How to Generate High-Quality French Audio for Ads with Our TTS API?

By AuraCast editorial team · Updated 2026-11-19

This guide explains how to generate French audio for advertising campaigns using the AuraCast TTS API. We detail the exact workflow for startups and agencies producing localized radio spots, covering script preparation, voice selection, and format optimization. The process relies on a standard JSON payload compatible with existing OpenAI SDKs, ensuring seamless integration into your production pipeline without requiring new libraries or complex setup procedures.

You will learn to specify the French language code explicitly, select appropriate voices like Aria for narration or Felix for promos, and manage character limits efficiently. The guide provides concrete examples for API configuration, error handling, and cost estimation based on our per-character pricing model. This resource is designed for technical teams who need reliable, consistent synthetic voices optimized for commercial use without enterprise overhead.

How to Generate High-Quality French Audio for Ads with Our TTS API?

Frequently asked questions

Do I need to specify the language for French?

Yes. Although the API guesses language from the script, Latin text defaults to English. To ensure correct French pronunciation, explicitly set the language parameter to "fr" in your request body. This guarantees the synthetic voice generates accurate French audio for your ads.

Which voice sounds most natural in French?

A: All eight voices support French. Aria offers a warm, clear narrator style. Nova provides a bright, upbeat tone. Luna is soft for storytelling, while Sage delivers calm, instructional clarity. Felix, Milo, Atlas, and Orion also speak French fluently. Each voice is based on licensed recordings.

Can I mix English and French in one request?

Yes. You can include both languages in the input text. However, if the script contains non-English Latin text, you must set the language parameter to "fr" to ensure correct pronunciation. The API processes the text up to 4096 characters in a single request.

How do I handle pauses in the script?

Use standard punctuation like periods, commas, or ellipses within the input text. The TTS engine interprets these pauses naturally. There is no SSML support, so you cannot use custom tags. Adjusting speed or voice selection may also influence the perceived rhythm of the audio.

Is SSML supported for French prosody?

No. The API does not support SSML, phoneme dictionaries, or custom pronunciation lexicons. Prosody is handled automatically by the model. For precise control, you must rely on standard punctuation in the input text or adjust the global speed parameter between 0.25 and 4.0.

Can I change the speed mid-script?

No. The speed parameter applies to the entire request. You can set a speed between 0.25 and 4.0, but it remains constant throughout the generated audio. To vary speed, split your script into multiple requests and join the resulting audio files manually.

What is the maximum character limit per call?

The maximum input length is 4096 characters per request. For longer scripts, split the text into parts of up to 4096 characters. Generate each part separately and join the resulting audio files. Failed requests are free, but successful ones incur charges based on character count.

How do I ensure high quality for radio?

Use the tts-1-hd model for higher fidelity. Select voices like Aria or Luna for professional narration. Ensure you set the language parameter correctly. The audio is returned directly in the response. Generated files carry an inaudible watermark marking them as synthetic speech for compliance.

Do I need to pay for each character separately?

No. Pricing is $10 per 1 million characters, or $0.01 per 1,000 characters. Spaces and punctuation count as characters. You pay based on the total characters in your input text. Failed requests are free, and the charge is returned in full if synthesis fails.

Can I use the API for live reading?

No. The API does not support streaming, WebSocket, or real-time partial audio. Requests are processed sequentially, and generation takes time roughly proportional to text length. It is designed for batch processing, not live broadcasting. Use standard HTTP requests to receive the final audio file.

How do I download the generated audio?

The audio is returned directly in the API response body. You can save it using the specified response format, such as mp3, wav, opus, flac, or pcm. No separate download endpoint exists. The audio is not stored by the service after the response is sent.

Is there a limit on concurrent requests?

Yes. The rate limit is 60 requests per minute per key. Requests are processed one after another, so under load, a request may wait in a queue. If you exceed the limit, you may experience delays. Monitor your usage via the dashboard to manage concurrency.

Can I use the audio for TV commercials?

A: Yes. The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain readings. The audio is synthetic and carries an inaudible watermark. It is suitable for commercial use.

Preparation

Sound wave illustration for: Preparation

Before generating audio, determine the character count of your script. Each request accepts up to 4,096 characters, including spaces and punctuation. Longer scripts require splitting into segments to ensure successful synthesis.

Select a voice that matches your ad’s tone, such as Aria for narration or Felix for promos. Choose a format like MP3 for web delivery or WAV for broadcast standards.

Script character count

Count every character in your source text, including spaces and punctuation. The API limits each request to 4,096 characters. Exceeding this limit causes the request to fail.

For scripts longer than 4,096 characters, split the text into smaller chunks. Generate each chunk separately and join the resulting audio files to create the final spot.

