DocsText to SpeechText Guidelines

Text Guidelines for AI Voice Generation

Why Text Structure Matters

AI text-to-speech systems rely on punctuation, sentence structure, and contextual cues to determine pacing, tone, and emphasis. Clear formatting leads to:

  • More natural pauses
  • Better rhythm and flow
  • Improved pronunciation
  • More realistic emotional delivery

Punctuation Best Practices

Use full stops and commas correctly

Proper punctuation helps the AI understand where to pause.

Bad

Welcome to our website today we are going to explain how this works

Good

Welcome to our website. Today, we are going to explain how this works.

Avoid overly long sentences

Extremely long sentences can result in unnatural pacing. Break them into shorter, clearer segments.

Use question marks and exclamation points intentionally

These symbols influence intonation and can make speech sound more dynamic.

Formatting Tips

  • Use line breaks to separate logical sections.
  • Avoid excessive special characters.
  • Write numbers clearly when pronunciation matters.
  • Spell out acronyms if needed for clarity.

Bad

Our KPI increased by 25% in Q4.

Good

Our key performance indicators increased by twenty five percent in quarter four.

Using Expressive Bracket Directives

Lexora supports bracket-based expressive directives that enhance emotional realism. You can insert specific cues inside square brackets to guide vocal delivery.

How it works

Add emotional or contextual instructions inside square brackets directly within the text. These directives influence tone and delivery.

Examples

Hello everyone. [laughs] I'm really excited to be here today.

I didn’t expect that… [sighs] but we’ll figure it out.

That was absolutely incredible! [excited]

Common expressive directives

  • [laughs]
  • [sighs]
  • [whispers]
  • [excited]
  • [serious tone]

Use directives naturally and sparingly. Overusing them may reduce realism.

Writing for Different Use Cases

Blog Articles

Use structured paragraphs and moderate pacing. Add subtle expressive cues if storytelling is involved.

Product Pages

Keep sentences clear and informative. Avoid excessive emotional directives.

Storytelling & Narration

Use emotional directives strategically to enhance immersion.

AI Tweaks

Lexora includes AI Tweaks to help you improve your text before generating speech. By clicking the button in the bottom-left of the editor, you can ask Lexora to rewrite your text with a specific goal.

What you can do with AI Tweaks

  • Make it catchier — improve hooks, rhythm, and clarity for a more engaging listen.
  • Make it shorter — compress the message while preserving the meaning.
  • Make it more emotional — add warmth, emphasis, and expressive delivery cues.

Best practices

  • Use AI Tweaks before choosing a voice, so you can match the final tone more accurately.
  • Review the rewritten text and keep the parts that match your brand voice.
  • For emotional outputs, consider adding bracket directives like [laughs] or [sighs] where it feels natural.

Credits and cost

AI Tweaks consume credits using the same rule as speech generation: 1 credit = 1 character. The final cost depends on the number of characters processed by the tweak (including spaces and punctuation).

Tip: if you only need a small improvement, tweak a paragraph at a time to keep credit usage predictable.

Importing Files into the Editor

Lexora also lets you import text directly from a document. Instead of pasting content manually, you can upload a supported file and use its contents as the text for your speech.

Supported file formats

  • .pdf
  • .docx
  • .txt

How it works

Click the Import content button in the editor and choose a supported document. Lexora will extract the text and replace the current editor content with the imported text.

Why this is useful

  • Quickly turn existing documents into speech
  • Avoid manual copy and paste
  • Reuse blog drafts, reports, scripts, or text documents

Best practices

  • Review the imported text before generating speech.
  • Clean up formatting issues such as broken line breaks or unnecessary symbols.
  • For best results, make sure the original file has clear punctuation and structure.

Once imported, the document content behaves exactly like normal editor text: you can edit it manually, apply AI Tweaks, choose a voice, and generate speech as usual.

Common Mistakes to Avoid

  • Missing punctuation
  • All caps text (may alter tone)
  • Overuse of expressive directives
  • Very dense paragraphs without pauses
Overview
Voices