Documentation Index
Fetch the complete documentation index at: https://docs.beyondwords.io/llms.txt
Use this file to discover all available pages before exploring further.
Overview
You can customize voices and associated settings in BeyondWords to better match your publishing needs. Voice models, voice settings, pronunciation rules, AI preprocessing, and pauses can all help you to improve the speech output.Set the voice model
When selecting or creating a voice in BeyondWords, you may need to choose an ElevenLabs voice model to use. The model determines how the voice sounds and behaves, shaping things like tone, pacing, and overall naturalness. There are up to three available models, depending on the language and voice type. We recommend using v3 where available:| Model | Description | Supported languages | Voice options |
|---|---|---|---|
| v3 | Most expressive and dynamic output. Best for engaging, varied narration. | 74 languages | |
| Multilingual v2 | Most stable and even output. Best for clear, consistent long-form narration. | 29 languages | |
| Flash v2.5 | Most accurate and controlled output. Best for replicating specific voices and delivery styles. | 32 languages |
voice_id, so you don’t need to set the model separately.
Change voice settings
Voice settings let you change how a voice sounds and performs, covering aspects such as speaking rate and expressiveness.Open voice settings
Go to “Content” → “Preferences” → “Voices” in your project.Then, click the settings icon alongside your chosen voice.
Configure settings
Adjust the settings as needed. Please note that some options are only available on selected voice models.
- Multilingual mode: When enabled, the voice automatically adapts to the detected language—ideal for content that mixes languages or includes foreign words
- Speaking rate: Controls the speed of speech. Values below 100% slow it down, while values above 100% speed it up.
- Expressiveness: Controls how expressive and varied the voice sounds. Lower values make speech steadier and more consistent. Higher values add emotion and natural variation but make delivery less consistent.
- Similarity: Higher similarity keeps the voice closer to the original speaker’s tone and character. Lower values allow more variation in delivery. Extremely high values may reduce naturalness.
- Style: Higher values make the voice delivery more stylized or expressive compared to the original. This can emphasize emotion or personality but may introduce less natural variation. Note that this is termed “style exaggeration” in ElevenLabs.
- Speaker boost: When enabled, this setting boosts the clarity and presence of the voice to better match the original speaker.
Add pronunciation rules
BeyondWords lets you define custom pronunciations for specific words or phrases. You can set pronunciation rules across all projects, individual projects, or individual content items. Each rule applies to every instance of the word within that scope. You can’t set a different pronunciation for a single occurrence, unless the word appears only once within that scope.Open pronunciation settings
If you’re using the Editor, highlight the word or phrase you’d like to add a pronunciation for and click “Add pronunciation rule”.Otherwise, go to “Content” → “Preferences” → “Pronunciations” in your project and click ”+ Pronunciation”.
Choose rule type
Choose what type of pronunciation rule you want to apply:
- Substitute: Pronounce as a different word or phrase. For example, you can set “CO2” to be pronounced as “carbon dioxide”.
- Say as word: Force an initialism to be pronounced as a word. For example, “UNESCO” would be pronounced as “Unesco” instead of “U.N.E.S.C.O.”.
- Say as letter sequence: Force an initialism to be pronounced as a series of letters. For example, “UNESCO” would be pronounced as “U.N.E.S.C.O.” instead of “Unesco”.
- Phonetic spelling (English only): Use IPA (International Phonetic Alphabet) to provide a phonetic spelling of your desired pronunciation. For example, you can set “Nice” to be pronounced as
ˈniːs. You can use the “Generate spelling” button to generate IPA suggestions.
Enter the word or phrase
Enter the word or phrase you want to change the pronunciation for, if applicable.
Define the change
If required, enter the required parameters for your rule type, such as the substitute word, target language, or IPA spelling.
Set the language and accent
Choose the language and accent (if applicable) that this rule should apply to.
Set the scope
Determine whether to apply the pronunciation rule across all projects, this project only, or this content item only.Please note that you can’t add a pronunciation rule for this occurrence only (i.e., the rule will apply to every instance of the word or phrase within the selected scope).
New pronunciations only affect existing content if that content is updated or regenerated.
Enable AI preprocessing
AI preprocessing (or LLM text normalization) adds an extra layer of intelligence to pronunciation, helping ensure words and phrases are spoken correctly based on context. For example, it can determine that “40–0” should be pronounced “forty–love” in the context of a tennis article.- Go to “Content” → “Preferences” → ” Pronunciations” in your project
- Click “AI preprocessing”
- Use the toggle to enable AI preprocessing
- Click “Done” to save changes
Add pauses
To further control the pacing of your narration, you can add default pauses after titles or between paragraphs, or insert pauses manually as needed.Add pauses by default
- Open your project and go to “Settings” → “General”
- Use the toggle(s) to enable “Pause after title paragraphs” and/or “Pause between body paragraphs”
- Enter the desired pause duration in seconds
- Click “Save changes”
Add pauses manually
- Open your content in the Editor
- Click at the end of a word or after punctuation, where you want to insert the pause
To add a pause at the end of a segment, insert the pause before a temporary word then delete the temporary word.
- Click “Add pause”
- Enter the desired pause duration in seconds, then click “Apply”
- Generate or regenerate your content