Talking Video

Talking Video is one place for the people you have already made, and one box to make them talk. You pick someone who is ready, or any picture as the first frame, give them a script or your own audio, and they say it. Direction lets you choose what they are doing, the emotion, the expression, the body movement, the hand gestures and the camera, and every choice goes into the video.

It used to be called Kriti Speak. Old links still open it.

What it does

The page has two tabs at the top of the feed.

• Ready shows your ready people as cards, the way Product Catalogue shows your products. Your AI Influencers come with their locked face (the Passport), their approved looks and their own voice. Your face models from Assets come with their voice. The Kritiverse avatars are there too, so you can start even before you have made anyone.
• Videos shows every talking video you made here. A video that is still being made shows Kriti working on it. A finished one opens in the preview, where you can play, download or post it.

Lip sync comes first. When the person talks to the camera with little movement, the video runs on the most accurate lip sync engine we tested side by side: the mouth matches the voice to the frame and the teeth stay clean. This is the same whether a voice reads your script or you add your own audio, and whether you give a first frame or Kriti makes one. When you ask for walking, lively movement or big gestures in Direction, a moving engine makes the video instead, because the lip sync engine keeps the body and the camera still. When the person holds a product, the video runs on the engine that keeps the product's label exactly as it is. A short line under the box always says which engine will run and why. If an engine is busy, the next one in line makes the video, at the same price.

People are made in AI Influencer and in Assets, Models. A card has a small link to the place it was made, and New influencer at the top takes you to AI Influencer.

When to use it

• You already have an AI Influencer or a model and want them to speak a script.
• You have a voice recording, a voice over or a song part, and want a person to say it.
• You want the same person in many short videos, with the same face and voice each time.
• You want the person to hold your product while they talk about it, or to show it beside them.
• You already have a picture you like, a generated one too, and want it to talk exactly as it is.
• To make a new person, use AI Influencer or Assets first, then come back here.

Step by step

1. Open Talking Video. If you have no videos yet, the page opens on Ready.
2. Press Use on a card. The person goes into the Model box with their voice and their default look. You can also press the Model box to pick from Assets.
3. Write what they say in the big box, or press the Audio box to add your own audio. To get help with a script, open Direction, then Script, and press Write one for me, or pick a script you saved in Script Writer.
4. Pick the language at the top: English, Hindi or Hinglish.
5. Press the shape and quality button to choose the shape (9:16, 1:1, 4:5 or 16:9) and the quality (Standard or HD).
6. Open Direction to change the cloth, add a product, choose what they are doing, the emotion, the movement, the gestures, the camera, the background, the framing and the voice.
7. Read the line under the box: how long the video is, and which engine will run.
8. Check the number on Make video. That number is the price. Press it.
9. The first time your own face talks, a short line asks you to confirm it is yours. Press I agree.
10. The video shows in Videos. It takes a few minutes, and you can leave the page.

Use your own audio

1. Press the Audio box, beside First frame. Assets opens on your audio. Pick a file, or upload one there.
2. Your audio is the speech, so no voice is made and the Voice section in Direction shows Not used. The script box is optional: what you write there is only the title.
3. The length and the price come from the real length of your audio. Your plan's longest video still applies.
4. A long audio is made in parts at the quiet moments between sentences and joined, as with a long script.
5. The small x on the Audio box clears it, and the voice comes back.

Hold a product

1. Press the Product box and pick a product in Assets, or open Direction, then Product.
2. Placement starts on In hand: the first picture is made with the person holding your product in one hand, the label facing the camera. With Head and shoulders framing, the product is held up beside the face so it stays in the picture. An action such as Hold it near the face and talk or Show it to camera says how it is held.
3. Pick Beside to show the product next to the person instead, on its own half of the video.
4. With In hand, the first picture is checked against your product before the person talks. When it does not pass, the video stops there and you pay nothing for the video part. The tile shows the picture, Use anyway with the price of the video, and Try again.
5. Make first frame with a product In hand makes the picture with the product held, so you can see it before it talks.