  • Count spaces and punctuation as characters
  • Limit: 4,096 characters per request
  • Split longer texts into multiple requests

Voice selection strategy

Choose a voice preset that aligns with your brand’s identity. Aria offers a warm, clear narrator style suitable for storytelling. Felix provides an energetic tone for promotional content.

Ensure the selected voice supports French. All eight presets can speak French, but consistency across spots strengthens brand recognition.

  • Aria: warm, clear narrator
  • Felix: energetic, promo style
  • Consistent voice for brand recognition

Format choice

Select MP3 for web delivery due to its small file size and broad compatibility. Use WAV for broadcast if higher fidelity is required.

Opus offers efficient compression for streaming applications. FLAC provides lossless quality without the large size of WAV.

  • MP3: web delivery
  • WAV: broadcast quality
  • Opus: efficient compression
  • FLAC: lossless quality

API Configuration

Sound wave illustration for: API Configuration

For advertising agencies leveraging AuraCast, correct API configuration is the foundation of reliable audio generation. The base URL points to our specific endpoint, ensuring requests are routed to the correct infrastructure without ambiguity.

Setting the base URL

Configure your HTTP client to target https://api.km12.info/v1. This base URL is essential for all subsequent requests, including speech synthesis and voice retrieval. Unlike generic endpoints, this address ensures compatibility with standard OpenAI SDKs while routing traffic through our specific routing layer.

Adding the Bearer token

Include your unique API key in the Authorization header as Bearer tts-.... This token authenticates your account and tracks character consumption against your balance. Ensure the key is transmitted securely, as it provides direct access to your generated audio assets and billing details.

  • Keys are issued per account and replace previous keys immediately upon renewal.

Specifying the model

Select either tts-1 or tts-1-hd in your request body. Both models use the same engine and pricing structure, delivering consistent synthetic voices optimized for commercial use. Choose based on your quality requirements, knowing that both support the full range of 23 languages including French for localized ad content.

Language Specification

Sound wave illustration for: Language Specification

Specify the language parameter explicitly when generating French audio. While the API guesses the language from Latin script by defaulting to English, advertising copy often contains English loanwords or technical terms that trigger incorrect voice selection. Providing the ISO code ensures consistent output.

Explicit language code

Use the two-letter code fr in your request body. This overrides the default English assumption for Latin-script text. It guarantees the engine applies the correct phonetic rules for French pronunciation, which is critical for accurate ad delivery.

  • Use "language": "fr" in the JSON payload.
  • Applies to all 8 voice presets.
  • Overrides default Latin-script behavior.

Why specify 'fr'

Without explicit specification, the system defaults to English for Latin text. This results in a French-sounding voice pronouncing words with an English accent, which damages brand credibility in localized campaigns. Explicit codes eliminate this ambiguity.

Avoiding misidentification

Text containing English words or mixed languages may be misidentified. Always set the language field to match the primary audio output. This prevents the engine from switching phoneme models mid-sentence, ensuring uniform quality across the entire ad spot.

  • Mixed-language scripts require explicit codes.
  • Prevents mid-sentence accent shifts.
  • Ensures consistent ad tone.

Voice Selection

Sound wave illustration for: Voice Selection

Selecting the right voice preset determines how your French advertisement is perceived by the listener. AuraCast offers eight distinct presets, each optimized for specific tonal requirements. Choosing the appropriate voice ensures your message aligns with the intended emotional context, from warm narration to energetic promotion.

Aria for narration

Aria provides a warm, clear narrator voice. Her tone is steady and engaging, making her suitable for longer scripts that require consistent listener retention. Use Aria when the narrative arc is central to the advertisement’s impact.

  • Best for: Story-driven commercials
  • Tone: Warm and clear
  • Use case: Brand storytelling

Felix for promos

Felix delivers an energetic, promo-oriented voice. His upbeat delivery captures attention quickly, making him a strong choice for short radio spots or digital ads. Use Felix to emphasize urgency and drive immediate listener action.

  • Best for: Promotional spots
  • Tone: Energetic and upbeat
  • Use case: Sales announcements

Atlas for news

Atlas offers a neutral, news-style voice. His delivery is straightforward and professional, suitable for informational ads or corporate announcements. Use Atlas when a balanced, trustworthy tone is required without emotional embellishment.

  • Best for: Corporate announcements
  • Tone: Neutral and professional
  • Use case: Informational spots

Speed Adjustment

Sound wave illustration for: Speed Adjustment

Adjusting the speed parameter allows you to fine-tune the delivery pace of your French radio spots, ensuring the voice matches the intended emotional tone and timing constraints of your advertisement.

