Prepare a two or three character voice script
Assign voices deliberately and keep dialogue understandable before generating audio.
Choose the cast before writing
In the voice generator, choose a single voice or a two or three character dialogue. Give each character a distinct label, then select the appropriate available voice and language. A label helps you assign lines; it is not automatically spoken in the finished audio. Keep the cast small enough that a listener can follow it.
Write a short first scene
For a two-person explainer, start with a question and an answer. Example: Host: “What should I check before publishing?” Guide: “Confirm the facts, remove unsupported claims, and read the draft aloud.” Assign each line to its speaker. Use actual dialogue rather than placing instructions such as “laugh here” inside text you do not want spoken.
Check the text and delivery settings
Review the character count before generating. The current studio accepts up to 12,000 characters overall and 30 dialogue lines. Long text is split for synthesis behind the scenes. Rate and pitch affect delivery, so test a short excerpt before producing a long narration. A voice that sounds natural in one language may not handle every name or abbreviation as expected.
Review the complete result
Listen for pronunciation, pauses and the correct speaker order. If a name is mispronounced, try a clearer spelling in a short test. Avoid changing every voice setting at once. After a successful generation you can play or download the combined audio and revisit saved results in your account. Generation requires an internet connection and can fail if the provider is unavailable.
Use voices responsibly
Only submit text and material you have permission to use. Do not present synthetic audio as a real person’s statement without appropriate disclosure and permission. Avoid including private information in a public narration. Review the output before sharing it, especially where pronunciation could change the meaning.
Plan speaker contrast without making dialogue distracting
Choose voices that remain understandable when played consecutively. A short host question followed by a longer guide answer can create contrast without extreme pitch or speed changes. Use a third voice only when it has a role, such as a customer giving an example. Write the purpose of each speaker before assigning lines. If two speakers give nearly identical explanations, combine them. Distinct roles often improve clarity more than dramatically different voices.
Prepare numbers, names and abbreviations
A voice can read a date, product code or abbreviation differently from what you intend. Write a short test line containing the difficult phrase. For an abbreviation, try the full phrase if spelling out its letters sounds wrong. For a price, make the currency explicit in the script. Keep the source facts unchanged while adjusting the spoken wording. Listen to the generated test instead of assuming that the written appearance guarantees correct pronunciation.
Estimate duration before generating
Use the public Word Counter to make a rough speaking-time estimate. A 300-word script at 150 words per minute gives about two minutes before extra pauses. This is a planning assumption, not an exact prediction of a selected voice. Dialogue pauses, language, punctuation and rate settings all affect the final duration. If you need a one-minute clip, shorten the content first rather than accelerating a long script until it becomes difficult to understand.
Recover from a failed generation
Keep a copy of your script before starting. If an error appears, check that every line has text, a speaker and an available voice. Try a short excerpt to distinguish an input problem from a service problem. When the provider is unavailable, waiting and trying later is more useful than submitting the full script repeatedly. Report the displayed error and approximate time through Contact if it continues. After a successful result, listen from beginning to end before downloading or using it in a public video.
A short three-character practice script
Illustrative example. Replace the sample facts with your own verified information.
Host: What are we making today? Guide: A short voice scene. First, assign a voice to each speaker. Learner: Can I test just one sentence? Guide: Yes. Check the names and pronunciation before generating the full script. Host: Then listen to the complete result before sharing it.
Before you use the result
- Is every dialogue line assigned to the intended speaker?
- Have you tested names and pronunciation with a short excerpt?
- Do you have permission to use the text, and is synthetic audio disclosed when appropriate?
Common questions
Will the character label be spoken?
The studio uses the assigned speaker to select a voice. Keep labels and stage directions out of the actual text field when you do not want them narrated.
Is the speaking-time estimate exact?
No. It uses your selected words-per-minute assumption. Pauses, language and the chosen voice can change the finished duration.