Frequently asked questions
What is the best voice for French narration?
All eight voices support French narration. Aria offers a warm, clear narrator style, while Luna provides a soft, storytelling tone. Sage is calm and instructional, and Nova is bright and upbeat. Felix, Milo, Atlas, and Orion also speak French fluently. Choose the preset that matches your podcast’s specific mood and pacing needs.
How long does it take to generate a 10-minute episode?
Generation takes roughly real-time, about 25 characters of text per second of generation. For a standard 10-minute episode, this takes approximately 10 minutes of processing time. Requests are processed sequentially, so under load, your request may wait in a queue before synthesis begins.
Can I change the voice mid-sentence?
No. Each API request generates audio for a single voice preset. To change voices, you must split your script into separate requests, generate each segment individually, and then join the resulting audio files together using your preferred editing software.
Does the API handle French accents correctly?
Yes. All eight voices can speak any of the 23 supported languages, including French. If you do not set the language parameter, the system guesses from the script, but for French text, explicitly setting the language ensures accurate pronunciation.
How do I split long scripts into chunks?
Split your text into parts of up to 4096 characters per request. Send each chunk via POST to the /v1/audio/speech endpoint. Then, join the returned audio files to create your final episode. This ensures each request stays within limits.
Is the audio MP3 or WAV better for podcasts?
MP3 is often preferred for podcasts due to its smaller file size and broad compatibility. You can choose between mp3, wav, opus, flac, or pcm formats. MP3 offers a good balance of quality and efficiency for streaming and distribution.
Can I use the generated audio on YouTube?
Yes. The generated audio files are returned directly in the API response and are not stored by us. You can download them and upload them to YouTube or any other platform. Each file carries an inaudible watermark marking it as synthetic.
How do I ensure consistent volume levels?
The API does not handle volume normalization. You should adjust volume levels in your own audio editing software after downloading the generated files. Use the speed parameter (0.25–4.0) to adjust pacing, which may indirectly affect perceived volume and rhythm.
What is the character limit for a single request?
Each request supports up to 4096 characters in the input field. Spaces and punctuation count toward this limit. For longer scripts, you must split the text into multiple requests and join the resulting audio segments manually.
Do I need technical skills to use the API?
Basic technical knowledge helps, but you can use the ready curl example provided on the /get-api-key/ page. You can also use the official OpenAI SDKs by changing the base_url. Plain HTTP requests work just as well for simple integrations.
How do I test different voices quickly?
Send short test texts to the /v1/audio/speech endpoint with different voice parameters. Check the /v1/voices endpoint to see all available presets. Use the free trial credit to test multiple voices without immediate cost concerns.
Can I generate multiple languages in one batch?
No. Each request generates audio for a single voice and language combination. To produce a multi-language episode, send separate requests for each language segment and join the resulting audio files together after generation is complete.
What happens if the server is slow?
Requests are processed one after another. Under load, your request may wait in a queue. Typical speed is roughly real-time, but short phrases may take a few seconds. Failed requests are free, so no charge occurs if synthesis fails.
Is there a way to preview audio before downloading?
The API returns the audio file directly in the response body. You can save this file immediately to your local machine or play it for preview. There is no separate preview endpoint; the output is the final audio.
Preparing Your Script
Generating French narration for your podcast starts with cleaning your script. The API accepts up to 4096 characters per request, so long scripts must be split into manageable segments. This ensures consistent audio quality.
Proper punctuation directly influences the AI’s pacing. Commas and ellipses create natural pauses, while periods signal full stops. Adjusting these elements helps the synthetic voice sound less robotic and more like a seasoned podcaster delivering a polished episode.
Text Length Optimization
Keep each input under 4096 characters. If your script is longer, split it into parts and join the resulting audio files later. This keeps processing time predictable, as requests are handled sequentially.
Short phrases may take a few seconds to generate, while longer texts approach real-time speed. Planning your segment lengths helps you estimate total production time accurately.
