Talking photo tool Make a photo talk with one portrait and a voice

The video follows a speech-audio track, supplied in either of two ways.

Use a recording you already have. Upload an MP3, M4A, WAV or OGG file, or select a voice track from Lab. No voice generation is needed; you go straight to the video quote.

Type the words and generate a voice. Open “No voice track yet? Type the words”, enter up to 5,000 characters in “Words to speak”, pick a voice from the menu, and press “Make the voice track”. The voice-track quote appears first; once the audio is ready it becomes the speech track for the render.

Either way, the words come from the audio. The optional “How should they say it?” field only shapes delivery — tone, gaze, expression.

Two things can be charged, and each is quoted separately before you confirm it.

Voice track from text: 45 credits per started 1,000 characters, only if you type a script and generate a voice.

Hedra Avatar video: 5 credits × source-audio duration rounded up to the next whole second, for every render.

Using your own recording? No voice-track charge — only the video quote. Typing a script? The voice-track quote comes first; the video quote follows once the audio exists, because the video price depends on its length. The combined total is not available until the voice track is ready.

One tested example: a two-sentence English test line was quoted 45 credits for the voice track and 30 credits for the Hedra render — 75 credits for that clip, not a standard package price. Costs can increase with script length and audio duration; the quote shows the applicable charge.

New accounts start with zero credits, so a talking-photo render needs credits before it runs. A subscription is optional — you can buy a one-time credit pack instead. Each generation is quoted before you confirm it. See pricing for current packs and plans.