Standard ad pace

The default speed of 1.0 provides a natural, clear narration style suitable for most standard ad formats. This baseline ensures high intelligibility for French listeners, maintaining professional quality without sounding rushed or overly slow.

  • Default speed: 1.0

Faster delivery options

Increasing the speed up to 4.0 allows for concise spot announcements or high-energy promos where time is critical. Faster delivery can fit longer scripts into shorter ad slots, reducing per-spot audio duration without losing content.

  • Max speed: 4.0

Clarity vs. Speed

While higher speeds reduce generation time and character consumption, excessive acceleration may compromise clarity for complex French phonetics. Test your specific script at various speeds to balance broadcast readiness with production efficiency.

Format Output

Sound wave illustration for: Format Output

The response_format field determines the audio container used for your generated radio spots. Selecting the correct format ensures compatibility with your ad servers and broadcast infrastructure while balancing file size and quality.

MP3 for web

MP3 is the standard for digital advertising. It offers a compact file size that loads quickly on websites and mobile apps. This format is ideal for online radio spots and social media ads where bandwidth efficiency matters.

  • Small file size for fast delivery
  • Universal support across all ad platforms
  • Ideal for web-based radio spots

WAV for broadcast

WAV provides uncompressed, high-fidelity audio suitable for professional radio broadcasting. Use this format when submitting files to traditional radio stations or for archival purposes where audio quality is prioritized over storage space.

  • Uncompressed audio quality
  • Standard for professional radio submission
  • Preserves all audio details

Opus for efficiency

Opus delivers high-quality audio at lower bitrates than MP3. It is efficient for applications where bandwidth is limited but quality remains critical. This format is useful for data-sensitive mobile advertising campaigns.

  • High quality at low bitrates
  • Optimal for data-constrained environments
  • Efficient for mobile ad campaigns

Testing

Sound wave illustration for: Testing

Before generating full ad spots, validate your French audio output using short phrase tests. Send concise texts to the API to verify voice selection and pronunciation accuracy without consuming significant budget.

Short phrase tests

Submit simple sentences to verify voice selection and pronunciation. This confirms the API correctly interprets your French text before committing to longer scripts.

Full script validation

Generate the entire 4096-character limit to ensure seamless audio joining. Verify that no gaps or errors occur at segment boundaries during batch processing.

Quality check

Listen for natural pacing and clear articulation. Ensure the selected voice matches your brand tone and that the audio file meets your broadcast standards.

Testing

Integration

Sound wave illustration for: Integration

Seamless integration allows your production pipeline to scale without manual intervention. By leveraging the standard HTTP endpoints, you can automate the generation of localized radio spots for multiple markets simultaneously.

Batch processing

Split long scripts into 4096-character segments and queue them sequentially. The API processes requests in order.

  • Queue multiple scripts for simultaneous generation
  • Handle up to 60 requests per minute
  • Automatic character counting for budget accuracy

Audio file management

Store generated MP3 or WAV files directly in your asset library. Since audio is returned in the response and not stored by AuraCast, you control the lifecycle of every file.

  • Download instantly via HTTP response
  • Support for MP3, WAV, OPUS, FLAC, PCM
  • No storage limits or expiration

Workflow automation

Integrate the API with your existing CMS or ad delivery systems using standard JSON payloads. This eliminates manual audio uploads and reduces time-to-market for localized campaigns.

  • Standard JSON request structure
  • Compatible with standard HTTP clients
  • Easy integration with CI/CD pipelines

Common Errors

Sound wave illustration for: Common Errors

API requests often fail due to simple configuration oversights. The most frequent issues involve incorrect voice identifiers, exceeding character limits, or hitting rate limits too quickly during testing.

Invalid voice code

Ensure you use one of the eight valid presets: aria, nova, luna, sage, felix, milo, atlas, or orion. Sending an unrecognized voice name returns an error immediately.

Text too long

Each request supports up to 4096 characters. If your script exceeds this, split the text into smaller parts and join the resulting audio files afterward.

Rate limit hits

New keys allow 10 requests per minute. After the first top-up, the limit increases to 60 requests per minute. Exceeding this results in temporary rejection.

Cost Estimation

Sound wave illustration for: Cost Estimation

Pricing is straightforward: $10 per 1 million characters, including spaces and punctuation. Failed requests incur no charge, ensuring you pay only for successful synthesis. This model eliminates hidden fees for startups producing localized radio spots.

Counting characters

Every character in your input text counts toward the total. A typical 30-second ad script contains approximately 60–70 characters per second of audio. For a standard 30-second spot, expect to use roughly 750 characters.