- Max 4096 characters per request
- Split long scripts into parts
- Join audio files post-generation
Punctuation for Pacing
Use commas for brief pauses and periods for longer breaks. The TTS engine interprets these as natural breathing points, making the French narration flow better for listeners.
Avoid excessive exclamation marks or unusual symbols, which might cause unnatural intonation. Stick to standard punctuation for the most predictable and professional result.
- Use commas for short pauses
- Use periods for full stops
- Avoid excessive exclamation marks
Removing Non-Spoken Text
Remove stage directions like [pause] or [laugh] unless you want them read aloud. The API does not support SSML, so clean text yields the best audio.
Ensure the script is purely the spoken words. This avoids awkward audio artifacts where the TTS voice reads out your editing notes, keeping the final podcast episode clean.
- Remove stage directions
- Exclude SSML tags
- Ensure pure spoken text
Choosing the Right Voice
Selecting the right voice is critical for maintaining listener engagement in your podcast. Our eight presets are curated to match specific tonal needs, ensuring your audio sounds professional without expensive studio time. Each voice offers a distinct character suitable for different segments of your episode.
Whether you need a warm narrator for intros or a calm instructor for deep dives, matching the voice to the content style enhances the overall listening experience. The following options cover the most common podcast formats.
Aria for Narration
Aria brings a warm, clear quality ideal for standard narration. Her tone is approachable yet polished, making her perfect for storytelling segments or episode intros where you want to establish a welcoming atmosphere immediately.
Use Aria when the focus is on your narrative voice. Her clarity ensures that every word is understood, reducing listener fatigue during longer monologues.
- Best for: Standard narration and intros
- Tone: Warm and clear
- Ideal length: Up to 4096 characters per request
Nova for Energetic Segments
Nova is bright and upbeat, designed to inject energy into your content. She is excellent for promotional segments, quick updates, or any part of the episode that requires a lively, engaging pace.
Her energetic delivery helps maintain high listener retention during transitions or when covering fast-paced topics. Use her to break up slower sections and keep the momentum going.
- Best for: Promos and energetic updates
- Tone: Bright and upbeat
- Use case: High-energy segments
Sage for Educational Content
Sage offers a calm, instructional voice that is perfect for educational content or detailed explanations. Her steady pace allows listeners to absorb complex information without feeling rushed.
She is ideal for tutorial segments, interviews, or any content where clarity and a relaxed tone are prioritized over high energy. Sage ensures your educational material sounds professional and authoritative.
- Best for: Educational content and tutorials
- Tone: Calm and instructional
- Use case: Complex explanations
Making the API Request
Generating French audio requires pointing your request to the correct API endpoint. By setting the base URL to https://api.playgb168.com/v1, you ensure compatibility with standard OpenAI SDKs. This simple configuration allows you to use familiar libraries while accessing our specific TTS engine for podcast narration.
The core request sends your script to the POST /v1/audio/speech endpoint. You must specify the input text, select a voice like Aria or Nova, and explicitly set the language field to fr. This ensures accurate pronunciation, as the system guesses Latin-script text as English by default.
Setting the Base URL
Configure your SDK’s base URL to https://api.playgb168.com/v1. This directs requests to our infrastructure while maintaining compatibility with existing OpenAI client libraries. Authentication is handled via the Authorization: Bearer header using your unique API key.
- Base URL: https://api.playgb168.com/v1
- Auth Header: Authorization: Bearer tts-...
Configuring the Payload
Send a JSON body containing model (tts-1 or tts-1-hd), input (your French script), and voice. Crucially, include language: "fr" to override default English detection. You can adjust speed between 0.25 and 4.0 to match your podcast’s pacing.
Use response_format to choose MP3 for web delivery or WAV for high-quality editing. Remember that the input text limit is 4096 characters per request. For longer episodes, split your script and concatenate the resulting audio files.
- Model: tts-1 or tts-1-hd
- Language: 'fr' (mandatory for non-English Latin text)
- Max Input: 4096 characters
Handling the Response
The API returns the audio file directly in the response body. If synthesis fails, no charge is applied, and you receive an error status. Successful requests return the binary audio data, which you can save directly as an .mp3 or .wav file for your production workflow.
