P-Video 2 Pro Avatar Guide: Turn One Portrait plus Audio or a Script into a Talking Video up to 3 Minutes

P-Video 2 Pro Avatar is Pruna AI's talking-video model. Give it one portrait plus either an audio clip or a written script, and it generates a video of that person speaking. This guide covers what it does, how to call it, what it costs, and when to pick a different model.

What it does

• One portrait is enough: the only required input is the portrait image_url (jpg, jpeg, png or webp). It becomes the first frame. No source video and no avatar training needed.
• Two ways to drive it: with audio_url, the video is as long as the audio. With only a script (up to 5,000 characters), the built-in voice reads it aloud. Each video can run up to 3 minutes; at 4k the audio is limited to 125 seconds.
• Built-in voices: 30 voices (each name shows the gender; default Zephyr (Female)) and 10 languages: English (US), English (UK), Spanish, French, German, Italian, Portuguese (Brazil), Japanese, Korean and Hindi. Chinese is not included. Use voice_prompt for tone, pace and emotion; it is never spoken.
• Prompt controls the picture: prompt (up to 2,000 characters) describes how the person looks, moves and behaves while speaking. The default is "The person is talking."
• Five resolutions: 480p, 768p, 1080p, 2k and 4k, default 768p.

Input rules

• Send audio or a script. If you send both, the audio wins and script, voice, voice_language and voice_prompt are ignored.
• If you send neither, the task fails with a message asking for one of them, and the charge is fully refunded.
• The content safety filter is always on. Content that violates the policy fails with CONTENT_POLICY_VIOLATION and is fully refunded.

How to call it

Submit an async task, then poll with the returned taskId:

curl https://api.hiapi.ai/v1/tasks \
  -H "Authorization: Bearer $HIAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "p-video-2-pro-avatar",
    "input": {
      "image_url": "https://example.com/portrait.png",
      "script": "Welcome to the course. Today we cover three ideas.",
      "voice": "Charon (Male)",
      "voice_language": "English (US)",
      "voice_prompt": "Calm and clear, moderate pace",
      "resolution": "768p"
    }
  }'

curl https://api.hiapi.ai/v1/tasks/<taskId> \
  -H "Authorization: Bearer $HIAPI_KEY"

To use your own recording, replace script and the voice fields with "audio_url": "https://example.com/voice.mp3". When the task succeeds, output holds the video URL.

Pricing

Billed by output video length; partial seconds are not charged:

• 480p: $0.0286/s
• 768p: $0.0572/s
• 1080p: $0.0858/s
• 2k: $0.1143/s
• 4k: $0.1715/s

For example, a 30-second 768p clip costs about $1.72, and 30 seconds at 480p about $0.86. With audio, the hold is based on the audio length. With a script only, the spoken length is unknown at submit time, so 3 minutes are held first; after completion you pay for the actual video length and the rest is returned. Failed tasks are fully refunded.

Prompt tips

Keep prompt about the picture, not the words. Write it in this order: appearance → amount of movement → expression and gaze → light or camera. For example:

• Course intro: A calm instructor in a dark shirt, small head nods, steady eye contact, soft studio light
• Gallery tour: A curator speaking warmly to camera, blinks naturally, warm gallery lighting

Put tone and pacing in voice_prompt, for example "Speak softly, small pauses between sentences".

When to pick something else

• Re-syncing the lips of an existing video to new audio, or image-to-video with no one speaking: that is not what P-Video 2 Pro Avatar does. Use a video editing or image-to-video model instead.
• Swapping the person in an existing video while keeping its motion and camera: use P-Video 2 Pro Replace from the same series.

To try it, open the HiAPI Playground, upload a portrait and write a short script.