (60 req/min) and transient failures. Failed requests are free, so retrying on 5xx errors is safe. Monitor your balance and character counts to avoid unexpected stops during high-volume dubbing batches.

  • Retry on 5xx errors
  • Free failed requests
  • Monitor rate limits (60 req/min)

Quality Control and Testing

Sound wave illustration for: Quality Control and Testing

Quality assurance is critical before deploying synthetic voices for dubbing. Pre-production tests ensure the selected preset matches the intended emotional tone and language accuracy. Agencies should generate short samples across different genres to verify consistency before scaling to full scripts. Monitoring latency helps identify queue delays during peak usage, ensuring timely delivery for tight production schedules.

Rate limit management prevents service interruptions. With a limit of 60 requests per minute, agencies must plan batch sizes accordingly. Understanding the 10 requests per minute starting limit is essential for new accounts. Proper pacing ensures smooth integration into existing dubbing pipelines without unexpected bottlenecks.

Pre-Production Voice Tests

Generate sample audio for each voice preset using diverse text types. Verify clarity, tone, and language detection accuracy. Use these samples to establish a baseline for quality control across different episodes and languages.

  • Test each of the 8 voices with genre-specific text
  • Verify language detection logic for mixed-script inputs
  • Document voice characteristics for team reference

Latency Monitoring

Track request processing time to identify queue delays. Short phrases may take a few seconds, while longer texts process at roughly real-time speed. Monitor these metrics to adjust batch processing times and maintain production efficiency.

Understand that requests are processed sequentially. Under load, waiting times may increase. Plan accordingly to avoid delaying final audio delivery.

  • Monitor average processing time per character
  • Identify peak latency periods
  • Adjust batch sizes based on queue behavior

Rate Limit Management

Adhere to the 60 requests per minute limit per key. New accounts start with 10 requests per minute until the first top-up. Plan API calls to stay within these limits, especially during high-volume dubbing projects.

Use the official OpenAI SDK or direct HTTP requests to manage rate limits effectively. Implement retry logic with exponential backoff to handle temporary throttling without losing progress.

  • Track requests per minute per API key
  • Plan batch sizes to avoid exceeding limits
  • Implement retry logic for temporary throttling

Legal and Licensing Compliance

Sound wave illustration for: Legal and Licensing Compliance

Dubbing Voice Metrics relies on voices derived from licensed recordings under CC BY 4.0 and public-domain readings. Understanding these licensing terms ensures your agency avoids infringement when commercializing synthetic content.

Understanding CC BY 4.0

The underlying voice models originate from LibriTTS-R, which uses CC BY 4.0 licensed data. This permits commercial use but requires proper attribution to the original dataset.

Attribution Requirements

Agencies must credit the voice sources in project documentation or credits. While the audio itself does not display text, your production pipeline should record the voice preset used for each scene.

  • Log voice preset (e.g., aria, atlas) per scene.
  • Cite LibriTTS-R and specific voice actor if available.

Watermark Detection Risks

Every generated file contains an inaudible watermark marking it as synthetic speech. This helps distinguish AI audio from human recordings but may trigger detection tools in media rights management systems.

Comparing Voice Personalities

Sound wave illustration for: Comparing Voice Personalities

Selecting the correct voice preset is critical for maintaining audience engagement across different dubbing contexts. This section compares key personality pairs to help agencies match tone to content type effectively.

Aria vs. Nova: Tone Differences

Aria offers a warm, clear narrator style ideal for storytelling or educational content. Nova provides a bright, upbeat energy suited for promotional or youthful material. Both support all 23 languages, but their distinct timbres affect listener retention differently.

Atlas vs. Orion: News vs. Documentary

Atlas delivers a neutral, news-ready delivery that ensures clarity and neutrality for informational scripts. Orion provides a deep, documentary-style voice that adds gravitas to long-form narratives. Choose Atlas for factual reporting and Orion for immersive storytelling.

Milo vs. Felix: Conversational vs. Promo

Milo is friendly and conversational, perfect for user guides or casual dialogue. Felix is energetic and promo-oriented, designed to capture attention quickly in advertisements. Both voices maintain consistent quality across short and long texts up to 4096 characters.