Post-Processing Audio
After retrieving the audio files from the API, you will need to concatenate the segments if your script exceeded 4096 characters. The API returns individual files for each part, which you must stitch together in your preferred audio editor to create a seamless narration track.
Concatenating Segments
Since the API processes texts in chunks of up to 4096 characters, longer scripts result in multiple audio files. Use an audio tool to merge these files sequentially, ensuring there are no unintended pauses between segments. This maintains the natural flow of your podcast narration.
- Split text into 4096-character chunks before sending requests
- Merge returned MP3 or WAV files in order
- Check for silence gaps between concatenated parts
Adding Background Music
Once your narration is assembled, layer background music or sound effects to enhance the listening experience. Since the API does not support SSML, you must handle this mixing step externally. Ensure the music volume does not overpower the generated voice track.
Exporting to MP3
The API supports MP3 as a response format, which is ideal for podcast distribution due to its small file size and wide compatibility. You can request MP3 directly from the API or convert other formats like WAV or FLAC using your audio software. MP3 ensures fast loading for your audience.
- Request
response_format: mp3for smaller files - Convert WAV or FLAC if higher quality is needed
- Verify audio integrity before uploading to your host
Quality Control
Quality control ensures your French podcast audio sounds natural and professional before publishing. Since the API guesses the language from script context, verifying the output is crucial, especially for mixed-language segments or proper nouns that might confuse the model.
Listening for Errors
Play back the generated audio to catch any unexpected pauses or mispronounced words. The API processes requests sequentially, so ensure the audio file is fully received before editing. If a segment sounds rushed or unnatural, you may need to adjust the input text or punctuation.
- Listen for abrupt cuts or unnatural breathing sounds
- Verify that special characters or numbers are rendered correctly
- Check for any background noise if using WAV format
Adjusting Speed
The speed parameter allows fine-tuning between 0.25 and 4.0 to match your podcast’s pacing. Standard narration often sounds best around 1.0–1.2, but you can slow it down for educational content or speed up for quick updates. Experiment with different speeds to find the optimal rhythm for your audience.
- Use speeds below 1.0 for clearer articulation of complex terms
- Increase speed slightly for energetic segments or intros
- Avoid extreme speeds that may distort the voice’s character
Checking Pronunciation
Even with the language parameter set to 'fr', some proper nouns or loanwords might be mispronounced. You can correct this by phonetically spelling the word in the input text. Always review the final audio to ensure names and places are pronounced correctly for your French-speaking audience.
- Phonetically spell difficult proper nouns in the input text
- Set the
languagefield explicitly to 'fr' for Latin script text - Verify that accent marks do not cause synthesis errors
Integration with Podcast Platforms
PodcastVoice generates MP3 audio files via a standard HTTP response, which you can upload directly to your preferred podcast hosting provider like Anchor, Libsyn, or Buzzsprout. Since the API returns the file rather than a URL, your workflow involves downloading the generated audio and uploading it as a new episode asset within your host’s dashboard.
Once uploaded, your hosting platform handles the distribution to Apple Podcasts, Spotify, and other directories via RSS feed. The API does not interact directly with these platforms; you simply ensure the generated audio file is correctly attached to the episode metadata in your host’s CMS before publishing.
Uploading to Hosting Services
Download the MP3 file returned from the POST /v1/audio/speech endpoint. Most podcast hosts accept standard MP3 uploads (128kbps or higher) for episode assets.
Log into your hosting dashboard and locate the episode editor. Upload the generated file as the main audio track, replacing any placeholder audio if needed.
- Supported formats: mp3, wav, opus, flac, pcm
- Direct file upload via standard web interfaces
- No proprietary SDK required for upload
RSS Feed Updates
After uploading the audio file to your host, the RSS feed is automatically updated by your hosting provider. You do not need to manually edit the XML.
Ensure your podcast title and description in the host are updated to reflect the new episode’s content before hitting publish.
