en-US · American English

English AI voiceover

An English AI voiceover here means eight American English voices, four styles in a male and a female version, reading the narration the director wrote for your video. Narration is billed by length, about half a credit for a typical scene line, and captions are derived from the same text automatically.

The English voices, with samples

Each sample reads the same line so you can compare them fairly.

Puck

energetic · male

This is how I sound reading your script at a normal pace, with no music behind me.

Laomedeia

energetic · female

This is how I sound reading your script at a normal pace, with no music behind me.

Charon

instructional · male

This is how I sound reading your script at a normal pace, with no music behind me.

Erinome

instructional · female

This is how I sound reading your script at a normal pace, with no music behind me.

Umbriel

warm · male

This is how I sound reading your script at a normal pace, with no music behind me.

Sulafat

warm · female

This is how I sound reading your script at a normal pace, with no music behind me.

Iapetus

neutral · male

This is how I sound reading your script at a normal pace, with no music behind me.

Aoede

neutral · female

This is how I sound reading your script at a normal pace, with no music behind me.

English is the default, so the field can stay empty

The generator's Language field accepts anything you type and offers 32 suggestions, but an empty field already resolves to American English. The director then writes the narration and every on-screen line in English from your brief, picks a style and a gender per video, and each scene arrives with its script attached. If you would rather have a British or Australian read, the voice dialog on any scene searches the full Google catalogue, which carries en-GB and en-AU voices well beyond these eight. Swapping the voice on one scene does not touch the others, which is how a two-presenter structure gets built.

Four styles, and where each one lands

Puck and Laomedeia are the energetic pair, the default for promos and UGC style ads. Charon and Erinome read instructions clearly and slowly, and tutorials are locked to that pair by a hard rule: even if the plan asks for an upbeat voice, a tutorial is forced back to the instructional one. Umbriel and Sulafat are warmer, suited to testimonials and local businesses. Iapetus and Aoede sit in the middle, corporate without being cold. The full voice name embeds the locale, en-US-Chirp3-HD-Charon for example, which is why a voice cannot be carried across languages.

The controls each scene gives you

Every scene holds its own narration text, a delay in seconds before the voice starts, and a speed between 0.85x and 1.25x. The music bed ducks to 22 percent of its volume while the voice speaks, ramping down and back over about half a second, so you never fight the soundtrack manually. Each synthesised clip is billed by length, about 1 credit per 330 characters with a minimum of half a credit, and regenerating after an edit is billed again. An export bills its render time, usually a small amount; longer cuts, or exports that still need clips made, need at least 1 credit left to start.

Writing lines the voice reads well

The voice reads exactly what is written, and there is no phonetic control, so a brand name it mangles gets fixed by respelling it in the narration box, Kiehl's as Keels, for instance. Write 2,500 as twenty five hundred if that is how you say it, because the digits alone leave the choice to the engine. Sentence punctuation drives the caption cues, so short sentences produce short readable pills. One more prompt rule worth knowing: the director never uses an em dash in narration, and pasting one into a line you edit by hand works against the pacing the captions expect.

Captions in this script

English captions type on word by word, timed to the moment the voice actually says each word. The premium voices refuse timing marks, so the app measures the same sentence on a sibling voice and scales those times to the clip; the spoken word brightens and slightly enlarges while the rest of the line waits at reduced opacity. The pill uses the video's own font and stays a fixed dark panel whatever the look, so white text remains readable over any scene.

A video narrated in English

Skincare serum UGC ad example still

Skincare serum UGC ad example

29s ugc

Questions people ask

Which English voices can I choose?

Eight American English voices are cast automatically: Puck, Laomedeia, Charon, Erinome, Umbriel, Sulafat, Iapetus and Aoede, covering energetic, instructional, warm and neutral styles in male and female versions. They are the automatic picks, not the ceiling; the per-scene voice dialog searches the entire Google catalogue, including British, Australian and Indian English voices.

Do I have to write my brief in English?

The brief is written in whatever language you think in; English is simply where the narration ends up when the Language field says English or stays empty. The director composes every spoken line and every on-screen headline itself, so the brief only needs the facts: the business, the offer, the tone you want.

Are captions automatic in English?

Yes. Captions are derived from the narration text, not from listening to the audio, so they never mishear a word. They are on by default and toggle off per scene in the voice dialog. Each word lights up as the voice reaches it, using timing measured against the actual clip.

How do I fix a mispronounced name?

Respell it phonetically in the narration text for that scene and regenerate the clip. There is no phonetic alphabet input, so the spelling is the control: write Nguyen as Win, or Loewe as Lo Ay Vay, until the read is right. The caption will show the respelled version, so consider turning captions off on that scene.

What does an English voiceover cost?

Narration is billed by length, about 1 credit per 330 characters, roughly 3 credits a minute of speech, with a minimum of half a credit per clip, and a typical video holds one clip per scene. Exporting bills its render time, usually a small amount; a longer export, or one that still has unsynthesised lines, needs at least 1 credit left to start. Regenerating a clip after rewriting its line is billed again.

Can two voices share one video?

Yes. The video-wide voice is just the default each scene inherits. Open the voice dialog on any scene and give that scene its own voice; the rest keep theirs. An opener and a calm explainer in different voices is two scene-level choices, nothing more.

Narrate a video in English

The button opens the video generator with the language already set in the brief. Pick a voice after the scenes are planned.

Make a video

Related pages

Portuguese AI voiceover

Choosing Portuguese gives you a Brazilian voice, one of eight pt-BR characters spanning energetic, instructional, warm and neutral reads. A team in Lisbon should know that up front: European Portuguese exists only through the per-scene catalogue search, not as the default.

Swahili AI voiceover

Swahili maps to sw-KE, the Kenyan locale, and it arrives with the complete cast of eight premium voices, a stronger tier than several European languages get. Captions are plain Latin, rendering natively in every look on the shelf.

Filipino AI voiceover

Filipino is one of three mapped languages without a premium voice tier, alongside Malay and Traditional Chinese: fil-PH offers fil-PH-Wavenet-C, male, and fil-PH-Wavenet-A, female. The script generation is full strength; only the voices are a generation older, and this page says what that means.

Every locale is listed onthe voiceover languages page.