Write for the ear
Lead with the point. A listener cannot skim a long sentence, so give each thought its own space. Read the script aloud once before generating it in Text to Speech.
- Spoken copy only: keep camera notes, timestamps and editing directions in a separate document.
- One thought at a time: split sentences that need a second breath.
- Keep useful directions deliberate: supported feeling tags belong in brackets; ordinary production notes may be spoken aloud.
- Choose a voice for the project: an energetic character and a calm narrator can make the same words feel very different.
Rewrite what looks good but sounds awkward
Before proceeding, users who have not yet completed the initial configuration process should navigate to the settings panel.
Open Settings first. Finish the setup, then come back here.
Put the action before the condition.
The offer ends on 03/04/27. Save 12–15% on the $1.2k bundle.
The offer ends on March fourth, twenty twenty-seven. Save twelve to fifteen percent on the twelve-hundred-dollar bundle.
Choose the intended date explicitly. Write out ambiguous numbers and money.
Export CSV via API/CLI.
Export a C-S-V file through the A-P-I, or use the command line.
Decide which abbreviations should be read as letters.
Includes: offline mode, 2FA, dark theme, PDF/DOCX.
It includes offline access, two-step sign-in, a dark theme, and PDF or Word exports.
Turn a visual list into a sentence a listener can follow.
Keep a pronunciation note
Text normalization is available in Settings for numbers and symbols. Check the actual output rather than assuming it resolves every case.
| Problem | Try | Example |
|---|---|---|
| Name unfamiliar to the voice | A phonetic spelling | Saoirse → Seer-sha |
| An ambiguous word | Clearer context or a different word | wind the cable → coil the cable |
| A measurement | Words rather than symbols | 3.5 kg → three and a half kilos |
| An abbreviation | Letters separated or the full phrase | ETA → estimated arrival time |
Estimate the length, then measure it
At an illustrative pace of 150 words per minute, 30 seconds holds about 75 words. Real voices, pauses and delivery vary; use these figures to draft, then time the generated take.
- Leave breathing room for a title, demonstration or closing image.
- If the clip runs long, cut repeated ideas before raising Speed.
- Use punctuation for natural phrasing; unsupported pause markup can become unwanted speech.
- Use feeling tags sparingly when the line needs a specific mood.
| Target | Starting word count |
|---|---|
| 15 seconds | About 38 words |
| 30 seconds | About 75 words |
| 60 seconds | About 150 words |
| 3 minutes | About 450 words |
Generate sections you can repair
Text to Speech accepts up to 5,000 characters per generation, counting spaces, punctuation and tags. For a long project, use complete scenes or paragraphs that fit within that limit.
Check the credit cost before Generate. If one sentence fails, revise that section instead of rerunning the entire project. New accounts receive 50 free credits once; pricing explains Creator and monthly credit resets.
Scene / section: Spoken script: Voice: Speed: Feeling tags: Special pronunciations: Approved file: Target duration / actual duration:
Review it where people will hear it
Listen on headphones for small glitches, then a phone speaker for clarity. Download the approved audio and assemble the project in your own editor. MorningAgent creates the voiceover; it does not edit or publish your video.
- Every word is present, with no unintended production notes.
- Names, dates and amounts are correct.
- The voice and delivery fit the intended audience.
- The clip has enough space around the opening and ending.
- Adjacent sections use matching voice and settings.
- The approved audio, script and pronunciation notes are saved together.
Frequently asked questions
How many words fit in a 30-second voiceover?
About 75 at 150 words per minute, before extra pauses. Time your actual take and shorten the script if it feels crowded.
How do I fix a mispronounced name?
Try a phonetic spelling in a short test sentence, then keep the working spelling in your production notes. Captions should keep the correct written name.
Should I use digits or words?
Write out dates, prices and abbreviations whenever the intended reading is ambiguous. Text normalization can help, but important details still need a listening check.
Do spaces and tags count toward the limit?
Yes. The 5,000-character Text to Speech limit includes spaces, punctuation and feeling tags.