- ['RSS feed updates are handled by your hosting provider', 'No direct API integration with RSS needed']
Metadata Management
Include relevant keywords in your podcast episode title and description to improve discoverability. The API does not manage metadata; this is done entirely within your hosting platform.
Ensure the episode title accurately reflects the French narration content to help listeners find your show in directories.
- Metadata is managed via your hosting provider
- API returns audio, not metadata
- Use standard MP3 tags if needed
Cost Management
Precise cost management ensures your podcast production remains sustainable. With pricing set at $0.01 per 1,000 characters, understanding your text volume is crucial for budgeting episodes accurately.
Estimating Character Counts
Calculate the total characters in your script, including spaces and punctuation. For a standard 10-minute podcast, expect to generate roughly 25,000 characters per minute at normal speed, but verify exact counts in your drafting tool.
Monitoring Usage
Track your consumption via the dashboard, which displays characters left and balance. Failed requests are free, so you can test variations without penalty. Monitor your 60 requests per minute limit to avoid queuing delays during high-volume batches.
Budgeting for Episodes
At $10 per million characters, a 50,000-character script costs just $0.50. Plan your monthly budget by multiplying your average episode character count by your publishing frequency. Use the free $0.50 trial credit when testing new voices.

Troubleshooting Common Issues
When generating French narration, you might encounter unexpected pauses or incorrect language detection. These issues usually stem from how the API interprets your input text or hits its usage limits.
Silence in Audio
If you hear silence, check if your text exceeded the 4096-character limit per request. Split longer scripts into smaller chunks and join the resulting audio files afterward.
Wrong Language Detection
The API guesses the language from the script, defaulting to English for Latin text. For French content, explicitly set the language parameter to fr to ensure accurate pronunciation.
API Rate Limits
You are limited to 60 requests per minute per key (10 before the first top-up). If you hit the limit, your requests will queue, causing delays. Monitor your usage dashboard to avoid bottlenecks during batch generation.
Advanced Techniques
Advanced usage unlocks dynamic audio structures without external editing tools. By adjusting the speed parameter between 0.25 and 4.0, you can create energetic intros or calm outros within a single episode, maintaining consistent voice quality while varying pacing.
Although we do not support SSML, you can simulate volume changes by generating segments at different speeds or using external normalization. Dynamic voice switching is achieved by splitting your script into parts, calling the API with different voice presets, and concatenating the resulting audio files.
Dynamic Voice Switching
Split your script into logical segments and send separate API requests for each, assigning different voices like Aria for narration and Felix for promotional breaks. Concatenate the returned audio files to create a seamless, multi-voice episode.
Ensure consistent background noise or silence between segments if your editing software requires it. Since each request is independent, you can experiment with combinations to find the perfect tonal balance for your podcast style.
- Use
tts-1ortts-1-hdfor consistent engine quality across voices - Split text at natural pauses to avoid cut-off audio
- Concatenate files using standard audio software
Speed Variation
Adjust the speed parameter (0.25–4.0) to alter delivery pace without changing the voice identity. Faster speeds suit quick updates or lists, while slower speeds enhance clarity for complex explanations or emotional moments.
Test different speeds for French text to ensure natural intonation remains intact. Extreme speeds may alter perceived emotion, so always listen to the generated audio before finalizing your episode structure.
- Speed 1.0 is standard; 2.0 doubles pace
- Speed 0.5 slows delivery for emphasis
- Keep speed within 0.25–4.0 range
Volume Normalization
The API returns raw audio; normalize volume externally using tools like FFmpeg or Audacity to ensure consistent loudness across episodes. This prevents listeners from adjusting their volume frequently during playback.
While we do not provide built-in volume controls, you can influence perceived loudness by adjusting speed or choosing voices with naturally higher energy, such as Nova or Felix, for intro segments.
- Use loudness standards like -16 LUFS for podcasts
- Normalize after concatenating voice segments
- Check levels before uploading to hosting services
Licensing Your Audio
All audio generated via our API is yours to use commercially for podcasts, videos, and other media without additional licensing fees. You retain full rights to the synthesized speech, making it safe for monetized content.