Limitations to Consider

Sound wave illustration for: Limitations to Consider

Dubbing workflows often rely on fine-grained control over delivery, but this API lacks support for SSML. Agencies cannot inject pause tags, adjust prosody, or modify emphasis via markup. This absence requires manual script adjustments or post-production editing to achieve desired pacing and tone, adding steps to the pipeline.

No SSML Support

Without SSML, precise timing and emphasis adjustments must be handled externally. Teams should plan for additional post-production time to manually align audio with video frames, as dynamic voice modulation is not possible at the API level.

No Real-Time Streaming

The API returns a complete audio file upon request completion, not a stream of partial audio. This means you must wait for the full synthesis to finish before receiving any data. For live dubbing or interactive applications requiring sub-second latency, this sequential processing model may introduce noticeable delays, especially under load.

Fixed Voice Parameters

Voices are static presets with no custom cloning or emotion tags. You select from eight fixed profiles, and each generated file carries an inaudible synthetic watermark. While this ensures consistency, it limits creative flexibility for unique character voices and may require specific detection strategies if watermark transparency is a legal concern.

Best Practices for Dubbing Agencies

Sound wave illustration for: Best Practices for Dubbing Agencies

Standardizing voice selection across projects reduces decision fatigue and ensures brand consistency. Dubbing agencies should define a mapping document that pairs specific content types—such as news or storytelling—with fixed voice presets like Atlas or Luna. This approach minimizes the risk of tonal mismatches when scaling production volumes.

Documenting API usage creates a reliable audit trail for billing and quality assurance. Teams should log character counts, selected voices, and response formats for every job. This data helps identify usage patterns, optimize budgeting for USDT top-ups, and quickly resolve any discrepancies in character billing or audio quality.

Training Teams on Limits

Teams must understand the 4096-character split limit and the need to join audio files manually. Proper handling of these splits ensures seamless playback without noticeable gaps or overlaps in the final dub.

Understanding the 60 requests per minute rate limit prevents pipeline bottlenecks. Training staff on batching requests or staggering uploads during peak hours maintains steady production flow without hitting caps.

  • Enforce 4096-character text splits before API submission
  • Monitor rate limits (60 req/min) to avoid throttling
  • Document voice presets per project type
  • Log character counts for accurate billing audits

Final Recommendations

Sound wave illustration for: Final Recommendations

For localization teams needing scalable, consistent synthetic narration, this API offers a cost-effective alternative to traditional recording pipelines. With support for 23 languages and a transparent pricing model, it streamlines the production of large-volume dubbing projects without the logistical overhead of coordinating human talent.

However, for nuanced emotional performances or specific brand voice requirements, human actors remain superior. Use this API for bulk content, e-learning, or standard narration where consistency and speed outweigh subtle emotional variation.

When to Use This API

Choose this API when you require rapid production of high-volume content across multiple languages. It is ideal for e-learning modules, audiobooks, and corporate training videos where voice consistency across episodes is more critical than unique emotional interpretation.

The ability to generate audio in 23 languages with a single key makes it a strong fit for global localization agencies managing diverse content libraries.

  • High-volume text-to-speech needs
  • Multi-language consistency requirements
  • Cost-effective bulk production

When to Choose Human Actors

Select human actors for projects requiring distinct emotional depth, unique character voices, or specific brand personality nuances that synthetic voices may not fully capture.

Use human talent for premium marketing campaigns or narrative-driven content where subtle vocal inflections significantly impact audience engagement.

Summary of Key Features

The API provides 8 distinct voice presets, 23 languages, and flexible audio formats (MP3, WAV, FLAC, etc.) via a simple REST interface.

With a clear pricing structure of $10 per million characters and USDT top-ups, it offers predictable costs for agencies scaling their dubbing operations.

  • 8 voice presets
  • 23 languages supported
  • Flexible audio formats
  • Transparent pricing ($10/1M chars)

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. Split logic should respect word boundaries to avoid cutting mid-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. LibriVox public-domain audiobooks
  4. ISO 639 language codes
  5. openai-node on GitHub
  6. openai-python on GitHub
  7. Creative Commons Attribution 4.0
  8. OpenAI API reference: create speech
  9. FFmpeg documentation
  10. W3C WAI: making audio and video accessible
  11. TRON developer hub: TRC-20
  12. LibriTTS-R corpus (OpenSLR 141)

More on this site