  • Include spaces and punctuation in your character count.
  • Text exceeding 4,096 characters must be split into multiple requests.
  • Failed requests are free and do not affect your balance.

Budgeting per spot

At $0.01 per 1,000 characters, a standard 30-second French ad costs approximately $0.0075. This low unit cost allows agencies to test multiple voice variations or language adaptations without significant overhead.

  • No monthly plans or minimum spend requirements.
  • No VAT or invoices; pay via USDT (TRC-20).
  • Balance does not expire, allowing for long-term campaign planning.

Volume discounts

The rate remains constant at $10 per million characters regardless of volume. There are no tiered discounts or enterprise contracts. However, the absence of monthly fees means you only pay for what you generate, scaling directly with your ad output.

  • No volume-based price reductions.
  • Pay-as-you-go structure suits variable campaign schedules.
  • Top-ups start at $5 with automatic crediting.

Best Practices

Sound wave illustration for: Best Practices

Consistent voice selection ensures brand recognition across your localized ad campaigns. By locking a single preset, you maintain auditory continuity whether the spot runs in French, German, or English. This consistency reduces listener confusion and strengthens the association between your brand identity and the synthetic voice.

Proper text preparation directly impacts synthesis quality. Clean, well-structured input reduces processing time and minimizes the risk of mispronunciation or awkward pacing. Advertisers should prioritize clarity over stylistic flair in the source script to ensure the TTS engine delivers a professional result.

Clean text formatting

Ensure your input text is free of unnecessary whitespace or line breaks that the API might interpret as pauses. Simple paragraphs work best for most ad scripts. This approach keeps the character count accurate and prevents unexpected audio artifacts.

Standardize punctuation for consistent pacing. Use periods for full stops and commas for brief pauses. Avoid mixing quotation styles unless necessary, as this can sometimes confuse the language detection model, especially when switching between Latin and non-Latin scripts.

  • Use simple paragraph breaks for distinct script sections.
  • Standardize punctuation to control pacing and pauses.
  • Verify text length stays within 4096 characters per request.

Avoiding special chars

Special characters such as em dashes, ellipses, or non-standard symbols can disrupt the synthesis engine. Replace them with standard ASCII equivalents to ensure reliable pronunciation. For example, use '...' instead of '…' and '--' instead of '—'.

If your script includes numbers, dates, or currency symbols, verify they convert correctly in your test runs. While the API handles most standard inputs, complex formatting may require manual adjustment to avoid audio glitches.

  • Replace em dashes with standard hyphens.
  • Use standard ellipses instead of special characters.
  • Verify number and currency formatting in test audio.

Consistent voice use

Stick to one voice preset per campaign to maintain brand consistency. Switching between 'Felix' for promos and 'Atlas' for news within the same ad series can confuse listeners. Consistency builds trust and recognition.

Document your chosen voice in your internal style guide. This ensures that any team member or automated workflow uses the same preset for all generated audio files, maintaining a uniform auditory experience across all platforms.

  • Lock one voice preset per campaign.
  • Document the chosen voice in your style guide.
  • Ensure all team members use the same preset.

Support

Sound wave illustration for: Support

Support for AuraCast is handled exclusively via email. We do not offer phone lines, live chat, or scheduled call sessions. Our team reviews all inquiries and responds during standard business hours, ensuring that technical and billing questions receive accurate, documented answers without the noise of real-time interruptions.

For immediate reference, consult our documentation which details every API parameter, error code, and integration pattern. We do not maintain a public community forum or Discord server; all knowledge is centralized in the official guides to ensure accuracy and prevent version drift across our partner network.

Email contact

Send all questions to [email protected]. Include your API key, request ID, and a clear description of the issue. We prioritize technical integration errors and billing discrepancies. Expect a response within one business day.

None

Documentation links

Access the complete API reference at /docs/. It includes cURL examples, voice descriptions, and format specifications. The documentation is updated with every API change and serves as the single source of truth for implementation details.

None

  • https://km12.info/docs/
  • https://api.km12.info/docs

Community resources

We do not host external community forums or user groups. All best practices, common error solutions, and integration tips are documented in the official guides. This ensures you always reference the correct, up-to-date information for your specific account setup.

None

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. This ensures that your radio spots maintain the same tonal quality and pacing across all episodes and campaigns.

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. Creative Commons Attribution 4.0
  3. LibriTTS-R corpus (OpenSLR 141)
  4. FFmpeg documentation
  5. ISO 639 language codes
  6. openai-node on GitHub
  7. curl documentation
  8. OpenAI API reference: create speech
  9. MDN: Web Speech API
  10. LibriTTS-R paper (arXiv 2305.18802)
  11. Tether (USDT)
  12. Python documentation

More on this site