Talk from your own first frame

1. Press the First frame box and pick a picture in Assets, or upload one there. A filled box opens Assets again and your pick replaces the picture. The small x clears it.
2. The Model box is optional now. With a model, it gives the voice and who the person is. With no model, the picture itself talks, and the first time a short line asks you to confirm the face is yours or that you may use it.
3. The video starts from exactly that picture, so Cloth, Background, Framing and the product's Placement are not used, and no product is added to the picture. Direction shows them as not used. The rest of Direction still changes how the person moves and feels while talking: what they are doing, the emotion, the expression, the movement, the gestures and the camera move.
4. The video keeps the shape of your picture. With a First frame the shape button shows Auto with the shape it uses in small text: the closest of 9:16, 1:1, 4:5 and 16:9. If you pick a shape by hand that is far from your picture's, your shape stays: the picture is first extended to it, with the person kept exactly as they are, and the line under the box says so. The video is never padded with black bands.
5. If the face is very small in your picture, for example a person far from the camera, the lips could not show. A closer first frame of the same person in the same place is made first, from the waist up, and the line under the box says so. The price of that one picture is in the number on Make video.
6. A picture with no clear face is refused before anything is charged.
7. The video ends with the speech. A short line is never stretched to a longer video.

Make the first frame first

1. Put a person in the Model box and set the look, the Direction, the shape and the script.
2. Press Make first frame, beside Make video. Only the first picture is made, at its own price, with what they are doing, the emotion and the camera you chose.
3. The picture shows in Videos with the same tools as the Assets feed. Press Make it talk on it.
4. The picture goes into the First frame box, and the box it was made with comes back: the model, the voice or your audio, the script, the Direction, the language and the shape. Press Make video.

Every option explained

Who can use it

Free plan: included. Pro plan: included.

What it costs

You pay for the talking video by the second of speech, from 54 credits a second, and one first picture of the person. HD is charged at 54 credits a second. A big movement from Direction (Lively, Walking, Expressive or pointing hands, a use and show action, an Orbit or Follow camera) runs on the moving engine, charged at 54 credits a second. With your own audio, the seconds are the real length of your audio. The voice, the lips fix and the finish are free. With your own first frame there is no first picture to make, so that part is not charged, unless the picture is extended to the shape you picked or a closer first frame is made for a small face: then one first picture is in the price. Make first frame charges only the first picture, and the number on it is its price. The number on Make video is the whole price before you press it, and it changes when a choice changes the engine. A finished video is charged, as with every tool.

Tips

• For the cleanest lips, keep Body movement on Still or Small gestures and the camera on Static: the lip sync engine runs.
• A first frame with the face large in the picture talks best and costs least: no closer frame is needed.
• A closer framing, such as Head and shoulders or Half body, keeps the face large, so the mouth moves cleanly. Half body is cut at the waist and Standing at the knees, so the face never gets too small to talk.
• Pick gestures before Make first frame, so the hands start free in the picture.
• Write the script the way people talk. Short sentences sound best.
• Use the person's own voice. It is picked for you when they have one.
• Your own audio works best when it is clear speech with little music behind it.
• Your plan sets the longest video. The line under the box shows how many seconds your script or audio is.

Common problems

"Pick a presenter first." Press Use on a card, press the Model box, or add a First frame.

"Write what the presenter says first." The script box is empty and there is no audio. Write the words, use Write one for me in Direction, or add your own audio.

"This script is longer than ... seconds." Your plan has a longest video. Make the script or the audio shorter.

"We could not find a clear face in this picture." The first frame has no clear face. Pick a picture of one person facing the camera.

"This model is not ready to use yet." The person is still being made. Finish them in AI Influencer or Assets, then come back.

"That picture was not found." with an audio in the box: the audio was deleted or is not yours. Pick another audio in Assets.

The price went up when I changed Direction. A big movement runs on the moving engine, which has its own price. The line under the box says so. Pick Still or Small gestures for the lip sync engine.

The person hardly moves. The lip sync engine keeps the body still for the best lips. Pick Lively or Walking in Body movement, or Expressive gestures, for the moving engine.