Commercial Use Rights
You can integrate the generated MP3, WAV, or FLAC files into any project, including paid subscriptions or ad-supported podcasts. There are no restrictions on the number of listeners or platforms where your audio can be published.
- Unlimited commercial usage
- No per-listener fees
- Suitable for monetized content
Attribution Requirements
While not strictly mandatory, we recommend crediting "PodcastVoice" in your show notes or video descriptions. This helps the community grow and acknowledges the synthetic nature of the voice.
- Credit recommended but optional
- Mention in show notes
Watermark Understanding
Every file includes an inaudible digital watermark marking it as synthetic speech. This invisible tag does not affect audio quality or listener experience but serves as proof of origin for licensing purposes.
Scaling Your Production
Scaling your podcast production means moving from manual scripts to automated pipelines. By leveraging the API’s 4096-character limit, you can programmatically split long articles into segments, synthesize them in parallel, and reassemble the audio files. This reduces the per-episode generation time significantly compared to single-request processing.
Automating Workflows
Use simple scripts to send text to POST https://api.playgb168.com/v1/audio/speech. Since requests are processed sequentially, batching multiple short phrases into a single script loop ensures efficient queue management without overwhelming the 60 requests-per-minute limit.
- Use cron jobs to trigger synthesis when new blog posts are published
- Automatically split text into 4096-character chunks before sending API requests
- Integrate with Zapier or Make to connect your CMS to the TTS endpoint
Batch Processing
Process multiple episodes simultaneously by queuing requests. Since the API charges per character, you can estimate costs upfront by counting characters in your script. This allows you to budget accurately for seasonal content spikes without surprise charges.
- Pre-calculate character counts to estimate costs before generation
- Queue requests to stay within the 60 requests-per-minute limit
- Retry failed requests automatically, as they are free if synthesis fails
Team Collaboration
Share a single API key among your team, but issue a new key if someone leaves to immediately invalidate the old one. Since accounts are per site, each team member can have their own usage history if you track keys, or share one key for a unified billing view.
- Issue new API keys instantly to revoke access for former team members
- Track usage via the dashboard to monitor individual contributions
- Use consistent voice presets like Aria or Nova for brand consistency across episodes
Future-Proofing Your Content
Future-proofing your podcast audio relies on choosing versatile formats and maintaining consistent voice identities across episodes. By leveraging standard outputs like MP3 or WAV, you ensure compatibility with major hosting platforms and streaming services as they evolve. Our API supports multiple response formats, allowing you to adapt quickly to new distribution requirements without re-recording your content.
Format Compatibility
Selecting the right audio format is crucial for long-term accessibility. We support MP3, WAV, OPUS, FLAC, and PCM, ensuring your files work with current and future podcast players. Using lossy formats like MP3 reduces storage costs while maintaining quality for standard streaming, whereas FLAC offers lossless compression for archival purposes.
- MP3 for broad compatibility
- WAV for high-quality editing
- FLAC for lossless archival
- OPUS for efficient streaming
Voice Consistency
Consistency builds listener trust. By sticking to one voice preset for your host segments, you create a recognizable brand sound. Our eight distinct voices, from Aria’s warm narration to Felix’s energetic promo style, allow you to maintain a specific tone across all your episodes, even as you scale production.
- Use Aria for narrative segments
- Use Felix for energetic intros
- Maintain one voice per role
- Switch voices for characters
Platform Updates
Podcast platforms frequently update their RSS feed requirements and metadata standards. Our API returns clean audio files that integrate seamlessly with standard publishing workflows. Since we do not lock you into a proprietary ecosystem, you can easily migrate your audio library between different hosting services as your platform preferences change.
- Standard RSS compatibility
- Easy migration between hosts
- No proprietary file locks
Listen to the 8 voices
Each sample below was generated by this API from the same sentence. Each voice is built from licensed LibriTTS-R recordings, providing a professional foundation for high-quality audio generation.
23 languages
Any voice speaks any of these languages.
